diff --git "a/checkpoint-20877/trainer_state.json" "b/checkpoint-20877/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-20877/trainer_state.json"
@@ -0,0 +1,125308 @@
+{
+  "best_metric": 0.03778468817472458,
+  "best_model_checkpoint": "fake-news-classifier/checkpoint-20877",
+  "epoch": 3.0,
+  "eval_steps": 1,
+  "global_step": 20877,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6512,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5407,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.6787,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8644,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.5971,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.6583,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.4614,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6233,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.6213,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.6342,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.6196,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5437,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.712,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.7779,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8056,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.727,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6981,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6763,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6443,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.6869,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 0.5695,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.5397,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6295,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6142,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6501,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.5185,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.467,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.518,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 0.5903,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 0.5408,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 0.5311,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.5239,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 0.5122,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.5558,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6173,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 0.4018,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 0.7879,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 0.4286,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.5833,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.5454,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.7909,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.602,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-05,
+      "loss": 0.5185,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.5906,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-05,
+      "loss": 0.5212,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.7412,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-05,
+      "loss": 0.8134,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 0.4267,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6518,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-05,
+      "loss": 0.6703,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997599270178134e-05,
+      "loss": 0.4829,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519854035627e-05,
+      "loss": 0.7828,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992797810534405e-05,
+      "loss": 0.6737,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990397080712536e-05,
+      "loss": 0.2957,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9987996350890675e-05,
+      "loss": 0.4616,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998559562106881e-05,
+      "loss": 0.3115,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9983194891246945e-05,
+      "loss": 0.5426,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998079416142508e-05,
+      "loss": 0.3101,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997839343160321e-05,
+      "loss": 0.2827,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997599270178134e-05,
+      "loss": 0.2166,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997359197195948e-05,
+      "loss": 0.1253,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997119124213761e-05,
+      "loss": 0.3338,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996879051231574e-05,
+      "loss": 0.5052,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996638978249388e-05,
+      "loss": 0.1393,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996398905267202e-05,
+      "loss": 0.2692,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996158832285015e-05,
+      "loss": 0.0734,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9959187593028284e-05,
+      "loss": 0.0524,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9956786863206415e-05,
+      "loss": 0.092,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9954386133384554e-05,
+      "loss": 0.0493,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9951985403562686e-05,
+      "loss": 0.0368,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994958467374082e-05,
+      "loss": 0.1706,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994718394391895e-05,
+      "loss": 1.0697,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994478321409709e-05,
+      "loss": 0.3634,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9942382484275227e-05,
+      "loss": 0.5572,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993998175445336e-05,
+      "loss": 0.2658,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993758102463149e-05,
+      "loss": 0.6432,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993518029480962e-05,
+      "loss": 0.03,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993277956498776e-05,
+      "loss": 0.0252,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993037883516589e-05,
+      "loss": 0.0284,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9927978105344024e-05,
+      "loss": 0.3563,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9925577375522156e-05,
+      "loss": 0.0852,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9923176645700295e-05,
+      "loss": 0.0174,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992077591587843e-05,
+      "loss": 0.8125,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9918375186056565e-05,
+      "loss": 0.0112,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99159744562347e-05,
+      "loss": 0.0522,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9913573726412835e-05,
+      "loss": 0.3429,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991117299659097e-05,
+      "loss": 0.0203,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99087722667691e-05,
+      "loss": 0.6747,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990637153694723e-05,
+      "loss": 0.0127,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990397080712537e-05,
+      "loss": 0.6777,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99015700773035e-05,
+      "loss": 0.4576,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989916934748164e-05,
+      "loss": 0.1814,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989676861765977e-05,
+      "loss": 0.0094,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.98943678878379e-05,
+      "loss": 0.035,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989196715801604e-05,
+      "loss": 0.065,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9889566428194174e-05,
+      "loss": 0.0051,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9887165698372305e-05,
+      "loss": 0.0059,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988476496855044e-05,
+      "loss": 0.0157,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9882364238728576e-05,
+      "loss": 0.9923,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987996350890671e-05,
+      "loss": 0.0063,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9877562779084846e-05,
+      "loss": 0.0081,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987516204926298e-05,
+      "loss": 0.0062,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9872761319441117e-05,
+      "loss": 0.0053,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987036058961925e-05,
+      "loss": 0.0036,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986795985979738e-05,
+      "loss": 0.0096,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986555912997551e-05,
+      "loss": 0.5232,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986315840015365e-05,
+      "loss": 0.2951,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.986075767033178e-05,
+      "loss": 0.0032,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9858356940509914e-05,
+      "loss": 0.3186,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985595621068805e-05,
+      "loss": 0.0021,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9853555480866185e-05,
+      "loss": 0.0027,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.985115475104432e-05,
+      "loss": 0.0032,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9848754021222455e-05,
+      "loss": 0.0084,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984635329140059e-05,
+      "loss": 0.7273,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984395256157872e-05,
+      "loss": 0.002,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984155183175686e-05,
+      "loss": 0.0016,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983915110193499e-05,
+      "loss": 0.7327,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983675037211312e-05,
+      "loss": 0.7459,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.983434964229126e-05,
+      "loss": 0.0062,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98319489124694e-05,
+      "loss": 0.7664,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982954818264753e-05,
+      "loss": 0.0024,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982714745282566e-05,
+      "loss": 0.3814,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982474672300379e-05,
+      "loss": 0.4255,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.982234599318193e-05,
+      "loss": 0.034,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9819945263360064e-05,
+      "loss": 0.0524,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9817544533538195e-05,
+      "loss": 0.2737,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.981514380371633e-05,
+      "loss": 0.0748,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9812743073894466e-05,
+      "loss": 0.0131,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9810342344072604e-05,
+      "loss": 0.0658,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9807941614250736e-05,
+      "loss": 0.0146,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980554088442887e-05,
+      "loss": 0.5147,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9803140154607e-05,
+      "loss": 0.0055,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.980073942478514e-05,
+      "loss": 0.0055,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979833869496327e-05,
+      "loss": 0.0035,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97959379651414e-05,
+      "loss": 0.0604,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979353723531954e-05,
+      "loss": 0.6987,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979113650549768e-05,
+      "loss": 0.0044,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978873577567581e-05,
+      "loss": 0.542,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978633504585394e-05,
+      "loss": 0.4818,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9783934316032074e-05,
+      "loss": 0.0361,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.978153358621021e-05,
+      "loss": 0.6117,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9779132856388345e-05,
+      "loss": 0.0304,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977673212656648e-05,
+      "loss": 0.0311,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977433139674461e-05,
+      "loss": 0.0551,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.977193066692275e-05,
+      "loss": 0.3541,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976952993710088e-05,
+      "loss": 0.3278,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976712920727902e-05,
+      "loss": 0.2326,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976472847745715e-05,
+      "loss": 0.0433,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.976232774763529e-05,
+      "loss": 0.0433,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975992701781342e-05,
+      "loss": 0.0484,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975752628799155e-05,
+      "loss": 0.0111,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975512555816968e-05,
+      "loss": 0.6462,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.975272482834782e-05,
+      "loss": 0.034,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9750324098525954e-05,
+      "loss": 0.0731,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9747923368704085e-05,
+      "loss": 0.3598,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9745522638882224e-05,
+      "loss": 0.0076,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9743121909060356e-05,
+      "loss": 0.0073,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9740721179238494e-05,
+      "loss": 0.0565,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9738320449416626e-05,
+      "loss": 1.8751,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973591971959476e-05,
+      "loss": 0.003,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973351898977289e-05,
+      "loss": 0.5829,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973111825995103e-05,
+      "loss": 0.0048,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972871753012916e-05,
+      "loss": 0.6552,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972631680030729e-05,
+      "loss": 0.8969,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972391607048543e-05,
+      "loss": 0.0054,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.972151534066357e-05,
+      "loss": 0.0062,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97191146108417e-05,
+      "loss": 0.005,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.971671388101983e-05,
+      "loss": 0.0081,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9714313151197964e-05,
+      "loss": 0.0043,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97119124213761e-05,
+      "loss": 0.0035,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9709511691554235e-05,
+      "loss": 0.0036,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970711096173237e-05,
+      "loss": 0.0034,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.97047102319105e-05,
+      "loss": 0.2939,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.970230950208864e-05,
+      "loss": 0.6701,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9699908772266776e-05,
+      "loss": 0.0026,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969750804244491e-05,
+      "loss": 0.003,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969510731262304e-05,
+      "loss": 0.0038,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969270658280117e-05,
+      "loss": 0.003,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.969030585297931e-05,
+      "loss": 0.4961,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968790512315744e-05,
+      "loss": 0.3667,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968550439333557e-05,
+      "loss": 0.0021,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9683103663513705e-05,
+      "loss": 0.0101,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.968070293369185e-05,
+      "loss": 0.0056,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.967830220386998e-05,
+      "loss": 0.0445,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9675901474048114e-05,
+      "loss": 0.0033,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9673500744226246e-05,
+      "loss": 0.0047,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9671100014404384e-05,
+      "loss": 0.0031,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9668699284582516e-05,
+      "loss": 0.2516,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.966629855476065e-05,
+      "loss": 0.0079,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.966389782493878e-05,
+      "loss": 0.5294,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.966149709511692e-05,
+      "loss": 0.0037,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965909636529506e-05,
+      "loss": 0.0043,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965669563547319e-05,
+      "loss": 0.0022,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965429490565132e-05,
+      "loss": 0.0025,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965189417582945e-05,
+      "loss": 0.0063,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964949344600759e-05,
+      "loss": 0.785,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.964709271618572e-05,
+      "loss": 0.0712,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9644691986363854e-05,
+      "loss": 0.0024,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9642291256541986e-05,
+      "loss": 0.0037,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9639890526720125e-05,
+      "loss": 0.6749,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.963748979689826e-05,
+      "loss": 0.0049,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9635089067076395e-05,
+      "loss": 0.0048,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.963268833725453e-05,
+      "loss": 0.0066,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9630287607432666e-05,
+      "loss": 0.8146,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96278868776108e-05,
+      "loss": 0.4914,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.962548614778893e-05,
+      "loss": 0.5075,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.962308541796706e-05,
+      "loss": 0.5155,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96206846881452e-05,
+      "loss": 0.7556,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961828395832333e-05,
+      "loss": 0.0367,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961588322850146e-05,
+      "loss": 0.3582,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96134824986796e-05,
+      "loss": 0.0117,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9611081768857734e-05,
+      "loss": 0.4425,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960868103903587e-05,
+      "loss": 0.3826,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9606280309214004e-05,
+      "loss": 1.3227,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9603879579392136e-05,
+      "loss": 0.0144,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9601478849570274e-05,
+      "loss": 0.019,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9599078119748406e-05,
+      "loss": 0.3583,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959667738992654e-05,
+      "loss": 0.1066,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959427666010467e-05,
+      "loss": 0.1206,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959187593028281e-05,
+      "loss": 0.0091,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958947520046095e-05,
+      "loss": 0.0167,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958707447063908e-05,
+      "loss": 0.095,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958467374081721e-05,
+      "loss": 0.0068,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958227301099534e-05,
+      "loss": 0.0084,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957987228117348e-05,
+      "loss": 0.0056,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.957747155135161e-05,
+      "loss": 0.7042,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9575070821529744e-05,
+      "loss": 0.0533,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9572670091707876e-05,
+      "loss": 0.0056,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9570269361886015e-05,
+      "loss": 0.007,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956786863206415e-05,
+      "loss": 0.0046,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9565467902242285e-05,
+      "loss": 0.0053,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956306717242042e-05,
+      "loss": 0.013,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9560666442598556e-05,
+      "loss": 0.004,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955826571277669e-05,
+      "loss": 0.0059,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955586498295482e-05,
+      "loss": 0.006,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955346425313295e-05,
+      "loss": 0.0097,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.955106352331109e-05,
+      "loss": 0.0037,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954866279348923e-05,
+      "loss": 0.4153,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954626206366736e-05,
+      "loss": 0.0026,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954386133384549e-05,
+      "loss": 0.0033,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9541460604023623e-05,
+      "loss": 0.0043,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953905987420176e-05,
+      "loss": 0.0022,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9536659144379894e-05,
+      "loss": 0.0024,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9534258414558026e-05,
+      "loss": 0.0032,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.953185768473616e-05,
+      "loss": 0.8145,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9529456954914296e-05,
+      "loss": 0.133,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9527056225092435e-05,
+      "loss": 0.0017,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9524655495270566e-05,
+      "loss": 0.0019,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.95222547654487e-05,
+      "loss": 0.7091,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951985403562684e-05,
+      "loss": 0.0025,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951745330580497e-05,
+      "loss": 0.0018,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.95150525759831e-05,
+      "loss": 0.7033,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951265184616123e-05,
+      "loss": 0.0344,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951025111633937e-05,
+      "loss": 0.4871,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.95078503865175e-05,
+      "loss": 0.3874,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.950544965669564e-05,
+      "loss": 0.0037,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.950304892687377e-05,
+      "loss": 0.7876,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9500648197051905e-05,
+      "loss": 0.0059,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949824746723004e-05,
+      "loss": 0.0045,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9495846737408175e-05,
+      "loss": 0.0431,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949344600758631e-05,
+      "loss": 0.0062,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949104527776444e-05,
+      "loss": 0.0387,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.948864454794258e-05,
+      "loss": 0.1615,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.948624381812071e-05,
+      "loss": 0.0038,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.948384308829885e-05,
+      "loss": 0.0118,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.948144235847698e-05,
+      "loss": 0.0141,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.947904162865512e-05,
+      "loss": 0.0049,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.947664089883325e-05,
+      "loss": 0.4944,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.947424016901138e-05,
+      "loss": 0.0029,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9471839439189513e-05,
+      "loss": 0.0035,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946943870936765e-05,
+      "loss": 0.1606,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9467037979545784e-05,
+      "loss": 0.6317,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9464637249723916e-05,
+      "loss": 0.0027,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946223651990205e-05,
+      "loss": 0.0028,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9459835790080186e-05,
+      "loss": 0.0023,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9457435060258325e-05,
+      "loss": 0.0017,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9455034330436456e-05,
+      "loss": 0.0049,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945263360061459e-05,
+      "loss": 0.0017,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945023287079272e-05,
+      "loss": 0.0025,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944783214097086e-05,
+      "loss": 0.0014,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944543141114899e-05,
+      "loss": 0.0024,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944303068132712e-05,
+      "loss": 0.6921,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9440629951505254e-05,
+      "loss": 0.0011,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94382292216834e-05,
+      "loss": 0.094,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943582849186153e-05,
+      "loss": 0.1843,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.943342776203966e-05,
+      "loss": 0.3134,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9431027032217795e-05,
+      "loss": 0.4046,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942862630239593e-05,
+      "loss": 0.0014,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9426225572574065e-05,
+      "loss": 0.5598,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.94238248427522e-05,
+      "loss": 0.0066,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.942142411293033e-05,
+      "loss": 0.0032,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941902338310847e-05,
+      "loss": 0.0412,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9416622653286606e-05,
+      "loss": 0.0085,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941422192346474e-05,
+      "loss": 0.0014,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941182119364287e-05,
+      "loss": 0.6863,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9409420463821e-05,
+      "loss": 0.0013,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940701973399914e-05,
+      "loss": 0.0245,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940461900417727e-05,
+      "loss": 0.0111,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9402218274355403e-05,
+      "loss": 0.8913,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939981754453354e-05,
+      "loss": 0.0032,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9397416814711674e-05,
+      "loss": 0.0134,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.939501608488981e-05,
+      "loss": 0.0191,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9392615355067944e-05,
+      "loss": 0.0034,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9390214625246076e-05,
+      "loss": 1.564,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9387813895424215e-05,
+      "loss": 0.0021,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9385413165602346e-05,
+      "loss": 0.7649,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938301243578048e-05,
+      "loss": 0.002,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938061170595861e-05,
+      "loss": 0.6081,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937821097613675e-05,
+      "loss": 0.0027,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937581024631488e-05,
+      "loss": 0.0034,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937340951649302e-05,
+      "loss": 0.4244,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937100878667115e-05,
+      "loss": 0.3924,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.936860805684929e-05,
+      "loss": 0.6352,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.936620732702742e-05,
+      "loss": 0.0082,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.936380659720555e-05,
+      "loss": 0.0057,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9361405867383685e-05,
+      "loss": 0.0059,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935900513756182e-05,
+      "loss": 0.6305,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9356604407739955e-05,
+      "loss": 0.7939,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.935420367791809e-05,
+      "loss": 0.0711,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9351802948096225e-05,
+      "loss": 0.3825,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.934940221827436e-05,
+      "loss": 0.3178,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9347001488452496e-05,
+      "loss": 0.337,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.934460075863063e-05,
+      "loss": 0.0682,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.934220002880876e-05,
+      "loss": 0.0316,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933979929898689e-05,
+      "loss": 0.3317,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933739856916503e-05,
+      "loss": 0.2353,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933499783934316e-05,
+      "loss": 0.051,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9332597109521293e-05,
+      "loss": 0.0247,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.933019637969943e-05,
+      "loss": 0.1259,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.932779564987757e-05,
+      "loss": 0.0395,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93253949200557e-05,
+      "loss": 0.0098,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9322994190233834e-05,
+      "loss": 0.006,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9320593460411966e-05,
+      "loss": 0.0063,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9318192730590105e-05,
+      "loss": 0.0054,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9315792000768236e-05,
+      "loss": 0.0042,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.931339127094637e-05,
+      "loss": 0.3493,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.93109905411245e-05,
+      "loss": 0.003,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.930858981130264e-05,
+      "loss": 0.0618,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.930618908148078e-05,
+      "loss": 0.2952,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.930378835165891e-05,
+      "loss": 0.0021,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.930138762183704e-05,
+      "loss": 0.0028,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929898689201517e-05,
+      "loss": 0.0017,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929658616219331e-05,
+      "loss": 0.0015,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.929418543237144e-05,
+      "loss": 0.0035,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9291784702549575e-05,
+      "loss": 0.0012,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9289383972727706e-05,
+      "loss": 0.0014,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9286983242905845e-05,
+      "loss": 0.0014,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9284582513083984e-05,
+      "loss": 0.0077,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9282181783262115e-05,
+      "loss": 0.0018,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927978105344025e-05,
+      "loss": 0.5153,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9277380323618386e-05,
+      "loss": 0.0009,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927497959379652e-05,
+      "loss": 0.0009,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927257886397465e-05,
+      "loss": 0.0008,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.927017813415278e-05,
+      "loss": 0.001,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926777740433092e-05,
+      "loss": 0.0007,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926537667450905e-05,
+      "loss": 0.0007,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926297594468719e-05,
+      "loss": 0.6539,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.926057521486532e-05,
+      "loss": 0.0007,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9258174485043454e-05,
+      "loss": 0.0008,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.925577375522159e-05,
+      "loss": 0.0014,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9253373025399724e-05,
+      "loss": 0.9487,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9250972295577856e-05,
+      "loss": 0.0011,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924857156575599e-05,
+      "loss": 0.7705,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9246170835934126e-05,
+      "loss": 0.0048,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.924377010611226e-05,
+      "loss": 0.6843,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92413693762904e-05,
+      "loss": 0.0123,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923896864646853e-05,
+      "loss": 0.016,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923656791664667e-05,
+      "loss": 0.4487,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92341671868248e-05,
+      "loss": 0.0093,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923176645700293e-05,
+      "loss": 0.4632,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922936572718106e-05,
+      "loss": 0.0103,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92269649973592e-05,
+      "loss": 0.0183,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.922456426753733e-05,
+      "loss": 0.0141,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9222163537715465e-05,
+      "loss": 0.0092,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92197628078936e-05,
+      "loss": 0.0087,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9217362078071735e-05,
+      "loss": 0.0037,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9214961348249874e-05,
+      "loss": 0.003,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9212560618428005e-05,
+      "loss": 0.0026,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921015988860614e-05,
+      "loss": 0.0018,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920775915878427e-05,
+      "loss": 0.0014,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920535842896241e-05,
+      "loss": 0.0011,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.920295769914054e-05,
+      "loss": 0.0009,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.920055696931867e-05,
+      "loss": 0.0016,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.919815623949681e-05,
+      "loss": 0.0011,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.919575550967495e-05,
+      "loss": 1.3584,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.919335477985308e-05,
+      "loss": 0.001,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.919095405003121e-05,
+      "loss": 0.5974,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9188553320209344e-05,
+      "loss": 0.0016,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.918615259038748e-05,
+      "loss": 0.0009,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9183751860565614e-05,
+      "loss": 0.0018,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9181351130743746e-05,
+      "loss": 0.0014,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.917895040092188e-05,
+      "loss": 0.002,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9176549671100016e-05,
+      "loss": 0.8262,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9174148941278155e-05,
+      "loss": 0.002,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.917174821145629e-05,
+      "loss": 0.3566,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916934748163442e-05,
+      "loss": 0.0021,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916694675181256e-05,
+      "loss": 0.0018,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916454602199069e-05,
+      "loss": 0.0033,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.916214529216882e-05,
+      "loss": 0.0026,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915974456234695e-05,
+      "loss": 0.0029,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915734383252509e-05,
+      "loss": 0.0021,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915494310270322e-05,
+      "loss": 0.0019,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915254237288136e-05,
+      "loss": 0.1863,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.915014164305949e-05,
+      "loss": 0.0014,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9147740913237625e-05,
+      "loss": 0.8022,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9145340183415764e-05,
+      "loss": 0.001,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9142939453593895e-05,
+      "loss": 0.0647,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.914053872377203e-05,
+      "loss": 0.7647,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.913813799395016e-05,
+      "loss": 0.0162,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91357372641283e-05,
+      "loss": 0.0013,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.913333653430643e-05,
+      "loss": 0.0018,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.913093580448457e-05,
+      "loss": 0.0018,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91285350746627e-05,
+      "loss": 0.0019,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.912613434484084e-05,
+      "loss": 0.0017,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.912373361501897e-05,
+      "loss": 0.0019,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.91213328851971e-05,
+      "loss": 0.0099,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9118932155375234e-05,
+      "loss": 0.0019,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.911653142555337e-05,
+      "loss": 0.0013,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9114130695731504e-05,
+      "loss": 1.5044,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9111729965909636e-05,
+      "loss": 0.0018,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9109329236087774e-05,
+      "loss": 0.0025,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9106928506265906e-05,
+      "loss": 0.6792,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9104527776444045e-05,
+      "loss": 0.0019,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.910212704662218e-05,
+      "loss": 0.6389,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909972631680031e-05,
+      "loss": 0.0016,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909732558697844e-05,
+      "loss": 0.0018,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909492485715658e-05,
+      "loss": 0.005,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909252412733471e-05,
+      "loss": 0.003,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.909012339751284e-05,
+      "loss": 0.0024,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908772266769098e-05,
+      "loss": 0.8327,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908532193786912e-05,
+      "loss": 0.0023,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908292120804725e-05,
+      "loss": 0.0028,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908052047822538e-05,
+      "loss": 0.0027,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9078119748403515e-05,
+      "loss": 0.0036,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9075719018581654e-05,
+      "loss": 0.0023,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9073318288759785e-05,
+      "loss": 0.0042,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.907091755893792e-05,
+      "loss": 0.0017,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906851682911605e-05,
+      "loss": 0.0037,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906611609929419e-05,
+      "loss": 0.6182,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9063715369472326e-05,
+      "loss": 0.0008,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.906131463965046e-05,
+      "loss": 0.0027,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905891390982859e-05,
+      "loss": 0.0094,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905651318000672e-05,
+      "loss": 0.0013,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905411245018486e-05,
+      "loss": 0.0015,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.905171172036299e-05,
+      "loss": 0.0016,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9049310990541124e-05,
+      "loss": 0.0021,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9046910260719255e-05,
+      "loss": 0.0007,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90445095308974e-05,
+      "loss": 0.0014,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.904210880107553e-05,
+      "loss": 0.5498,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9039708071253664e-05,
+      "loss": 0.0025,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9037307341431796e-05,
+      "loss": 0.6885,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9034906611609935e-05,
+      "loss": 0.0044,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9032505881788067e-05,
+      "loss": 0.0017,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.90301051519662e-05,
+      "loss": 0.0013,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.902770442214433e-05,
+      "loss": 0.0093,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.902530369232247e-05,
+      "loss": 0.7557,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.90229029625006e-05,
+      "loss": 0.0016,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.902050223267874e-05,
+      "loss": 0.0013,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901810150285687e-05,
+      "loss": 0.7313,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9015700773035e-05,
+      "loss": 0.0239,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901330004321314e-05,
+      "loss": 0.454,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.901089931339127e-05,
+      "loss": 0.2613,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9008498583569405e-05,
+      "loss": 0.0172,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9006097853747544e-05,
+      "loss": 0.0423,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9003697123925675e-05,
+      "loss": 0.0287,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.900129639410381e-05,
+      "loss": 0.0018,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8998895664281946e-05,
+      "loss": 0.0029,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.899649493446008e-05,
+      "loss": 0.0023,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8994094204638216e-05,
+      "loss": 0.0034,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.899169347481635e-05,
+      "loss": 0.0024,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.898929274499448e-05,
+      "loss": 0.0014,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.898689201517261e-05,
+      "loss": 0.0032,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.898449128535075e-05,
+      "loss": 0.7297,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.898209055552888e-05,
+      "loss": 0.0038,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8979689825707014e-05,
+      "loss": 0.5757,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897728909588515e-05,
+      "loss": 0.0196,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8974888366063284e-05,
+      "loss": 0.0015,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897248763624142e-05,
+      "loss": 0.0024,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8970086906419554e-05,
+      "loss": 0.0041,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8967686176597686e-05,
+      "loss": 0.0034,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8965285446775825e-05,
+      "loss": 0.0009,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8962884716953957e-05,
+      "loss": 0.0031,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.896048398713209e-05,
+      "loss": 0.6109,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895808325731022e-05,
+      "loss": 0.5464,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895568252748836e-05,
+      "loss": 0.0051,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89532817976665e-05,
+      "loss": 0.0035,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.895088106784463e-05,
+      "loss": 0.0186,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894848033802276e-05,
+      "loss": 0.0092,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894607960820089e-05,
+      "loss": 0.5214,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894367887837903e-05,
+      "loss": 0.0267,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.894127814855716e-05,
+      "loss": 0.0049,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8938877418735295e-05,
+      "loss": 0.0079,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893647668891343e-05,
+      "loss": 0.0183,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.893407595909157e-05,
+      "loss": 0.006,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8931675229269704e-05,
+      "loss": 0.0259,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8929274499447836e-05,
+      "loss": 0.0027,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892687376962597e-05,
+      "loss": 0.0023,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8924473039804106e-05,
+      "loss": 0.0031,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.892207230998224e-05,
+      "loss": 0.0044,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891967158016037e-05,
+      "loss": 0.6648,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.89172708503385e-05,
+      "loss": 0.0041,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891487012051664e-05,
+      "loss": 0.9847,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891246939069478e-05,
+      "loss": 0.5109,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.891006866087291e-05,
+      "loss": 0.0049,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890766793105104e-05,
+      "loss": 0.002,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8905267201229174e-05,
+      "loss": 0.0025,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890286647140731e-05,
+      "loss": 0.5419,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8900465741585444e-05,
+      "loss": 0.002,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8898065011763576e-05,
+      "loss": 0.0016,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.889566428194171e-05,
+      "loss": 0.0029,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8893263552119847e-05,
+      "loss": 0.6161,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8890862822297985e-05,
+      "loss": 0.0035,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888846209247612e-05,
+      "loss": 0.0028,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888606136265425e-05,
+      "loss": 0.0065,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888366063283239e-05,
+      "loss": 0.0057,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.888125990301052e-05,
+      "loss": 0.0011,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887885917318865e-05,
+      "loss": 0.0048,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887645844336678e-05,
+      "loss": 0.2909,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887405771354492e-05,
+      "loss": 0.0022,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.887165698372305e-05,
+      "loss": 0.0233,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8869256253901185e-05,
+      "loss": 0.0012,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8866855524079323e-05,
+      "loss": 0.1249,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8864454794257455e-05,
+      "loss": 0.6851,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8862054064435594e-05,
+      "loss": 0.002,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8859653334613726e-05,
+      "loss": 0.0019,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.885725260479186e-05,
+      "loss": 0.0021,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.885485187496999e-05,
+      "loss": 0.0013,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.885245114514813e-05,
+      "loss": 0.0016,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.885005041532626e-05,
+      "loss": 0.0049,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.884764968550439e-05,
+      "loss": 0.0023,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.884524895568253e-05,
+      "loss": 0.0015,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.884284822586067e-05,
+      "loss": 0.7769,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.88404474960388e-05,
+      "loss": 0.002,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.883804676621693e-05,
+      "loss": 0.0018,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8835646036395064e-05,
+      "loss": 0.0014,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.88332453065732e-05,
+      "loss": 0.0013,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8830844576751334e-05,
+      "loss": 1.1834,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8828443846929466e-05,
+      "loss": 0.4867,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.88260431171076e-05,
+      "loss": 0.0016,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8823642387285737e-05,
+      "loss": 0.4131,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8821241657463875e-05,
+      "loss": 0.2964,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.881884092764201e-05,
+      "loss": 0.0041,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.881644019782014e-05,
+      "loss": 0.0032,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.881403946799827e-05,
+      "loss": 0.0047,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.881163873817641e-05,
+      "loss": 0.1299,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880923800835454e-05,
+      "loss": 0.0055,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880683727853267e-05,
+      "loss": 0.0183,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880443654871081e-05,
+      "loss": 0.0126,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.880203581888895e-05,
+      "loss": 0.0209,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.879963508906708e-05,
+      "loss": 0.0025,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8797234359245213e-05,
+      "loss": 0.43,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8794833629423345e-05,
+      "loss": 0.0036,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8792432899601484e-05,
+      "loss": 0.1089,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8790032169779616e-05,
+      "loss": 0.0037,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878763143995775e-05,
+      "loss": 0.0033,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878523071013588e-05,
+      "loss": 0.477,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.878282998031402e-05,
+      "loss": 0.0018,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8780429250492156e-05,
+      "loss": 0.002,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877802852067029e-05,
+      "loss": 0.5912,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877562779084842e-05,
+      "loss": 0.0183,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877322706102656e-05,
+      "loss": 0.7718,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.877082633120469e-05,
+      "loss": 0.0041,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876842560138282e-05,
+      "loss": 0.0023,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8766024871560954e-05,
+      "loss": 0.0017,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.876362414173909e-05,
+      "loss": 0.0013,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8761223411917224e-05,
+      "loss": 0.0286,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.875882268209536e-05,
+      "loss": 0.0023,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8756421952273495e-05,
+      "loss": 0.0013,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8754021222451626e-05,
+      "loss": 0.3185,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8751620492629765e-05,
+      "loss": 0.0006,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87492197628079e-05,
+      "loss": 0.0168,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874681903298603e-05,
+      "loss": 0.0011,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874441830316416e-05,
+      "loss": 0.0048,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87420175733423e-05,
+      "loss": 0.004,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873961684352043e-05,
+      "loss": 0.6554,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873721611369857e-05,
+      "loss": 0.0037,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.87348153838767e-05,
+      "loss": 0.0055,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873241465405484e-05,
+      "loss": 0.0009,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.873001392423297e-05,
+      "loss": 0.001,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8727613194411103e-05,
+      "loss": 0.001,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8725212464589235e-05,
+      "loss": 0.0026,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8722811734767374e-05,
+      "loss": 0.0006,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8720411004945506e-05,
+      "loss": 0.0007,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871801027512364e-05,
+      "loss": 0.0013,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871560954530177e-05,
+      "loss": 0.0027,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871320881547991e-05,
+      "loss": 0.0005,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8710808085658046e-05,
+      "loss": 0.0008,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870840735583618e-05,
+      "loss": 0.4709,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870600662601431e-05,
+      "loss": 0.285,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870360589619244e-05,
+      "loss": 0.0005,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870120516637058e-05,
+      "loss": 0.0019,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.869880443654871e-05,
+      "loss": 0.004,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8696403706726844e-05,
+      "loss": 0.0007,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8694002976904976e-05,
+      "loss": 0.0005,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.869160224708312e-05,
+      "loss": 0.0005,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.868920151726125e-05,
+      "loss": 0.0005,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8686800787439385e-05,
+      "loss": 0.0006,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8684400057617516e-05,
+      "loss": 0.0004,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8681999327795655e-05,
+      "loss": 0.0006,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.867959859797379e-05,
+      "loss": 0.0004,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.867719786815192e-05,
+      "loss": 0.0073,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.867479713833005e-05,
+      "loss": 0.0004,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.867239640850819e-05,
+      "loss": 0.0005,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866999567868633e-05,
+      "loss": 0.9537,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866759494886446e-05,
+      "loss": 0.9531,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866519421904259e-05,
+      "loss": 0.0082,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866279348922072e-05,
+      "loss": 0.0004,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.866039275939886e-05,
+      "loss": 0.0011,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.865799202957699e-05,
+      "loss": 0.0009,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8655591299755125e-05,
+      "loss": 0.0012,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.865319056993326e-05,
+      "loss": 0.0009,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8650789840111396e-05,
+      "loss": 0.0004,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8648389110289534e-05,
+      "loss": 1.5713,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8645988380467666e-05,
+      "loss": 0.0008,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.86435876506458e-05,
+      "loss": 0.0005,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8641186920823936e-05,
+      "loss": 0.0013,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.863878619100207e-05,
+      "loss": 0.0046,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.86363854611802e-05,
+      "loss": 0.0013,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.863398473135833e-05,
+      "loss": 0.001,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.863158400153647e-05,
+      "loss": 0.1039,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.86291832717146e-05,
+      "loss": 0.001,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862678254189274e-05,
+      "loss": 0.0021,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.862438181207087e-05,
+      "loss": 0.0016,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8621981082249004e-05,
+      "loss": 0.0016,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861958035242714e-05,
+      "loss": 0.002,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8617179622605275e-05,
+      "loss": 0.1946,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8614778892783406e-05,
+      "loss": 0.448,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.861237816296154e-05,
+      "loss": 0.0014,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860997743313968e-05,
+      "loss": 0.4122,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860757670331781e-05,
+      "loss": 0.0033,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860517597349595e-05,
+      "loss": 0.001,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860277524367408e-05,
+      "loss": 0.0282,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.860037451385222e-05,
+      "loss": 0.2338,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859797378403035e-05,
+      "loss": 0.0051,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859557305420848e-05,
+      "loss": 0.0023,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859317232438661e-05,
+      "loss": 0.1344,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.859077159456475e-05,
+      "loss": 0.0006,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.858837086474288e-05,
+      "loss": 0.0011,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8585970134921015e-05,
+      "loss": 0.0625,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8583569405099154e-05,
+      "loss": 0.0017,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8581168675277286e-05,
+      "loss": 0.0012,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8578767945455424e-05,
+      "loss": 0.0018,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8576367215633556e-05,
+      "loss": 0.0014,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.857396648581169e-05,
+      "loss": 0.001,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8571565755989826e-05,
+      "loss": 0.0008,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856916502616796e-05,
+      "loss": 0.0012,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856676429634609e-05,
+      "loss": 0.7733,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856436356652422e-05,
+      "loss": 0.0009,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.856196283670236e-05,
+      "loss": 0.0015,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85595621068805e-05,
+      "loss": 0.0013,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855716137705863e-05,
+      "loss": 0.5673,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855476064723676e-05,
+      "loss": 0.002,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8552359917414894e-05,
+      "loss": 0.0014,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854995918759303e-05,
+      "loss": 0.0006,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8547558457771165e-05,
+      "loss": 0.0765,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8545157727949296e-05,
+      "loss": 0.0071,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854275699812743e-05,
+      "loss": 0.0048,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.854035626830557e-05,
+      "loss": 0.003,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8537955538483705e-05,
+      "loss": 0.0037,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853555480866184e-05,
+      "loss": 0.4294,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853315407883997e-05,
+      "loss": 0.0024,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.853075334901811e-05,
+      "loss": 0.0039,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.852835261919624e-05,
+      "loss": 0.2375,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.852595188937437e-05,
+      "loss": 0.0021,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.85235511595525e-05,
+      "loss": 0.0029,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.852115042973064e-05,
+      "loss": 0.1812,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.851874969990877e-05,
+      "loss": 0.0198,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.851634897008691e-05,
+      "loss": 0.8531,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8513948240265044e-05,
+      "loss": 0.002,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8511547510443175e-05,
+      "loss": 0.0028,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8509146780621314e-05,
+      "loss": 0.0514,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8506746050799446e-05,
+      "loss": 0.0015,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.850434532097758e-05,
+      "loss": 0.0592,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.850194459115571e-05,
+      "loss": 0.0021,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849954386133385e-05,
+      "loss": 0.0025,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849714313151198e-05,
+      "loss": 1.08,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849474240169012e-05,
+      "loss": 0.0011,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.849234167186825e-05,
+      "loss": 0.2308,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848994094204639e-05,
+      "loss": 0.0054,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848754021222452e-05,
+      "loss": 0.0277,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848513948240265e-05,
+      "loss": 0.0015,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8482738752580784e-05,
+      "loss": 0.0024,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.848033802275892e-05,
+      "loss": 0.0022,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8477937292937055e-05,
+      "loss": 0.0016,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8475536563115186e-05,
+      "loss": 0.0016,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8473135833293325e-05,
+      "loss": 0.0019,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.847073510347146e-05,
+      "loss": 0.0014,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8468334373649595e-05,
+      "loss": 0.0011,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.846593364382773e-05,
+      "loss": 0.001,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.846353291400586e-05,
+      "loss": 0.0012,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.846113218418399e-05,
+      "loss": 0.001,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.845873145436213e-05,
+      "loss": 0.0016,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.845633072454026e-05,
+      "loss": 0.0009,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.845392999471839e-05,
+      "loss": 0.001,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.845152926489653e-05,
+      "loss": 0.0009,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.844912853507467e-05,
+      "loss": 0.0059,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84467278052528e-05,
+      "loss": 0.0617,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8444327075430934e-05,
+      "loss": 0.0015,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8441926345609065e-05,
+      "loss": 0.0069,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8439525615787204e-05,
+      "loss": 0.0052,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8437124885965336e-05,
+      "loss": 0.0057,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.843472415614347e-05,
+      "loss": 0.0006,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.84323234263216e-05,
+      "loss": 0.0007,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842992269649974e-05,
+      "loss": 0.0009,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8427521966677877e-05,
+      "loss": 0.0006,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842512123685601e-05,
+      "loss": 0.0006,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842272050703414e-05,
+      "loss": 0.0006,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.842031977721227e-05,
+      "loss": 0.0005,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841791904739041e-05,
+      "loss": 0.0004,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841551831756854e-05,
+      "loss": 1.0209,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8413117587746674e-05,
+      "loss": 0.0008,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.841071685792481e-05,
+      "loss": 0.7563,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8408316128102945e-05,
+      "loss": 0.0766,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840591539828108e-05,
+      "loss": 0.7244,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8403514668459215e-05,
+      "loss": 0.0068,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.840111393863735e-05,
+      "loss": 0.0067,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8398713208815485e-05,
+      "loss": 0.0076,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839631247899362e-05,
+      "loss": 0.0105,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839391174917175e-05,
+      "loss": 0.0094,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.839151101934988e-05,
+      "loss": 0.006,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838911028952802e-05,
+      "loss": 0.0031,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838670955970615e-05,
+      "loss": 0.0034,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838430882988429e-05,
+      "loss": 0.0021,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838190810006242e-05,
+      "loss": 0.0026,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837950737024055e-05,
+      "loss": 0.0042,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.837710664041869e-05,
+      "loss": 0.003,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8374705910596824e-05,
+      "loss": 0.003,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8372305180774955e-05,
+      "loss": 0.0029,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8369904450953094e-05,
+      "loss": 0.6065,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8367503721131226e-05,
+      "loss": 0.0021,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.836510299130936e-05,
+      "loss": 0.0021,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8362702261487496e-05,
+      "loss": 0.0015,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.836030153166563e-05,
+      "loss": 0.0015,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8357900801843767e-05,
+      "loss": 0.0031,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83555000720219e-05,
+      "loss": 0.7245,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.835309934220003e-05,
+      "loss": 0.0024,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.835069861237816e-05,
+      "loss": 0.0017,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83482978825563e-05,
+      "loss": 0.5279,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.834589715273443e-05,
+      "loss": 0.0009,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8343496422912564e-05,
+      "loss": 0.0018,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.83410956930907e-05,
+      "loss": 0.0021,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.833869496326884e-05,
+      "loss": 0.0034,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.833629423344697e-05,
+      "loss": 0.0014,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8333893503625105e-05,
+      "loss": 0.001,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.833149277380324e-05,
+      "loss": 0.0051,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8329092043981375e-05,
+      "loss": 0.0009,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.832669131415951e-05,
+      "loss": 0.0047,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.832429058433764e-05,
+      "loss": 0.0009,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.832188985451577e-05,
+      "loss": 0.8214,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.831948912469391e-05,
+      "loss": 0.0007,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.831708839487205e-05,
+      "loss": 0.0009,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.831468766505018e-05,
+      "loss": 0.0008,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.831228693522831e-05,
+      "loss": 0.0019,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830988620540644e-05,
+      "loss": 0.0074,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830748547558458e-05,
+      "loss": 0.0021,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8305084745762714e-05,
+      "loss": 0.4486,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8302684015940845e-05,
+      "loss": 0.0006,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.830028328611898e-05,
+      "loss": 0.0007,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.829788255629712e-05,
+      "loss": 0.0011,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8295481826475254e-05,
+      "loss": 0.0026,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8293081096653386e-05,
+      "loss": 0.0005,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.829068036683152e-05,
+      "loss": 0.0019,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8288279637009657e-05,
+      "loss": 0.0005,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.828587890718779e-05,
+      "loss": 0.0004,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.828347817736592e-05,
+      "loss": 0.0007,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.828107744754405e-05,
+      "loss": 0.0004,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827867671772219e-05,
+      "loss": 0.0005,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827627598790032e-05,
+      "loss": 0.0004,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827387525807846e-05,
+      "loss": 0.0078,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.827147452825659e-05,
+      "loss": 0.0037,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8269073798434724e-05,
+      "loss": 0.0007,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826667306861286e-05,
+      "loss": 0.0019,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8264272338790995e-05,
+      "loss": 0.0004,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.826187160896913e-05,
+      "loss": 0.0005,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825947087914726e-05,
+      "loss": 0.0004,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82570701493254e-05,
+      "loss": 0.0003,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825466941950353e-05,
+      "loss": 0.0008,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.825226868968167e-05,
+      "loss": 0.0003,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82498679598598e-05,
+      "loss": 0.5687,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824746723003794e-05,
+      "loss": 0.8011,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824506650021607e-05,
+      "loss": 0.0003,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82426657703942e-05,
+      "loss": 0.0003,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824026504057233e-05,
+      "loss": 0.1466,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.823786431075047e-05,
+      "loss": 0.0003,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8235463580928604e-05,
+      "loss": 0.0003,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8233062851106735e-05,
+      "loss": 0.0004,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8230662121284874e-05,
+      "loss": 0.443,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8228261391463006e-05,
+      "loss": 0.0003,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8225860661641144e-05,
+      "loss": 0.0003,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8223459931819276e-05,
+      "loss": 0.0003,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.822105920199741e-05,
+      "loss": 0.0092,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821865847217554e-05,
+      "loss": 0.0022,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821625774235368e-05,
+      "loss": 0.0003,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821385701253181e-05,
+      "loss": 0.0003,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.821145628270994e-05,
+      "loss": 0.0019,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820905555288808e-05,
+      "loss": 0.0003,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820665482306622e-05,
+      "loss": 0.0016,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820425409324435e-05,
+      "loss": 0.0003,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820185336342248e-05,
+      "loss": 0.0004,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8199452633600614e-05,
+      "loss": 0.0217,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819705190377875e-05,
+      "loss": 0.0003,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8194651173956885e-05,
+      "loss": 1.06,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819225044413502e-05,
+      "loss": 0.0003,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818984971431315e-05,
+      "loss": 0.0003,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818744898449129e-05,
+      "loss": 0.003,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8185048254669426e-05,
+      "loss": 0.4549,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818264752484756e-05,
+      "loss": 0.0003,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818024679502569e-05,
+      "loss": 0.643,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817784606520382e-05,
+      "loss": 0.0004,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817544533538196e-05,
+      "loss": 0.0184,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817304460556009e-05,
+      "loss": 0.0003,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817064387573822e-05,
+      "loss": 0.0046,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816824314591636e-05,
+      "loss": 0.0009,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81658424160945e-05,
+      "loss": 0.0254,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816344168627263e-05,
+      "loss": 0.0004,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8161040956450764e-05,
+      "loss": 0.0006,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8158640226628896e-05,
+      "loss": 0.0003,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8156239496807034e-05,
+      "loss": 0.9343,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8153838766985166e-05,
+      "loss": 0.0003,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.81514380371633e-05,
+      "loss": 0.0003,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814903730734143e-05,
+      "loss": 0.0006,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814663657751957e-05,
+      "loss": 0.0007,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814423584769771e-05,
+      "loss": 0.0014,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814183511787584e-05,
+      "loss": 0.3623,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813943438805397e-05,
+      "loss": 0.0006,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813703365823211e-05,
+      "loss": 0.0007,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813463292841024e-05,
+      "loss": 0.8074,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813223219858837e-05,
+      "loss": 0.0024,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8129831468766504e-05,
+      "loss": 0.5294,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812743073894464e-05,
+      "loss": 0.0005,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8125030009122775e-05,
+      "loss": 0.0006,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812262927930091e-05,
+      "loss": 0.0069,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8120228549479045e-05,
+      "loss": 0.6946,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811782781965718e-05,
+      "loss": 0.0015,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8115427089835316e-05,
+      "loss": 0.01,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811302636001345e-05,
+      "loss": 0.2274,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811062563019158e-05,
+      "loss": 0.0061,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810822490036971e-05,
+      "loss": 0.0009,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810582417054785e-05,
+      "loss": 0.6277,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810342344072598e-05,
+      "loss": 0.0078,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810102271090411e-05,
+      "loss": 0.001,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809862198108225e-05,
+      "loss": 0.0067,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809622125126039e-05,
+      "loss": 0.0033,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809382052143852e-05,
+      "loss": 0.0005,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8091419791616654e-05,
+      "loss": 0.0087,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8089019061794786e-05,
+      "loss": 0.0006,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8086618331972924e-05,
+      "loss": 0.0007,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8084217602151056e-05,
+      "loss": 0.0025,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.808181687232919e-05,
+      "loss": 0.0008,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807941614250732e-05,
+      "loss": 0.0006,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807701541268546e-05,
+      "loss": 0.0004,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80746146828636e-05,
+      "loss": 0.0021,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807221395304173e-05,
+      "loss": 0.0004,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806981322321986e-05,
+      "loss": 0.0004,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806741249339799e-05,
+      "loss": 0.0004,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806501176357613e-05,
+      "loss": 0.0002,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806261103375426e-05,
+      "loss": 0.0051,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8060210303932394e-05,
+      "loss": 0.0003,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8057809574110526e-05,
+      "loss": 0.0004,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805540884428867e-05,
+      "loss": 0.639,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80530081144668e-05,
+      "loss": 0.0003,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8050607384644935e-05,
+      "loss": 0.0002,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804820665482307e-05,
+      "loss": 0.0002,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8045805925001206e-05,
+      "loss": 0.0024,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804340519517934e-05,
+      "loss": 0.0012,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.804100446535747e-05,
+      "loss": 0.0002,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.80386037355356e-05,
+      "loss": 0.0002,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803620300571374e-05,
+      "loss": 0.0002,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.803380227589188e-05,
+      "loss": 0.0002,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803140154607001e-05,
+      "loss": 0.0018,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802900081624814e-05,
+      "loss": 0.0002,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8026600086426273e-05,
+      "loss": 0.0033,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.802419935660441e-05,
+      "loss": 0.0009,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8021798626782544e-05,
+      "loss": 0.0002,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8019397896960676e-05,
+      "loss": 0.009,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801699716713881e-05,
+      "loss": 0.0003,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8014596437316946e-05,
+      "loss": 0.0015,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8012195707495085e-05,
+      "loss": 0.003,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8009794977673216e-05,
+      "loss": 0.0002,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800739424785135e-05,
+      "loss": 0.0002,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800499351802949e-05,
+      "loss": 0.0003,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800259278820762e-05,
+      "loss": 0.0016,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800019205838575e-05,
+      "loss": 0.5506,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799779132856388e-05,
+      "loss": 0.0001,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799539059874202e-05,
+      "loss": 0.0001,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799298986892015e-05,
+      "loss": 0.0001,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799058913909829e-05,
+      "loss": 0.0001,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798818840927642e-05,
+      "loss": 0.0002,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7985787679454555e-05,
+      "loss": 0.0013,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798338694963269e-05,
+      "loss": 0.001,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7980986219810825e-05,
+      "loss": 0.0001,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797858548998896e-05,
+      "loss": 0.0001,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7976184760167096e-05,
+      "loss": 0.0002,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797378403034523e-05,
+      "loss": 0.0005,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797138330052336e-05,
+      "loss": 0.0001,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796898257070149e-05,
+      "loss": 0.0002,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796658184087963e-05,
+      "loss": 0.0001,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796418111105777e-05,
+      "loss": 0.0015,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79617803812359e-05,
+      "loss": 0.0001,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795937965141403e-05,
+      "loss": 0.0002,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7956978921592163e-05,
+      "loss": 0.0001,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79545781917703e-05,
+      "loss": 0.0003,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7952177461948434e-05,
+      "loss": 0.0001,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7949776732126566e-05,
+      "loss": 0.0009,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79473760023047e-05,
+      "loss": 0.0001,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7944975272482836e-05,
+      "loss": 0.0001,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7942574542660975e-05,
+      "loss": 1.1309,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7940173812839106e-05,
+      "loss": 0.0001,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793777308301724e-05,
+      "loss": 0.9916,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793537235319538e-05,
+      "loss": 0.0006,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793297162337351e-05,
+      "loss": 0.0003,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793057089355164e-05,
+      "loss": 0.0004,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792817016372977e-05,
+      "loss": 0.3377,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792576943390791e-05,
+      "loss": 0.0006,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792336870408605e-05,
+      "loss": 0.0008,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792096797426418e-05,
+      "loss": 0.2347,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791856724444231e-05,
+      "loss": 0.2838,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7916166514620445e-05,
+      "loss": 0.004,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791376578479858e-05,
+      "loss": 0.0011,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7911365054976715e-05,
+      "loss": 0.0009,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790896432515485e-05,
+      "loss": 0.0008,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790656359533298e-05,
+      "loss": 0.6597,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790416286551112e-05,
+      "loss": 0.7073,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7901762135689256e-05,
+      "loss": 0.0013,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789936140586739e-05,
+      "loss": 0.0036,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789696067604552e-05,
+      "loss": 0.0027,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789455994622366e-05,
+      "loss": 0.0058,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789215921640179e-05,
+      "loss": 0.0038,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788975848657992e-05,
+      "loss": 0.0033,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7887357756758053e-05,
+      "loss": 0.0028,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788495702693619e-05,
+      "loss": 0.0046,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7882556297114324e-05,
+      "loss": 0.0025,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788015556729246e-05,
+      "loss": 0.0033,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7877754837470594e-05,
+      "loss": 0.0157,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7875354107648726e-05,
+      "loss": 0.0015,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7872953377826865e-05,
+      "loss": 0.001,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7870552648004996e-05,
+      "loss": 0.0018,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786815191818313e-05,
+      "loss": 0.002,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786575118836126e-05,
+      "loss": 0.0008,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78633504585394e-05,
+      "loss": 0.001,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.786094972871753e-05,
+      "loss": 0.3451,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785854899889567e-05,
+      "loss": 0.002,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78561482690738e-05,
+      "loss": 0.0016,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785374753925194e-05,
+      "loss": 0.0021,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.785134680943007e-05,
+      "loss": 0.0008,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78489460796082e-05,
+      "loss": 0.0014,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7846545349786335e-05,
+      "loss": 0.0014,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784414461996447e-05,
+      "loss": 0.0012,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7841743890142605e-05,
+      "loss": 0.0009,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783934316032074e-05,
+      "loss": 0.0007,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7836942430498875e-05,
+      "loss": 0.001,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783454170067701e-05,
+      "loss": 0.0006,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7832140970855146e-05,
+      "loss": 0.0005,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782974024103328e-05,
+      "loss": 0.0006,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782733951121141e-05,
+      "loss": 0.0015,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782493878138954e-05,
+      "loss": 0.0008,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782253805156768e-05,
+      "loss": 0.0007,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.782013732174581e-05,
+      "loss": 0.0004,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7817736591923943e-05,
+      "loss": 0.0015,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781533586210208e-05,
+      "loss": 0.0003,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781293513228022e-05,
+      "loss": 0.0006,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.781053440245835e-05,
+      "loss": 0.0009,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7808133672636484e-05,
+      "loss": 0.0006,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7805732942814616e-05,
+      "loss": 0.0009,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7803332212992755e-05,
+      "loss": 0.0005,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7800931483170886e-05,
+      "loss": 0.0003,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779853075334902e-05,
+      "loss": 0.0007,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779613002352715e-05,
+      "loss": 0.0019,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779372929370529e-05,
+      "loss": 0.0012,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.779132856388343e-05,
+      "loss": 0.0007,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778892783406156e-05,
+      "loss": 0.0003,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778652710423969e-05,
+      "loss": 0.0008,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778412637441782e-05,
+      "loss": 0.0004,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778172564459596e-05,
+      "loss": 0.0002,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777932491477409e-05,
+      "loss": 0.0006,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7776924184952225e-05,
+      "loss": 0.0003,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777452345513036e-05,
+      "loss": 0.0002,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7772122725308495e-05,
+      "loss": 0.0003,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7769721995486634e-05,
+      "loss": 0.0003,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7767321265664765e-05,
+      "loss": 0.0003,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77649205358429e-05,
+      "loss": 0.3259,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7762519806021036e-05,
+      "loss": 0.0003,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776011907619917e-05,
+      "loss": 0.0004,
+      "step": 983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77577183463773e-05,
+      "loss": 0.0002,
+      "step": 984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775531761655543e-05,
+      "loss": 0.0002,
+      "step": 985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775291688673357e-05,
+      "loss": 0.0003,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77505161569117e-05,
+      "loss": 0.0003,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774811542708984e-05,
+      "loss": 0.0002,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774571469726797e-05,
+      "loss": 0.0005,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774331396744611e-05,
+      "loss": 0.0003,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774091323762424e-05,
+      "loss": 0.0246,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7738512507802374e-05,
+      "loss": 0.0002,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7736111777980506e-05,
+      "loss": 0.0003,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7733711048158645e-05,
+      "loss": 0.0002,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7731310318336776e-05,
+      "loss": 0.0006,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772890958851491e-05,
+      "loss": 0.0001,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772650885869305e-05,
+      "loss": 0.0007,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772410812887118e-05,
+      "loss": 0.0004,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772170739904932e-05,
+      "loss": 0.0004,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771930666922745e-05,
+      "loss": 0.0002,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771690593940558e-05,
+      "loss": 0.0008,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771450520958371e-05,
+      "loss": 0.0212,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771210447976185e-05,
+      "loss": 0.7942,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770970374993998e-05,
+      "loss": 0.0008,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7707303020118115e-05,
+      "loss": 0.8486,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770490229029625e-05,
+      "loss": 0.0001,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770250156047439e-05,
+      "loss": 0.0009,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7700100830652524e-05,
+      "loss": 0.0002,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7697700100830655e-05,
+      "loss": 0.0001,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.769529937100879e-05,
+      "loss": 0.8459,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7692898641186926e-05,
+      "loss": 0.0002,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.769049791136506e-05,
+      "loss": 0.0006,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768809718154319e-05,
+      "loss": 0.0002,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768569645172132e-05,
+      "loss": 0.0005,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.768329572189946e-05,
+      "loss": 0.0062,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76808949920776e-05,
+      "loss": 0.0003,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767849426225573e-05,
+      "loss": 0.0005,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767609353243386e-05,
+      "loss": 0.0003,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7673692802611994e-05,
+      "loss": 0.0006,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.767129207279013e-05,
+      "loss": 0.0028,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7668891342968264e-05,
+      "loss": 0.0011,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7666490613146396e-05,
+      "loss": 0.003,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.766408988332453e-05,
+      "loss": 0.0008,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7661689153502666e-05,
+      "loss": 0.001,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7659288423680805e-05,
+      "loss": 0.0004,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.765688769385894e-05,
+      "loss": 0.0024,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.765448696403707e-05,
+      "loss": 0.0005,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.765208623421521e-05,
+      "loss": 0.0003,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764968550439334e-05,
+      "loss": 0.0008,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764728477457147e-05,
+      "loss": 0.0003,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.76448840447496e-05,
+      "loss": 0.0018,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764248331492774e-05,
+      "loss": 0.0002,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.764008258510587e-05,
+      "loss": 0.0003,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.763768185528401e-05,
+      "loss": 0.0003,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.763528112546214e-05,
+      "loss": 0.0004,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7632880395640275e-05,
+      "loss": 0.5501,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7630479665818414e-05,
+      "loss": 0.0001,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7628078935996545e-05,
+      "loss": 0.0002,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762567820617468e-05,
+      "loss": 0.0002,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762327747635281e-05,
+      "loss": 0.5295,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.762087674653095e-05,
+      "loss": 0.0003,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761847601670908e-05,
+      "loss": 0.0003,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761607528688722e-05,
+      "loss": 0.0002,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761367455706535e-05,
+      "loss": 0.0014,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.761127382724349e-05,
+      "loss": 0.0003,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760887309742162e-05,
+      "loss": 0.0002,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760647236759975e-05,
+      "loss": 1.3254,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7604071637777884e-05,
+      "loss": 0.001,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.760167090795602e-05,
+      "loss": 0.0003,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7599270178134154e-05,
+      "loss": 0.0005,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7596869448312286e-05,
+      "loss": 0.0003,
+      "step": 1051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7594468718490424e-05,
+      "loss": 0.4154,
+      "step": 1052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7592067988668556e-05,
+      "loss": 0.0026,
+      "step": 1053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7589667258846695e-05,
+      "loss": 0.0055,
+      "step": 1054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758726652902483e-05,
+      "loss": 0.0012,
+      "step": 1055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758486579920296e-05,
+      "loss": 0.0017,
+      "step": 1056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758246506938109e-05,
+      "loss": 0.0018,
+      "step": 1057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.758006433955923e-05,
+      "loss": 0.0005,
+      "step": 1058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757766360973736e-05,
+      "loss": 0.334,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757526287991549e-05,
+      "loss": 0.0005,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757286215009363e-05,
+      "loss": 0.0003,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.757046142027177e-05,
+      "loss": 0.0123,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75680606904499e-05,
+      "loss": 0.0005,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.756565996062803e-05,
+      "loss": 0.0006,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7563259230806165e-05,
+      "loss": 0.2414,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7560858500984304e-05,
+      "loss": 0.0003,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7558457771162435e-05,
+      "loss": 0.0054,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755605704134057e-05,
+      "loss": 0.0101,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.75536563115187e-05,
+      "loss": 0.6239,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755125558169684e-05,
+      "loss": 0.0004,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7548854851874976e-05,
+      "loss": 0.0003,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754645412205311e-05,
+      "loss": 0.0006,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754405339223124e-05,
+      "loss": 0.0004,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754165266240938e-05,
+      "loss": 0.0012,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753925193258751e-05,
+      "loss": 0.0004,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753685120276564e-05,
+      "loss": 0.0005,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7534450472943774e-05,
+      "loss": 0.4533,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.753204974312191e-05,
+      "loss": 0.5409,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7529649013300044e-05,
+      "loss": 0.0362,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.752724828347818e-05,
+      "loss": 0.0011,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7524847553656314e-05,
+      "loss": 0.0004,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7522446823834446e-05,
+      "loss": 0.0003,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7520046094012585e-05,
+      "loss": 0.0004,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7517645364190717e-05,
+      "loss": 0.0003,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.751524463436885e-05,
+      "loss": 0.0003,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.751284390454698e-05,
+      "loss": 0.0001,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.751044317472512e-05,
+      "loss": 0.4293,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750804244490325e-05,
+      "loss": 0.0003,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750564171508139e-05,
+      "loss": 0.0002,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750324098525952e-05,
+      "loss": 0.0002,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.750084025543766e-05,
+      "loss": 0.0001,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.749843952561579e-05,
+      "loss": 0.0011,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.749603879579392e-05,
+      "loss": 0.0003,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7493638065972055e-05,
+      "loss": 0.0021,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7491237336150194e-05,
+      "loss": 0.8914,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7488836606328325e-05,
+      "loss": 0.4058,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.748643587650646e-05,
+      "loss": 0.0091,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7484035146684596e-05,
+      "loss": 0.0005,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.748163441686273e-05,
+      "loss": 0.0009,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7479233687040866e-05,
+      "loss": 0.0035,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7476832957219e-05,
+      "loss": 0.1512,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.747443222739713e-05,
+      "loss": 0.0037,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.747203149757526e-05,
+      "loss": 0.0017,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74696307677534e-05,
+      "loss": 0.0016,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.746723003793153e-05,
+      "loss": 0.0018,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7464829308109664e-05,
+      "loss": 0.0024,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.74624285782878e-05,
+      "loss": 0.0006,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.746002784846594e-05,
+      "loss": 0.0035,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.0013,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7455226388822204e-05,
+      "loss": 0.4664,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7452825659000336e-05,
+      "loss": 0.0051,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7450424929178475e-05,
+      "loss": 0.0352,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7448024199356607e-05,
+      "loss": 0.0039,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744562346953474e-05,
+      "loss": 0.0014,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744322273971287e-05,
+      "loss": 0.0003,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.744082200989101e-05,
+      "loss": 0.0005,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743842128006915e-05,
+      "loss": 0.0006,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743602055024728e-05,
+      "loss": 0.0008,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743361982042541e-05,
+      "loss": 0.0132,
+      "step": 1119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.743121909060354e-05,
+      "loss": 0.0005,
+      "step": 1120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742881836078168e-05,
+      "loss": 0.0002,
+      "step": 1121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742641763095981e-05,
+      "loss": 0.0004,
+      "step": 1122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7424016901137945e-05,
+      "loss": 0.0004,
+      "step": 1123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.742161617131608e-05,
+      "loss": 0.0004,
+      "step": 1124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741921544149422e-05,
+      "loss": 0.0017,
+      "step": 1125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7416814711672354e-05,
+      "loss": 0.0015,
+      "step": 1126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7414413981850486e-05,
+      "loss": 0.0002,
+      "step": 1127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.741201325202862e-05,
+      "loss": 0.0004,
+      "step": 1128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7409612522206756e-05,
+      "loss": 0.0002,
+      "step": 1129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740721179238489e-05,
+      "loss": 0.6316,
+      "step": 1130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740481106256302e-05,
+      "loss": 0.0003,
+      "step": 1131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740241033274115e-05,
+      "loss": 0.0004,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740000960291929e-05,
+      "loss": 0.7414,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739760887309743e-05,
+      "loss": 0.0003,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739520814327556e-05,
+      "loss": 0.0002,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.739280741345369e-05,
+      "loss": 0.0005,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7390406683631824e-05,
+      "loss": 0.7002,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.738800595380996e-05,
+      "loss": 0.0005,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7385605223988094e-05,
+      "loss": 0.0002,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7383204494166226e-05,
+      "loss": 0.001,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7380803764344365e-05,
+      "loss": 0.0019,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7378403034522497e-05,
+      "loss": 0.0007,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737600230470063e-05,
+      "loss": 0.0007,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737360157487877e-05,
+      "loss": 0.0005,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73712008450569e-05,
+      "loss": 0.401,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736880011523504e-05,
+      "loss": 0.0049,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736639938541317e-05,
+      "loss": 0.0006,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73639986555913e-05,
+      "loss": 0.0003,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.736159792576943e-05,
+      "loss": 0.0003,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735919719594757e-05,
+      "loss": 0.0007,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.73567964661257e-05,
+      "loss": 0.0024,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7354395736303835e-05,
+      "loss": 0.0003,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7351995006481973e-05,
+      "loss": 0.0013,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7349594276660105e-05,
+      "loss": 0.0007,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7347193546838244e-05,
+      "loss": 0.0006,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7344792817016376e-05,
+      "loss": 0.0009,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.734239208719451e-05,
+      "loss": 0.0006,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7339991357372646e-05,
+      "loss": 0.0006,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733759062755078e-05,
+      "loss": 0.0002,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733518989772891e-05,
+      "loss": 0.0003,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733278916790704e-05,
+      "loss": 0.0013,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.733038843808518e-05,
+      "loss": 0.0003,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732798770826332e-05,
+      "loss": 0.0001,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732558697844145e-05,
+      "loss": 0.0003,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732318624861958e-05,
+      "loss": 0.0229,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7320785518797714e-05,
+      "loss": 0.726,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731838478897585e-05,
+      "loss": 0.0003,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7315984059153984e-05,
+      "loss": 0.001,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7313583329332116e-05,
+      "loss": 0.0003,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.731118259951025e-05,
+      "loss": 0.0002,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730878186968839e-05,
+      "loss": 0.0002,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7306381139866525e-05,
+      "loss": 0.0002,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730398041004466e-05,
+      "loss": 0.0002,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730157968022279e-05,
+      "loss": 0.0003,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729917895040093e-05,
+      "loss": 0.0002,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729677822057906e-05,
+      "loss": 0.0002,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729437749075719e-05,
+      "loss": 0.1164,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.729197676093532e-05,
+      "loss": 0.0002,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.728957603111346e-05,
+      "loss": 0.0007,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72871753012916e-05,
+      "loss": 0.0001,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.728477457146973e-05,
+      "loss": 0.0009,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7282373841647863e-05,
+      "loss": 0.5192,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7279973111825995e-05,
+      "loss": 0.0001,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7277572382004134e-05,
+      "loss": 0.0002,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7275171652182266e-05,
+      "loss": 0.0001,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.72727709223604e-05,
+      "loss": 0.001,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.727037019253853e-05,
+      "loss": 0.0002,
+      "step": 1187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726796946271667e-05,
+      "loss": 0.0001,
+      "step": 1188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7265568732894806e-05,
+      "loss": 0.0002,
+      "step": 1189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726316800307294e-05,
+      "loss": 0.0002,
+      "step": 1190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.726076727325107e-05,
+      "loss": 0.0591,
+      "step": 1191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725836654342921e-05,
+      "loss": 0.1971,
+      "step": 1192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725596581360734e-05,
+      "loss": 0.0004,
+      "step": 1193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.725356508378547e-05,
+      "loss": 0.0004,
+      "step": 1194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7251164353963604e-05,
+      "loss": 0.0003,
+      "step": 1195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724876362414174e-05,
+      "loss": 0.0002,
+      "step": 1196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7246362894319874e-05,
+      "loss": 0.0033,
+      "step": 1197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724396216449801e-05,
+      "loss": 0.0035,
+      "step": 1198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7241561434676145e-05,
+      "loss": 0.0002,
+      "step": 1199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7239160704854276e-05,
+      "loss": 0.0006,
+      "step": 1200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7236759975032415e-05,
+      "loss": 0.0003,
+      "step": 1201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723435924521055e-05,
+      "loss": 0.0004,
+      "step": 1202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.723195851538868e-05,
+      "loss": 0.0002,
+      "step": 1203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722955778556681e-05,
+      "loss": 0.0003,
+      "step": 1204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722715705574495e-05,
+      "loss": 0.0001,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722475632592308e-05,
+      "loss": 0.0001,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.722235559610121e-05,
+      "loss": 0.0001,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721995486627935e-05,
+      "loss": 0.0001,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721755413645749e-05,
+      "loss": 0.0001,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721515340663562e-05,
+      "loss": 0.0001,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7212752676813753e-05,
+      "loss": 0.0001,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7210351946991885e-05,
+      "loss": 0.0001,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7207951217170024e-05,
+      "loss": 0.0241,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7205550487348156e-05,
+      "loss": 0.0001,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720314975752629e-05,
+      "loss": 0.0001,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.720074902770442e-05,
+      "loss": 0.0001,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.719834829788256e-05,
+      "loss": 0.0001,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7195947568060696e-05,
+      "loss": 0.0001,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719354683823883e-05,
+      "loss": 0.0001,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719114610841696e-05,
+      "loss": 0.0001,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718874537859509e-05,
+      "loss": 1.0838,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718634464877323e-05,
+      "loss": 0.967,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.718394391895136e-05,
+      "loss": 0.0002,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7181543189129494e-05,
+      "loss": 0.0003,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717914245930763e-05,
+      "loss": 0.0005,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717674172948577e-05,
+      "loss": 0.0006,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71743409996639e-05,
+      "loss": 0.0009,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7171940269842035e-05,
+      "loss": 0.0002,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7169539540020166e-05,
+      "loss": 0.0006,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7167138810198305e-05,
+      "loss": 0.6113,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.716473808037644e-05,
+      "loss": 0.0004,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.716233735055457e-05,
+      "loss": 0.038,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71599366207327e-05,
+      "loss": 0.0006,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715753589091084e-05,
+      "loss": 0.0007,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715513516108898e-05,
+      "loss": 0.0007,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715273443126711e-05,
+      "loss": 0.8547,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715033370144524e-05,
+      "loss": 0.0008,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714793297162338e-05,
+      "loss": 0.0008,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714553224180151e-05,
+      "loss": 0.0006,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714313151197964e-05,
+      "loss": 0.0005,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7140730782157775e-05,
+      "loss": 0.0009,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7138330052335914e-05,
+      "loss": 0.0035,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7135929322514046e-05,
+      "loss": 0.0013,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7133528592692184e-05,
+      "loss": 0.0024,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7131127862870316e-05,
+      "loss": 0.0006,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712872713304845e-05,
+      "loss": 0.0073,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7126326403226586e-05,
+      "loss": 0.0006,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712392567340472e-05,
+      "loss": 0.0007,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712152494358285e-05,
+      "loss": 0.0006,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711912421376098e-05,
+      "loss": 0.0019,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711672348393912e-05,
+      "loss": 0.0006,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711432275411725e-05,
+      "loss": 0.005,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.711192202429539e-05,
+      "loss": 0.0028,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710952129447352e-05,
+      "loss": 0.0011,
+      "step": 1254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710712056465166e-05,
+      "loss": 0.0049,
+      "step": 1255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.710471983482979e-05,
+      "loss": 0.002,
+      "step": 1256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7102319105007925e-05,
+      "loss": 0.031,
+      "step": 1257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7099918375186056e-05,
+      "loss": 0.0007,
+      "step": 1258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7097517645364195e-05,
+      "loss": 0.0008,
+      "step": 1259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709511691554233e-05,
+      "loss": 0.0027,
+      "step": 1260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.709271618572046e-05,
+      "loss": 0.0006,
+      "step": 1261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70903154558986e-05,
+      "loss": 0.0013,
+      "step": 1262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708791472607673e-05,
+      "loss": 0.0006,
+      "step": 1263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708551399625487e-05,
+      "loss": 0.0006,
+      "step": 1264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7083113266433e-05,
+      "loss": 0.0002,
+      "step": 1265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.708071253661113e-05,
+      "loss": 0.0003,
+      "step": 1266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707831180678926e-05,
+      "loss": 0.0003,
+      "step": 1267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70759110769674e-05,
+      "loss": 0.0005,
+      "step": 1268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707351034714553e-05,
+      "loss": 0.6701,
+      "step": 1269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7071109617323665e-05,
+      "loss": 0.0009,
+      "step": 1270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70687088875018e-05,
+      "loss": 0.0007,
+      "step": 1271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706630815767994e-05,
+      "loss": 0.0029,
+      "step": 1272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7063907427858074e-05,
+      "loss": 0.0005,
+      "step": 1273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7061506698036206e-05,
+      "loss": 0.0002,
+      "step": 1274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705910596821434e-05,
+      "loss": 0.8761,
+      "step": 1275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7056705238392476e-05,
+      "loss": 0.0003,
+      "step": 1276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705430450857061e-05,
+      "loss": 0.5649,
+      "step": 1277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705190377874874e-05,
+      "loss": 0.0003,
+      "step": 1278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704950304892687e-05,
+      "loss": 0.8211,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704710231910501e-05,
+      "loss": 0.0004,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704470158928315e-05,
+      "loss": 0.0005,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704230085946128e-05,
+      "loss": 0.0013,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703990012963941e-05,
+      "loss": 0.0013,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7037499399817544e-05,
+      "loss": 0.0014,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703509866999568e-05,
+      "loss": 0.0043,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7032697940173815e-05,
+      "loss": 0.0022,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7030297210351946e-05,
+      "loss": 0.0007,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.702789648053008e-05,
+      "loss": 0.0015,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.702549575070822e-05,
+      "loss": 0.0013,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7023095020886355e-05,
+      "loss": 0.0014,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.702069429106449e-05,
+      "loss": 0.0016,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.701829356124262e-05,
+      "loss": 0.0021,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.701589283142076e-05,
+      "loss": 0.0007,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.701349210159889e-05,
+      "loss": 0.5286,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.701109137177702e-05,
+      "loss": 0.0014,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.700869064195515e-05,
+      "loss": 0.003,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.700628991213329e-05,
+      "loss": 0.078,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.700388918231142e-05,
+      "loss": 0.0014,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.700148845248956e-05,
+      "loss": 0.0016,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6999087722667694e-05,
+      "loss": 0.0016,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6996686992845825e-05,
+      "loss": 0.0013,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6994286263023964e-05,
+      "loss": 0.0008,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6991885533202096e-05,
+      "loss": 0.0008,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698948480338023e-05,
+      "loss": 0.0003,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698708407355836e-05,
+      "loss": 0.0012,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69846833437365e-05,
+      "loss": 0.9808,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.698228261391463e-05,
+      "loss": 0.0011,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697988188409277e-05,
+      "loss": 0.0088,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69774811542709e-05,
+      "loss": 0.0006,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697508042444904e-05,
+      "loss": 0.0012,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.697267969462717e-05,
+      "loss": 0.0015,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.69702789648053e-05,
+      "loss": 0.7593,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6967878234983434e-05,
+      "loss": 0.0027,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.696547750516157e-05,
+      "loss": 0.0037,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6963076775339705e-05,
+      "loss": 0.0017,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6960676045517836e-05,
+      "loss": 0.0007,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6958275315695975e-05,
+      "loss": 0.0007,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.695587458587411e-05,
+      "loss": 0.0005,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6953473856052245e-05,
+      "loss": 0.0008,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.695107312623038e-05,
+      "loss": 0.0008,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694867239640851e-05,
+      "loss": 0.0043,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694627166658665e-05,
+      "loss": 0.7681,
+      "step": 1322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694387093676478e-05,
+      "loss": 0.9012,
+      "step": 1323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.694147020694291e-05,
+      "loss": 0.0013,
+      "step": 1324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693906947712104e-05,
+      "loss": 0.7551,
+      "step": 1325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693666874729918e-05,
+      "loss": 0.001,
+      "step": 1326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693426801747732e-05,
+      "loss": 0.0008,
+      "step": 1327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.693186728765545e-05,
+      "loss": 0.0291,
+      "step": 1328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6929466557833584e-05,
+      "loss": 0.0053,
+      "step": 1329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6927065828011715e-05,
+      "loss": 0.6186,
+      "step": 1330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6924665098189854e-05,
+      "loss": 0.0012,
+      "step": 1331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6922264368367986e-05,
+      "loss": 0.116,
+      "step": 1332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691986363854612e-05,
+      "loss": 0.0067,
+      "step": 1333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691746290872425e-05,
+      "loss": 0.3727,
+      "step": 1334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691506217890239e-05,
+      "loss": 0.0107,
+      "step": 1335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6912661449080527e-05,
+      "loss": 0.006,
+      "step": 1336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691026071925866e-05,
+      "loss": 0.3567,
+      "step": 1337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690785998943679e-05,
+      "loss": 0.0065,
+      "step": 1338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690545925961493e-05,
+      "loss": 0.007,
+      "step": 1339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690305852979306e-05,
+      "loss": 0.0103,
+      "step": 1340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690065779997119e-05,
+      "loss": 0.0041,
+      "step": 1341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6898257070149324e-05,
+      "loss": 0.0166,
+      "step": 1342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689585634032746e-05,
+      "loss": 0.004,
+      "step": 1343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6893455610505595e-05,
+      "loss": 0.0043,
+      "step": 1344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.689105488068373e-05,
+      "loss": 0.0335,
+      "step": 1345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6888654150861865e-05,
+      "loss": 0.001,
+      "step": 1346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688625342104e-05,
+      "loss": 0.0025,
+      "step": 1347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6883852691218135e-05,
+      "loss": 0.0013,
+      "step": 1348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688145196139627e-05,
+      "loss": 0.04,
+      "step": 1349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68790512315744e-05,
+      "loss": 0.0151,
+      "step": 1350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687665050175253e-05,
+      "loss": 0.0009,
+      "step": 1351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.687424977193067e-05,
+      "loss": 0.0008,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68718490421088e-05,
+      "loss": 0.001,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686944831228694e-05,
+      "loss": 0.0005,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686704758246507e-05,
+      "loss": 0.0007,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686464685264321e-05,
+      "loss": 0.0005,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.686224612282134e-05,
+      "loss": 0.0021,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6859845392999474e-05,
+      "loss": 0.0006,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6857444663177605e-05,
+      "loss": 0.0006,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6855043933355744e-05,
+      "loss": 0.0007,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6852643203533876e-05,
+      "loss": 0.0004,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.685024247371201e-05,
+      "loss": 0.0006,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6847841743890146e-05,
+      "loss": 0.0002,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.684544101406828e-05,
+      "loss": 0.0004,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6843040284246417e-05,
+      "loss": 0.89,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.684063955442455e-05,
+      "loss": 0.0004,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683823882460268e-05,
+      "loss": 0.0018,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683583809478081e-05,
+      "loss": 0.0007,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683343736495895e-05,
+      "loss": 0.0005,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.683103663513708e-05,
+      "loss": 0.0007,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6828635905315214e-05,
+      "loss": 0.2033,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.682623517549335e-05,
+      "loss": 0.0006,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.682383444567149e-05,
+      "loss": 0.0014,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.682143371584962e-05,
+      "loss": 0.0027,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6819032986027755e-05,
+      "loss": 0.0005,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.681663225620589e-05,
+      "loss": 0.0004,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6814231526384025e-05,
+      "loss": 0.0004,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.681183079656216e-05,
+      "loss": 0.0004,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.680943006674029e-05,
+      "loss": 0.0006,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.680702933691842e-05,
+      "loss": 0.0016,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.680462860709656e-05,
+      "loss": 0.0004,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.68022278772747e-05,
+      "loss": 0.0016,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679982714745283e-05,
+      "loss": 0.0006,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679742641763096e-05,
+      "loss": 0.0011,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679502568780909e-05,
+      "loss": 0.0014,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.679262495798723e-05,
+      "loss": 0.0007,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6790224228165364e-05,
+      "loss": 0.0006,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6787823498343495e-05,
+      "loss": 0.0006,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6785422768521634e-05,
+      "loss": 0.0003,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.678302203869977e-05,
+      "loss": 0.0015,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6780621308877904e-05,
+      "loss": 0.0005,
+      "step": 1391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6778220579056036e-05,
+      "loss": 0.0006,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677581984923417e-05,
+      "loss": 0.0013,
+      "step": 1393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6773419119412307e-05,
+      "loss": 0.0005,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.677101838959044e-05,
+      "loss": 0.0005,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676861765976857e-05,
+      "loss": 0.0006,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.67662169299467e-05,
+      "loss": 0.7099,
+      "step": 1397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676381620012484e-05,
+      "loss": 0.0005,
+      "step": 1398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.676141547030297e-05,
+      "loss": 0.0004,
+      "step": 1399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675901474048111e-05,
+      "loss": 0.3319,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675661401065924e-05,
+      "loss": 1.2106,
+      "step": 1401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6754213280837374e-05,
+      "loss": 0.0016,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.675181255101551e-05,
+      "loss": 0.002,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6749411821193645e-05,
+      "loss": 0.0012,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674701109137178e-05,
+      "loss": 0.0014,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6744610361549915e-05,
+      "loss": 0.0024,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.674220963172805e-05,
+      "loss": 0.0025,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673980890190618e-05,
+      "loss": 0.0025,
+      "step": 1408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673740817208432e-05,
+      "loss": 0.0023,
+      "step": 1409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673500744226245e-05,
+      "loss": 0.0016,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673260671244059e-05,
+      "loss": 0.0015,
+      "step": 1411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.673020598261872e-05,
+      "loss": 0.2369,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672780525279685e-05,
+      "loss": 0.0016,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672540452297498e-05,
+      "loss": 0.0012,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.672300379315312e-05,
+      "loss": 0.0012,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6720603063331254e-05,
+      "loss": 0.0012,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6718202333509385e-05,
+      "loss": 0.0013,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6715801603687524e-05,
+      "loss": 0.0015,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671340087386566e-05,
+      "loss": 0.0013,
+      "step": 1419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6711000144043794e-05,
+      "loss": 0.0011,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6708599414221926e-05,
+      "loss": 0.0011,
+      "step": 1421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670619868440006e-05,
+      "loss": 0.0012,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6703797954578197e-05,
+      "loss": 0.0013,
+      "step": 1423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670139722475633e-05,
+      "loss": 0.0012,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669899649493446e-05,
+      "loss": 0.0008,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.669659576511259e-05,
+      "loss": 0.0007,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.669419503529073e-05,
+      "loss": 0.0009,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.669179430546887e-05,
+      "loss": 0.0009,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6689393575647e-05,
+      "loss": 0.0008,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.668699284582513e-05,
+      "loss": 0.0014,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6684592116003264e-05,
+      "loss": 0.0007,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.66821913861814e-05,
+      "loss": 0.0006,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6679790656359535e-05,
+      "loss": 0.0006,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.667738992653767e-05,
+      "loss": 0.0006,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.66749891967158e-05,
+      "loss": 0.8107,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6672588466893944e-05,
+      "loss": 0.8937,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6670187737072076e-05,
+      "loss": 0.0006,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666778700725021e-05,
+      "loss": 0.0007,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666538627742834e-05,
+      "loss": 0.0007,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666298554760648e-05,
+      "loss": 0.0065,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.666058481778461e-05,
+      "loss": 0.0008,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665818408796274e-05,
+      "loss": 0.0007,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665578335814087e-05,
+      "loss": 0.0018,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665338262831901e-05,
+      "loss": 0.0017,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.665098189849715e-05,
+      "loss": 0.0014,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.664858116867528e-05,
+      "loss": 0.0052,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6646180438853414e-05,
+      "loss": 0.0007,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6643779709031546e-05,
+      "loss": 0.0007,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6641378979209684e-05,
+      "loss": 0.0025,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6638978249387816e-05,
+      "loss": 0.0011,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.663657751956595e-05,
+      "loss": 0.0012,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.663417678974408e-05,
+      "loss": 0.0008,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.663177605992222e-05,
+      "loss": 0.0009,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662937533010035e-05,
+      "loss": 0.001,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662697460027849e-05,
+      "loss": 0.0016,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662457387045662e-05,
+      "loss": 0.0014,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.662217314063476e-05,
+      "loss": 0.0025,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661977241081289e-05,
+      "loss": 0.966,
+      "step": 1458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661737168099102e-05,
+      "loss": 0.6515,
+      "step": 1459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6614970951169154e-05,
+      "loss": 0.0013,
+      "step": 1460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.661257022134729e-05,
+      "loss": 0.0011,
+      "step": 1461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6610169491525425e-05,
+      "loss": 0.0014,
+      "step": 1462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660776876170356e-05,
+      "loss": 0.0016,
+      "step": 1463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6605368031881695e-05,
+      "loss": 0.0012,
+      "step": 1464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.660296730205983e-05,
+      "loss": 0.0049,
+      "step": 1465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6600566572237966e-05,
+      "loss": 0.0011,
+      "step": 1466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65981658424161e-05,
+      "loss": 0.0012,
+      "step": 1467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659576511259423e-05,
+      "loss": 0.0013,
+      "step": 1468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.659336438277236e-05,
+      "loss": 0.0012,
+      "step": 1469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65909636529505e-05,
+      "loss": 0.0024,
+      "step": 1470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658856292312863e-05,
+      "loss": 0.0013,
+      "step": 1471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658616219330676e-05,
+      "loss": 0.0017,
+      "step": 1472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65837614634849e-05,
+      "loss": 0.0013,
+      "step": 1473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.658136073366304e-05,
+      "loss": 0.0023,
+      "step": 1474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.657896000384117e-05,
+      "loss": 0.0014,
+      "step": 1475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6576559274019304e-05,
+      "loss": 0.0012,
+      "step": 1476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6574158544197436e-05,
+      "loss": 0.001,
+      "step": 1477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6571757814375574e-05,
+      "loss": 0.0012,
+      "step": 1478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6569357084553706e-05,
+      "loss": 0.0015,
+      "step": 1479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656695635473184e-05,
+      "loss": 0.0017,
+      "step": 1480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656455562490997e-05,
+      "loss": 0.0011,
+      "step": 1481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656215489508811e-05,
+      "loss": 0.0053,
+      "step": 1482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655975416526625e-05,
+      "loss": 0.0012,
+      "step": 1483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655735343544438e-05,
+      "loss": 0.0009,
+      "step": 1484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655495270562251e-05,
+      "loss": 0.0011,
+      "step": 1485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655255197580065e-05,
+      "loss": 0.0011,
+      "step": 1486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.655015124597878e-05,
+      "loss": 0.0008,
+      "step": 1487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654775051615691e-05,
+      "loss": 0.0008,
+      "step": 1488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6545349786335044e-05,
+      "loss": 0.0007,
+      "step": 1489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654294905651318e-05,
+      "loss": 0.4401,
+      "step": 1490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654054832669132e-05,
+      "loss": 0.0007,
+      "step": 1491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653814759686945e-05,
+      "loss": 0.002,
+      "step": 1492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6535746867047585e-05,
+      "loss": 0.0043,
+      "step": 1493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.653334613722572e-05,
+      "loss": 0.0006,
+      "step": 1494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6530945407403856e-05,
+      "loss": 0.0005,
+      "step": 1495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652854467758199e-05,
+      "loss": 0.0006,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.652614394776012e-05,
+      "loss": 0.0005,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.652374321793825e-05,
+      "loss": 0.0006,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.652134248811639e-05,
+      "loss": 0.0007,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651894175829453e-05,
+      "loss": 0.8143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651654102847266e-05,
+      "loss": 0.0193,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651414029865079e-05,
+      "loss": 0.0005,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.651173956882893e-05,
+      "loss": 0.0005,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.650933883900706e-05,
+      "loss": 0.8784,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6506938109185194e-05,
+      "loss": 0.0016,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6504537379363326e-05,
+      "loss": 0.0007,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6502136649541464e-05,
+      "loss": 0.0009,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6499735919719596e-05,
+      "loss": 0.0009,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6497335189897735e-05,
+      "loss": 0.0011,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6494934460075866e-05,
+      "loss": 0.0142,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6492533730254e-05,
+      "loss": 0.0024,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.649013300043214e-05,
+      "loss": 0.0017,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.648773227061027e-05,
+      "loss": 0.0006,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64853315407884e-05,
+      "loss": 0.0012,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.648293081096653e-05,
+      "loss": 0.0008,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.648053008114467e-05,
+      "loss": 0.0062,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64781293513228e-05,
+      "loss": 0.0017,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6475728621500934e-05,
+      "loss": 0.001,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.647332789167907e-05,
+      "loss": 0.8623,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.647092716185721e-05,
+      "loss": 0.0015,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.646852643203534e-05,
+      "loss": 0.8475,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6466125702213475e-05,
+      "loss": 0.0012,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.646372497239161e-05,
+      "loss": 0.0017,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6461324242569746e-05,
+      "loss": 0.001,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645892351274788e-05,
+      "loss": 0.0018,
+      "step": 1525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645652278292601e-05,
+      "loss": 0.0019,
+      "step": 1526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645412205310414e-05,
+      "loss": 0.6659,
+      "step": 1527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.645172132328228e-05,
+      "loss": 0.7143,
+      "step": 1528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644932059346042e-05,
+      "loss": 0.0023,
+      "step": 1529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644691986363855e-05,
+      "loss": 0.0049,
+      "step": 1530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.644451913381668e-05,
+      "loss": 0.0032,
+      "step": 1531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6442118403994813e-05,
+      "loss": 0.0029,
+      "step": 1532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643971767417295e-05,
+      "loss": 0.0036,
+      "step": 1533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6437316944351084e-05,
+      "loss": 0.0048,
+      "step": 1534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6434916214529216e-05,
+      "loss": 0.0035,
+      "step": 1535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643251548470735e-05,
+      "loss": 0.0033,
+      "step": 1536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.643011475488549e-05,
+      "loss": 0.0024,
+      "step": 1537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6427714025063625e-05,
+      "loss": 0.0029,
+      "step": 1538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6425313295241756e-05,
+      "loss": 0.0023,
+      "step": 1539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642291256541989e-05,
+      "loss": 0.0018,
+      "step": 1540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.642051183559803e-05,
+      "loss": 0.0035,
+      "step": 1541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641811110577616e-05,
+      "loss": 0.0022,
+      "step": 1542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641571037595429e-05,
+      "loss": 0.0012,
+      "step": 1543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641330964613242e-05,
+      "loss": 0.0027,
+      "step": 1544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641090891631056e-05,
+      "loss": 0.0031,
+      "step": 1545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.64085081864887e-05,
+      "loss": 0.0023,
+      "step": 1546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640610745666683e-05,
+      "loss": 0.0008,
+      "step": 1547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.640370672684496e-05,
+      "loss": 0.5543,
+      "step": 1548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6401305997023095e-05,
+      "loss": 0.0006,
+      "step": 1549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639890526720123e-05,
+      "loss": 0.0007,
+      "step": 1550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6396504537379365e-05,
+      "loss": 0.0009,
+      "step": 1551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.63941038075575e-05,
+      "loss": 0.0028,
+      "step": 1552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.639170307773563e-05,
+      "loss": 0.0006,
+      "step": 1553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638930234791377e-05,
+      "loss": 0.0005,
+      "step": 1554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6386901618091906e-05,
+      "loss": 0.001,
+      "step": 1555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638450088827004e-05,
+      "loss": 0.0129,
+      "step": 1556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.638210015844817e-05,
+      "loss": 0.0006,
+      "step": 1557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637969942862631e-05,
+      "loss": 0.0005,
+      "step": 1558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637729869880444e-05,
+      "loss": 0.002,
+      "step": 1559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637489796898257e-05,
+      "loss": 0.0014,
+      "step": 1560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6372497239160703e-05,
+      "loss": 0.001,
+      "step": 1561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637009650933884e-05,
+      "loss": 0.0006,
+      "step": 1562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6367695779516974e-05,
+      "loss": 0.0004,
+      "step": 1563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.636529504969511e-05,
+      "loss": 0.0003,
+      "step": 1564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6362894319873244e-05,
+      "loss": 0.0011,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6360493590051376e-05,
+      "loss": 0.0003,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6358092860229515e-05,
+      "loss": 0.0024,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6355692130407646e-05,
+      "loss": 0.0004,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.635329140058578e-05,
+      "loss": 0.0012,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.635089067076392e-05,
+      "loss": 0.0009,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634848994094205e-05,
+      "loss": 0.0002,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634608921112018e-05,
+      "loss": 0.2991,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634368848129832e-05,
+      "loss": 0.0005,
+      "step": 1573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.634128775147645e-05,
+      "loss": 0.0063,
+      "step": 1574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.633888702165459e-05,
+      "loss": 0.0002,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.633648629183272e-05,
+      "loss": 0.0002,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.633408556201085e-05,
+      "loss": 0.0003,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6331684832188985e-05,
+      "loss": 0.0002,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.632928410236712e-05,
+      "loss": 0.0002,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6326883372545255e-05,
+      "loss": 0.0008,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.632448264272339e-05,
+      "loss": 0.0002,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.632208191290152e-05,
+      "loss": 0.0002,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6319681183079664e-05,
+      "loss": 0.0011,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6317280453257796e-05,
+      "loss": 0.0006,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.631487972343593e-05,
+      "loss": 0.0002,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.631247899361406e-05,
+      "loss": 0.0002,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.63100782637922e-05,
+      "loss": 0.0002,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.630767753397033e-05,
+      "loss": 0.0002,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.630527680414846e-05,
+      "loss": 0.0001,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6302876074326593e-05,
+      "loss": 0.0002,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.630047534450473e-05,
+      "loss": 0.0002,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.629807461468287e-05,
+      "loss": 0.0003,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6295673884861e-05,
+      "loss": 0.0002,
+      "step": 1593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6293273155039134e-05,
+      "loss": 0.0002,
+      "step": 1594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6290872425217266e-05,
+      "loss": 0.0005,
+      "step": 1595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6288471695395405e-05,
+      "loss": 2.0346,
+      "step": 1596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6286070965573536e-05,
+      "loss": 0.0017,
+      "step": 1597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.628367023575167e-05,
+      "loss": 0.0002,
+      "step": 1598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62812695059298e-05,
+      "loss": 0.0004,
+      "step": 1599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627886877610794e-05,
+      "loss": 0.0002,
+      "step": 1600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627646804628608e-05,
+      "loss": 0.0004,
+      "step": 1601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627406731646421e-05,
+      "loss": 0.0004,
+      "step": 1602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.627166658664234e-05,
+      "loss": 0.0009,
+      "step": 1603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626926585682048e-05,
+      "loss": 0.0003,
+      "step": 1604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626686512699861e-05,
+      "loss": 0.0005,
+      "step": 1605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.626446439717674e-05,
+      "loss": 0.0006,
+      "step": 1606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6262063667354875e-05,
+      "loss": 0.0003,
+      "step": 1607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625966293753301e-05,
+      "loss": 0.0005,
+      "step": 1608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6257262207711145e-05,
+      "loss": 1.0548,
+      "step": 1609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6254861477889284e-05,
+      "loss": 0.0004,
+      "step": 1610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6252460748067415e-05,
+      "loss": 0.0015,
+      "step": 1611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.625006001824555e-05,
+      "loss": 0.0013,
+      "step": 1612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6247659288423686e-05,
+      "loss": 0.001,
+      "step": 1613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624525855860182e-05,
+      "loss": 0.0017,
+      "step": 1614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624285782877995e-05,
+      "loss": 0.0014,
+      "step": 1615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.624045709895808e-05,
+      "loss": 0.0015,
+      "step": 1616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623805636913622e-05,
+      "loss": 0.0014,
+      "step": 1617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623565563931435e-05,
+      "loss": 0.0014,
+      "step": 1618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623325490949249e-05,
+      "loss": 0.4592,
+      "step": 1619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623085417967062e-05,
+      "loss": 0.0009,
+      "step": 1620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622845344984876e-05,
+      "loss": 0.002,
+      "step": 1621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.622605272002689e-05,
+      "loss": 0.0014,
+      "step": 1622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6223651990205024e-05,
+      "loss": 0.0014,
+      "step": 1623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6221251260383156e-05,
+      "loss": 0.0012,
+      "step": 1624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6218850530561295e-05,
+      "loss": 0.0011,
+      "step": 1625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6216449800739426e-05,
+      "loss": 0.0012,
+      "step": 1626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.621404907091756e-05,
+      "loss": 0.0009,
+      "step": 1627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62116483410957e-05,
+      "loss": 0.0011,
+      "step": 1628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620924761127383e-05,
+      "loss": 0.0016,
+      "step": 1629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620684688145197e-05,
+      "loss": 0.001,
+      "step": 1630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62044461516301e-05,
+      "loss": 0.5908,
+      "step": 1631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.620204542180823e-05,
+      "loss": 0.0009,
+      "step": 1632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619964469198636e-05,
+      "loss": 0.0042,
+      "step": 1633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61972439621645e-05,
+      "loss": 0.0009,
+      "step": 1634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619484323234263e-05,
+      "loss": 0.001,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6192442502520765e-05,
+      "loss": 0.0009,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61900417726989e-05,
+      "loss": 0.0013,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.618764104287704e-05,
+      "loss": 0.7302,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6185240313055174e-05,
+      "loss": 0.0017,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6182839583233305e-05,
+      "loss": 0.0026,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.618043885341144e-05,
+      "loss": 0.0023,
+      "step": 1641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6178038123589576e-05,
+      "loss": 0.002,
+      "step": 1642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.617563739376771e-05,
+      "loss": 0.6718,
+      "step": 1643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.617323666394584e-05,
+      "loss": 0.0031,
+      "step": 1644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.617083593412397e-05,
+      "loss": 0.5842,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616843520430211e-05,
+      "loss": 0.0015,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616603447448025e-05,
+      "loss": 0.0021,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616363374465838e-05,
+      "loss": 0.0028,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.616123301483651e-05,
+      "loss": 0.0013,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6158832285014644e-05,
+      "loss": 0.0014,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.615643155519278e-05,
+      "loss": 0.0014,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6154030825370914e-05,
+      "loss": 0.0014,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6151630095549046e-05,
+      "loss": 0.004,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6149229365727184e-05,
+      "loss": 0.0057,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6146828635905316e-05,
+      "loss": 0.0012,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6144427906083455e-05,
+      "loss": 0.0014,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.614202717626159e-05,
+      "loss": 0.4588,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613962644643972e-05,
+      "loss": 0.1749,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613722571661786e-05,
+      "loss": 0.0008,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613482498679599e-05,
+      "loss": 0.0012,
+      "step": 1660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613242425697412e-05,
+      "loss": 0.0025,
+      "step": 1661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.613002352715225e-05,
+      "loss": 0.002,
+      "step": 1662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.612762279733039e-05,
+      "loss": 0.0027,
+      "step": 1663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.612522206750852e-05,
+      "loss": 0.0008,
+      "step": 1664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.612282133768666e-05,
+      "loss": 0.0007,
+      "step": 1665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.612042060786479e-05,
+      "loss": 0.0008,
+      "step": 1666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611801987804293e-05,
+      "loss": 0.0008,
+      "step": 1667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6115619148221064e-05,
+      "loss": 0.0007,
+      "step": 1668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6113218418399195e-05,
+      "loss": 0.0008,
+      "step": 1669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.611081768857733e-05,
+      "loss": 0.0007,
+      "step": 1670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6108416958755466e-05,
+      "loss": 0.0006,
+      "step": 1671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.61060162289336e-05,
+      "loss": 0.0007,
+      "step": 1672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610361549911173e-05,
+      "loss": 0.0011,
+      "step": 1673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.610121476928987e-05,
+      "loss": 0.0005,
+      "step": 1674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6098814039468e-05,
+      "loss": 0.0017,
+      "step": 1675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609641330964614e-05,
+      "loss": 0.0007,
+      "step": 1676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.609401257982427e-05,
+      "loss": 0.0006,
+      "step": 1677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60916118500024e-05,
+      "loss": 0.588,
+      "step": 1678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6089211120180534e-05,
+      "loss": 0.0005,
+      "step": 1679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.608681039035867e-05,
+      "loss": 0.0046,
+      "step": 1680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6084409660536804e-05,
+      "loss": 0.0019,
+      "step": 1681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6082008930714936e-05,
+      "loss": 0.0005,
+      "step": 1682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6079608200893074e-05,
+      "loss": 0.0031,
+      "step": 1683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607720747107121e-05,
+      "loss": 0.0023,
+      "step": 1684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6074806741249345e-05,
+      "loss": 0.0035,
+      "step": 1685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607240601142748e-05,
+      "loss": 0.0026,
+      "step": 1686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.607000528160561e-05,
+      "loss": 0.0004,
+      "step": 1687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606760455178375e-05,
+      "loss": 0.0005,
+      "step": 1688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606520382196188e-05,
+      "loss": 0.0024,
+      "step": 1689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606280309214001e-05,
+      "loss": 0.0004,
+      "step": 1690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606040236231814e-05,
+      "loss": 0.0015,
+      "step": 1691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605800163249628e-05,
+      "loss": 0.0006,
+      "step": 1692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605560090267442e-05,
+      "loss": 0.0004,
+      "step": 1693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605320017285255e-05,
+      "loss": 0.0013,
+      "step": 1694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.605079944303068e-05,
+      "loss": 0.3853,
+      "step": 1695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6048398713208815e-05,
+      "loss": 0.0004,
+      "step": 1696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6045997983386954e-05,
+      "loss": 0.4843,
+      "step": 1697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6043597253565085e-05,
+      "loss": 0.0008,
+      "step": 1698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.604119652374322e-05,
+      "loss": 0.0004,
+      "step": 1699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603879579392135e-05,
+      "loss": 0.0004,
+      "step": 1700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6036395064099494e-05,
+      "loss": 0.0004,
+      "step": 1701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6033994334277626e-05,
+      "loss": 0.0033,
+      "step": 1702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.603159360445576e-05,
+      "loss": 0.257,
+      "step": 1703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.602919287463389e-05,
+      "loss": 0.0008,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.602679214481203e-05,
+      "loss": 0.0003,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.602439141499016e-05,
+      "loss": 0.0004,
+      "step": 1706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.602199068516829e-05,
+      "loss": 0.0003,
+      "step": 1707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6019589955346424e-05,
+      "loss": 0.0003,
+      "step": 1708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.601718922552456e-05,
+      "loss": 0.0047,
+      "step": 1709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6014788495702694e-05,
+      "loss": 0.0034,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.601238776588083e-05,
+      "loss": 0.0004,
+      "step": 1711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6009987036058964e-05,
+      "loss": 0.0051,
+      "step": 1712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6007586306237096e-05,
+      "loss": 0.0003,
+      "step": 1713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6005185576415235e-05,
+      "loss": 0.045,
+      "step": 1714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6002784846593367e-05,
+      "loss": 0.0006,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.60003841167715e-05,
+      "loss": 0.6825,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.599798338694963e-05,
+      "loss": 0.2475,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.599558265712777e-05,
+      "loss": 0.0003,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59931819273059e-05,
+      "loss": 0.002,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.599078119748404e-05,
+      "loss": 0.0017,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598838046766217e-05,
+      "loss": 0.0003,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598597973784031e-05,
+      "loss": 0.0003,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598357900801844e-05,
+      "loss": 0.0002,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.598117827819657e-05,
+      "loss": 0.0003,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5978777548374705e-05,
+      "loss": 0.0003,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5976376818552844e-05,
+      "loss": 0.0003,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5973976088730975e-05,
+      "loss": 0.0003,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.597157535890911e-05,
+      "loss": 0.0003,
+      "step": 1728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5969174629087246e-05,
+      "loss": 0.0003,
+      "step": 1729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.596677389926538e-05,
+      "loss": 0.0003,
+      "step": 1730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5964373169443516e-05,
+      "loss": 0.0003,
+      "step": 1731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.596197243962165e-05,
+      "loss": 0.0003,
+      "step": 1732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595957170979978e-05,
+      "loss": 0.0003,
+      "step": 1733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595717097997792e-05,
+      "loss": 0.0002,
+      "step": 1734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595477025015605e-05,
+      "loss": 0.0003,
+      "step": 1735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.595236952033418e-05,
+      "loss": 0.9968,
+      "step": 1736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5949968790512314e-05,
+      "loss": 0.003,
+      "step": 1737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594756806069045e-05,
+      "loss": 0.0003,
+      "step": 1738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594516733086859e-05,
+      "loss": 0.0016,
+      "step": 1739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.594276660104672e-05,
+      "loss": 0.0198,
+      "step": 1740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5940365871224854e-05,
+      "loss": 0.0004,
+      "step": 1741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5937965141402986e-05,
+      "loss": 0.0005,
+      "step": 1742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5935564411581125e-05,
+      "loss": 0.0005,
+      "step": 1743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5933163681759257e-05,
+      "loss": 0.0005,
+      "step": 1744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.593076295193739e-05,
+      "loss": 0.0006,
+      "step": 1745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592836222211552e-05,
+      "loss": 0.0004,
+      "step": 1746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592596149229366e-05,
+      "loss": 0.0004,
+      "step": 1747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59235607624718e-05,
+      "loss": 0.0006,
+      "step": 1748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.592116003264993e-05,
+      "loss": 0.0004,
+      "step": 1749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591875930282806e-05,
+      "loss": 0.0015,
+      "step": 1750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.59163585730062e-05,
+      "loss": 0.5841,
+      "step": 1751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591395784318433e-05,
+      "loss": 0.0006,
+      "step": 1752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.591155711336246e-05,
+      "loss": 0.0014,
+      "step": 1753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5909156383540595e-05,
+      "loss": 0.0028,
+      "step": 1754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5906755653718733e-05,
+      "loss": 0.0008,
+      "step": 1755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590435492389687e-05,
+      "loss": 0.0011,
+      "step": 1756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5901954194075004e-05,
+      "loss": 0.0005,
+      "step": 1757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5899553464253136e-05,
+      "loss": 0.0008,
+      "step": 1758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589715273443127e-05,
+      "loss": 0.0005,
+      "step": 1759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5894752004609406e-05,
+      "loss": 0.0037,
+      "step": 1760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.589235127478754e-05,
+      "loss": 0.0058,
+      "step": 1761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588995054496567e-05,
+      "loss": 0.0004,
+      "step": 1762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58875498151438e-05,
+      "loss": 0.0005,
+      "step": 1763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588514908532194e-05,
+      "loss": 0.0004,
+      "step": 1764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588274835550007e-05,
+      "loss": 0.0004,
+      "step": 1765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588034762567821e-05,
+      "loss": 0.0004,
+      "step": 1766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587794689585634e-05,
+      "loss": 0.0005,
+      "step": 1767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587554616603448e-05,
+      "loss": 0.0004,
+      "step": 1768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587314543621261e-05,
+      "loss": 0.0005,
+      "step": 1769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5870744706390744e-05,
+      "loss": 0.0041,
+      "step": 1770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5868343976568876e-05,
+      "loss": 0.0008,
+      "step": 1771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5865943246747015e-05,
+      "loss": 0.0004,
+      "step": 1772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5863542516925147e-05,
+      "loss": 0.0004,
+      "step": 1773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586114178710328e-05,
+      "loss": 0.0004,
+      "step": 1774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.585874105728142e-05,
+      "loss": 0.0003,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.585634032745955e-05,
+      "loss": 0.0008,
+      "step": 1776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.585393959763769e-05,
+      "loss": 0.0004,
+      "step": 1777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.585153886781582e-05,
+      "loss": 0.0017,
+      "step": 1778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584913813799395e-05,
+      "loss": 0.0004,
+      "step": 1779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584673740817208e-05,
+      "loss": 0.002,
+      "step": 1780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584433667835022e-05,
+      "loss": 0.0007,
+      "step": 1781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.584193594852835e-05,
+      "loss": 0.0003,
+      "step": 1782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5839535218706485e-05,
+      "loss": 0.0003,
+      "step": 1783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5837134488884623e-05,
+      "loss": 0.0004,
+      "step": 1784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.583473375906276e-05,
+      "loss": 0.0035,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5832333029240894e-05,
+      "loss": 0.001,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5829932299419026e-05,
+      "loss": 0.5765,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.582753156959716e-05,
+      "loss": 0.0003,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5825130839775296e-05,
+      "loss": 0.0003,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.582273010995343e-05,
+      "loss": 0.0003,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.582032938013156e-05,
+      "loss": 0.0003,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.581792865030969e-05,
+      "loss": 0.0003,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.581552792048783e-05,
+      "loss": 0.5692,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.581312719066597e-05,
+      "loss": 0.0087,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.58107264608441e-05,
+      "loss": 0.0003,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.580832573102223e-05,
+      "loss": 0.0003,
+      "step": 1796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5805925001200364e-05,
+      "loss": 0.0046,
+      "step": 1797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.58035242713785e-05,
+      "loss": 0.0003,
+      "step": 1798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5801123541556634e-05,
+      "loss": 0.0003,
+      "step": 1799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5798722811734766e-05,
+      "loss": 0.0003,
+      "step": 1800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57963220819129e-05,
+      "loss": 0.3632,
+      "step": 1801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.579392135209104e-05,
+      "loss": 0.0003,
+      "step": 1802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5791520622269175e-05,
+      "loss": 0.003,
+      "step": 1803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578911989244731e-05,
+      "loss": 0.0003,
+      "step": 1804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578671916262544e-05,
+      "loss": 0.205,
+      "step": 1805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578431843280358e-05,
+      "loss": 0.0003,
+      "step": 1806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.578191770298171e-05,
+      "loss": 0.0003,
+      "step": 1807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577951697315984e-05,
+      "loss": 0.0004,
+      "step": 1808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577711624333797e-05,
+      "loss": 0.0018,
+      "step": 1809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577471551351611e-05,
+      "loss": 0.0003,
+      "step": 1810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.577231478369425e-05,
+      "loss": 0.0002,
+      "step": 1811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.576991405387238e-05,
+      "loss": 0.0006,
+      "step": 1812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5767513324050513e-05,
+      "loss": 0.0003,
+      "step": 1813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5765112594228645e-05,
+      "loss": 0.0003,
+      "step": 1814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5762711864406784e-05,
+      "loss": 0.0003,
+      "step": 1815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5760311134584916e-05,
+      "loss": 0.0007,
+      "step": 1816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575791040476305e-05,
+      "loss": 0.0003,
+      "step": 1817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5755509674941186e-05,
+      "loss": 0.2443,
+      "step": 1818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.575310894511932e-05,
+      "loss": 0.0003,
+      "step": 1819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5750708215297456e-05,
+      "loss": 0.444,
+      "step": 1820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574830748547559e-05,
+      "loss": 0.0002,
+      "step": 1821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574590675565372e-05,
+      "loss": 0.0002,
+      "step": 1822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574350602583186e-05,
+      "loss": 0.0002,
+      "step": 1823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.574110529600999e-05,
+      "loss": 0.0003,
+      "step": 1824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573870456618812e-05,
+      "loss": 0.0002,
+      "step": 1825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5736303836366254e-05,
+      "loss": 0.0016,
+      "step": 1826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.573390310654439e-05,
+      "loss": 0.0003,
+      "step": 1827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5731502376722524e-05,
+      "loss": 0.4617,
+      "step": 1828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5729101646900656e-05,
+      "loss": 0.0012,
+      "step": 1829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5726700917078795e-05,
+      "loss": 0.0004,
+      "step": 1830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5724300187256926e-05,
+      "loss": 0.0003,
+      "step": 1831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5721899457435065e-05,
+      "loss": 0.0002,
+      "step": 1832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57194987276132e-05,
+      "loss": 0.0002,
+      "step": 1833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571709799779133e-05,
+      "loss": 0.0002,
+      "step": 1834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571469726796947e-05,
+      "loss": 0.0002,
+      "step": 1835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.57122965381476e-05,
+      "loss": 0.3936,
+      "step": 1836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570989580832573e-05,
+      "loss": 0.0002,
+      "step": 1837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570749507850386e-05,
+      "loss": 1.044,
+      "step": 1838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5705094348682e-05,
+      "loss": 0.0003,
+      "step": 1839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570269361886014e-05,
+      "loss": 0.9444,
+      "step": 1840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.570029288903827e-05,
+      "loss": 0.0012,
+      "step": 1841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5697892159216403e-05,
+      "loss": 0.0013,
+      "step": 1842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5695491429394535e-05,
+      "loss": 0.0061,
+      "step": 1843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5693090699572674e-05,
+      "loss": 0.0094,
+      "step": 1844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5690689969750806e-05,
+      "loss": 0.2951,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.568828923992894e-05,
+      "loss": 0.0069,
+      "step": 1846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.568588851010707e-05,
+      "loss": 0.0033,
+      "step": 1847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5683487780285215e-05,
+      "loss": 0.0078,
+      "step": 1848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5681087050463346e-05,
+      "loss": 0.0028,
+      "step": 1849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567868632064148e-05,
+      "loss": 0.0024,
+      "step": 1850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567628559081961e-05,
+      "loss": 0.0021,
+      "step": 1851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567388486099775e-05,
+      "loss": 0.0072,
+      "step": 1852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.567148413117588e-05,
+      "loss": 0.0014,
+      "step": 1853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.566908340135401e-05,
+      "loss": 0.6331,
+      "step": 1854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5666682671532144e-05,
+      "loss": 0.7941,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.566428194171028e-05,
+      "loss": 0.0022,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.566188121188842e-05,
+      "loss": 0.0149,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.565948048206655e-05,
+      "loss": 0.0025,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5657079752244685e-05,
+      "loss": 0.4367,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5654679022422816e-05,
+      "loss": 0.1561,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5652278292600955e-05,
+      "loss": 0.007,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564987756277909e-05,
+      "loss": 0.0034,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564747683295722e-05,
+      "loss": 0.0036,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564507610313535e-05,
+      "loss": 0.0082,
+      "step": 1864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564267537331349e-05,
+      "loss": 0.0062,
+      "step": 1865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.564027464349163e-05,
+      "loss": 0.0044,
+      "step": 1866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563787391366976e-05,
+      "loss": 0.0026,
+      "step": 1867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563547318384789e-05,
+      "loss": 0.0007,
+      "step": 1868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563307245402603e-05,
+      "loss": 0.0016,
+      "step": 1869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.563067172420416e-05,
+      "loss": 0.001,
+      "step": 1870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.562827099438229e-05,
+      "loss": 0.0011,
+      "step": 1871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5625870264560425e-05,
+      "loss": 0.002,
+      "step": 1872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5623469534738564e-05,
+      "loss": 0.0011,
+      "step": 1873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5621068804916696e-05,
+      "loss": 0.001,
+      "step": 1874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5618668075094834e-05,
+      "loss": 0.0012,
+      "step": 1875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5616267345272966e-05,
+      "loss": 0.0006,
+      "step": 1876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56138666154511e-05,
+      "loss": 0.0008,
+      "step": 1877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5611465885629236e-05,
+      "loss": 0.0007,
+      "step": 1878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560906515580737e-05,
+      "loss": 0.0024,
+      "step": 1879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.56066644259855e-05,
+      "loss": 0.0008,
+      "step": 1880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560426369616363e-05,
+      "loss": 0.0006,
+      "step": 1881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.560186296634177e-05,
+      "loss": 0.0009,
+      "step": 1882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55994622365199e-05,
+      "loss": 0.001,
+      "step": 1883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559706150669804e-05,
+      "loss": 0.9464,
+      "step": 1884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559466077687617e-05,
+      "loss": 0.0005,
+      "step": 1885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.559226004705431e-05,
+      "loss": 0.0012,
+      "step": 1886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558985931723244e-05,
+      "loss": 0.0018,
+      "step": 1887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5587458587410575e-05,
+      "loss": 0.8318,
+      "step": 1888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5585057857588706e-05,
+      "loss": 0.001,
+      "step": 1889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5582657127766845e-05,
+      "loss": 0.0013,
+      "step": 1890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.558025639794498e-05,
+      "loss": 0.0013,
+      "step": 1891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557785566812311e-05,
+      "loss": 0.0017,
+      "step": 1892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557545493830124e-05,
+      "loss": 0.0014,
+      "step": 1893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557305420847938e-05,
+      "loss": 0.0016,
+      "step": 1894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557065347865752e-05,
+      "loss": 0.0014,
+      "step": 1895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556825274883565e-05,
+      "loss": 0.0315,
+      "step": 1896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556585201901378e-05,
+      "loss": 0.0016,
+      "step": 1897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556345128919191e-05,
+      "loss": 0.0261,
+      "step": 1898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.556105055937005e-05,
+      "loss": 0.0017,
+      "step": 1899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555864982954818e-05,
+      "loss": 0.0017,
+      "step": 1900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5556249099726315e-05,
+      "loss": 0.0016,
+      "step": 1901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5553848369904454e-05,
+      "loss": 0.002,
+      "step": 1902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.555144764008259e-05,
+      "loss": 0.002,
+      "step": 1903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5549046910260724e-05,
+      "loss": 0.0018,
+      "step": 1904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5546646180438856e-05,
+      "loss": 0.0014,
+      "step": 1905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.554424545061699e-05,
+      "loss": 0.0012,
+      "step": 1906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5541844720795126e-05,
+      "loss": 0.6761,
+      "step": 1907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553944399097326e-05,
+      "loss": 0.0013,
+      "step": 1908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553704326115139e-05,
+      "loss": 0.0012,
+      "step": 1909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553464253132952e-05,
+      "loss": 0.0016,
+      "step": 1910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553224180150766e-05,
+      "loss": 0.0017,
+      "step": 1911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.55298410716858e-05,
+      "loss": 0.0022,
+      "step": 1912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.552744034186393e-05,
+      "loss": 0.7158,
+      "step": 1913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.552503961204206e-05,
+      "loss": 0.001,
+      "step": 1914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.55226388822202e-05,
+      "loss": 0.0013,
+      "step": 1915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.552023815239833e-05,
+      "loss": 0.7841,
+      "step": 1916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5517837422576465e-05,
+      "loss": 0.0012,
+      "step": 1917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5515436692754596e-05,
+      "loss": 0.0022,
+      "step": 1918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5513035962932735e-05,
+      "loss": 0.0012,
+      "step": 1919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.551063523311087e-05,
+      "loss": 0.0017,
+      "step": 1920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5508234503289005e-05,
+      "loss": 0.6719,
+      "step": 1921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.550583377346714e-05,
+      "loss": 0.0025,
+      "step": 1922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.550343304364527e-05,
+      "loss": 0.0035,
+      "step": 1923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.550103231382341e-05,
+      "loss": 0.0034,
+      "step": 1924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.549863158400154e-05,
+      "loss": 0.7139,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.549623085417967e-05,
+      "loss": 0.0042,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54938301243578e-05,
+      "loss": 0.0045,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.549142939453594e-05,
+      "loss": 0.0065,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.548902866471407e-05,
+      "loss": 0.0056,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.548662793489221e-05,
+      "loss": 0.0058,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5484227205070344e-05,
+      "loss": 0.9682,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.548182647524848e-05,
+      "loss": 0.0048,
+      "step": 1932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5479425745426614e-05,
+      "loss": 0.007,
+      "step": 1933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5477025015604746e-05,
+      "loss": 0.0052,
+      "step": 1934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.547462428578288e-05,
+      "loss": 0.0025,
+      "step": 1935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5472223555961016e-05,
+      "loss": 0.0111,
+      "step": 1936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546982282613915e-05,
+      "loss": 0.2134,
+      "step": 1937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546742209631728e-05,
+      "loss": 0.6276,
+      "step": 1938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546502136649542e-05,
+      "loss": 0.0029,
+      "step": 1939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546262063667355e-05,
+      "loss": 0.0021,
+      "step": 1940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.546021990685169e-05,
+      "loss": 0.0025,
+      "step": 1941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545781917702982e-05,
+      "loss": 0.0023,
+      "step": 1942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545541844720795e-05,
+      "loss": 0.0513,
+      "step": 1943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5453017717386084e-05,
+      "loss": 0.0021,
+      "step": 1944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.545061698756422e-05,
+      "loss": 0.4751,
+      "step": 1945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5448216257742355e-05,
+      "loss": 0.0013,
+      "step": 1946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5445815527920486e-05,
+      "loss": 0.0033,
+      "step": 1947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5443414798098625e-05,
+      "loss": 0.0019,
+      "step": 1948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5441014068276764e-05,
+      "loss": 0.0044,
+      "step": 1949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5438613338454895e-05,
+      "loss": 0.0011,
+      "step": 1950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543621260863303e-05,
+      "loss": 0.0078,
+      "step": 1951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.543381187881116e-05,
+      "loss": 0.0009,
+      "step": 1952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54314111489893e-05,
+      "loss": 0.0014,
+      "step": 1953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542901041916743e-05,
+      "loss": 0.0015,
+      "step": 1954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542660968934556e-05,
+      "loss": 0.0007,
+      "step": 1955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542420895952369e-05,
+      "loss": 0.0008,
+      "step": 1956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.542180822970183e-05,
+      "loss": 0.0022,
+      "step": 1957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.541940749987997e-05,
+      "loss": 0.0007,
+      "step": 1958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54170067700581e-05,
+      "loss": 0.0006,
+      "step": 1959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5414606040236234e-05,
+      "loss": 0.0008,
+      "step": 1960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5412205310414365e-05,
+      "loss": 0.0011,
+      "step": 1961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5409804580592504e-05,
+      "loss": 0.0006,
+      "step": 1962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5407403850770636e-05,
+      "loss": 0.0007,
+      "step": 1963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540500312094877e-05,
+      "loss": 0.0012,
+      "step": 1964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.54026023911269e-05,
+      "loss": 0.0013,
+      "step": 1965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540020166130504e-05,
+      "loss": 0.5785,
+      "step": 1966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5397800931483177e-05,
+      "loss": 0.0005,
+      "step": 1967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539540020166131e-05,
+      "loss": 0.8889,
+      "step": 1968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539299947183944e-05,
+      "loss": 0.0013,
+      "step": 1969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.539059874201758e-05,
+      "loss": 0.001,
+      "step": 1970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538819801219571e-05,
+      "loss": 0.0016,
+      "step": 1971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538579728237384e-05,
+      "loss": 0.0029,
+      "step": 1972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5383396552551974e-05,
+      "loss": 0.0055,
+      "step": 1973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538099582273011e-05,
+      "loss": 0.003,
+      "step": 1974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5378595092908245e-05,
+      "loss": 0.0014,
+      "step": 1975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537619436308638e-05,
+      "loss": 0.0003,
+      "step": 1976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5373793633264515e-05,
+      "loss": 0.0026,
+      "step": 1977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.537139290344265e-05,
+      "loss": 0.6776,
+      "step": 1978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5368992173620785e-05,
+      "loss": 0.0022,
+      "step": 1979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536659144379892e-05,
+      "loss": 0.0037,
+      "step": 1980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536419071397705e-05,
+      "loss": 0.0036,
+      "step": 1981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.536178998415519e-05,
+      "loss": 0.0055,
+      "step": 1982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535938925433332e-05,
+      "loss": 0.0025,
+      "step": 1983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.535698852451145e-05,
+      "loss": 0.0054,
+      "step": 1984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.535458779468959e-05,
+      "loss": 0.0023,
+      "step": 1985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.535218706486772e-05,
+      "loss": 0.0341,
+      "step": 1986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.534978633504586e-05,
+      "loss": 0.0042,
+      "step": 1987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.534738560522399e-05,
+      "loss": 0.0014,
+      "step": 1988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5344984875402124e-05,
+      "loss": 0.6782,
+      "step": 1989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5342584145580255e-05,
+      "loss": 0.0034,
+      "step": 1990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5340183415758394e-05,
+      "loss": 0.0029,
+      "step": 1991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5337782685936526e-05,
+      "loss": 0.003,
+      "step": 1992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.533538195611466e-05,
+      "loss": 0.0032,
+      "step": 1993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5332981226292796e-05,
+      "loss": 0.0023,
+      "step": 1994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.533058049647093e-05,
+      "loss": 0.0029,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5328179766649067e-05,
+      "loss": 0.0031,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.53257790368272e-05,
+      "loss": 0.0026,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.532337830700533e-05,
+      "loss": 0.0016,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.532097757718347e-05,
+      "loss": 0.0006,
+      "step": 1999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.53185768473616e-05,
+      "loss": 0.0012,
+      "step": 2000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.531617611753973e-05,
+      "loss": 0.0005,
+      "step": 2001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5313775387717864e-05,
+      "loss": 0.0004,
+      "step": 2002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5311374657896e-05,
+      "loss": 0.7667,
+      "step": 2003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.530897392807414e-05,
+      "loss": 0.0007,
+      "step": 2004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.530657319825227e-05,
+      "loss": 0.0005,
+      "step": 2005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5304172468430405e-05,
+      "loss": 0.0012,
+      "step": 2006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.530177173860854e-05,
+      "loss": 0.0005,
+      "step": 2007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5299371008786675e-05,
+      "loss": 0.0005,
+      "step": 2008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.529697027896481e-05,
+      "loss": 0.0005,
+      "step": 2009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.529456954914294e-05,
+      "loss": 0.0006,
+      "step": 2010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.529216881932107e-05,
+      "loss": 0.0004,
+      "step": 2011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5289768089499216e-05,
+      "loss": 0.0007,
+      "step": 2012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528736735967735e-05,
+      "loss": 0.0007,
+      "step": 2013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528496662985548e-05,
+      "loss": 0.7591,
+      "step": 2014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528256590003361e-05,
+      "loss": 0.0009,
+      "step": 2015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.528016517021175e-05,
+      "loss": 0.0004,
+      "step": 2016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.527776444038988e-05,
+      "loss": 0.0007,
+      "step": 2017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5275363710568014e-05,
+      "loss": 0.001,
+      "step": 2018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5272962980746145e-05,
+      "loss": 0.0005,
+      "step": 2019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5270562250924284e-05,
+      "loss": 0.4467,
+      "step": 2020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5268161521102416e-05,
+      "loss": 0.0004,
+      "step": 2021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5265760791280554e-05,
+      "loss": 0.0004,
+      "step": 2022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5263360061458686e-05,
+      "loss": 0.0005,
+      "step": 2023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.526095933163682e-05,
+      "loss": 0.0009,
+      "step": 2024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5258558601814957e-05,
+      "loss": 0.0013,
+      "step": 2025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525615787199309e-05,
+      "loss": 0.0011,
+      "step": 2026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525375714217122e-05,
+      "loss": 0.0011,
+      "step": 2027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.525135641234935e-05,
+      "loss": 0.0007,
+      "step": 2028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524895568252749e-05,
+      "loss": 0.0015,
+      "step": 2029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524655495270562e-05,
+      "loss": 0.0003,
+      "step": 2030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524415422288376e-05,
+      "loss": 0.0009,
+      "step": 2031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.524175349306189e-05,
+      "loss": 0.0004,
+      "step": 2032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523935276324003e-05,
+      "loss": 0.0004,
+      "step": 2033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523695203341816e-05,
+      "loss": 0.0004,
+      "step": 2034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5234551303596295e-05,
+      "loss": 0.0011,
+      "step": 2035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523215057377443e-05,
+      "loss": 0.0007,
+      "step": 2036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5229749843952565e-05,
+      "loss": 0.0019,
+      "step": 2037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52273491141307e-05,
+      "loss": 0.0002,
+      "step": 2038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522494838430883e-05,
+      "loss": 0.0003,
+      "step": 2039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.522254765448697e-05,
+      "loss": 0.0003,
+      "step": 2040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52201469246651e-05,
+      "loss": 0.0014,
+      "step": 2041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521774619484324e-05,
+      "loss": 0.0003,
+      "step": 2042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521534546502137e-05,
+      "loss": 0.0009,
+      "step": 2043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.52129447351995e-05,
+      "loss": 0.0003,
+      "step": 2044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.521054400537763e-05,
+      "loss": 0.0002,
+      "step": 2045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520814327555577e-05,
+      "loss": 0.0003,
+      "step": 2046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5205742545733904e-05,
+      "loss": 0.0007,
+      "step": 2047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5203341815912035e-05,
+      "loss": 0.0012,
+      "step": 2048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5200941086090174e-05,
+      "loss": 0.0002,
+      "step": 2049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.519854035626831e-05,
+      "loss": 0.0006,
+      "step": 2050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5196139626446444e-05,
+      "loss": 1.0217,
+      "step": 2051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5193738896624576e-05,
+      "loss": 0.0007,
+      "step": 2052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.519133816680271e-05,
+      "loss": 0.0003,
+      "step": 2053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5188937436980847e-05,
+      "loss": 0.0006,
+      "step": 2054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.518653670715898e-05,
+      "loss": 0.0008,
+      "step": 2055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.518413597733711e-05,
+      "loss": 0.0011,
+      "step": 2056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.518173524751524e-05,
+      "loss": 0.0004,
+      "step": 2057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517933451769338e-05,
+      "loss": 0.0013,
+      "step": 2058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517693378787152e-05,
+      "loss": 0.0013,
+      "step": 2059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517453305804965e-05,
+      "loss": 0.0007,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.517213232822778e-05,
+      "loss": 0.0008,
+      "step": 2061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5169731598405914e-05,
+      "loss": 0.0008,
+      "step": 2062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.516733086858405e-05,
+      "loss": 0.0012,
+      "step": 2063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5164930138762185e-05,
+      "loss": 0.0089,
+      "step": 2064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.516252940894032e-05,
+      "loss": 0.4638,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5160128679118455e-05,
+      "loss": 0.0012,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5157727949296594e-05,
+      "loss": 0.0006,
+      "step": 2067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5155327219474726e-05,
+      "loss": 0.0012,
+      "step": 2068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.515292648965286e-05,
+      "loss": 0.0005,
+      "step": 2069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.515052575983099e-05,
+      "loss": 0.0003,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.514812503000913e-05,
+      "loss": 0.0049,
+      "step": 2071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.514572430018726e-05,
+      "loss": 0.0006,
+      "step": 2072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.514332357036539e-05,
+      "loss": 0.0007,
+      "step": 2073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.514092284054352e-05,
+      "loss": 0.0017,
+      "step": 2074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.513852211072166e-05,
+      "loss": 0.0004,
+      "step": 2075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51361213808998e-05,
+      "loss": 0.0011,
+      "step": 2076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.513372065107793e-05,
+      "loss": 0.009,
+      "step": 2077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5131319921256064e-05,
+      "loss": 0.0005,
+      "step": 2078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5128919191434196e-05,
+      "loss": 0.0003,
+      "step": 2079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5126518461612334e-05,
+      "loss": 0.0008,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5124117731790466e-05,
+      "loss": 0.0007,
+      "step": 2081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.51217170019686e-05,
+      "loss": 0.0003,
+      "step": 2082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5119316272146736e-05,
+      "loss": 0.1061,
+      "step": 2083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511691554232487e-05,
+      "loss": 0.0006,
+      "step": 2084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5114514812503e-05,
+      "loss": 0.0007,
+      "step": 2085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.511211408268114e-05,
+      "loss": 0.0013,
+      "step": 2086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510971335285927e-05,
+      "loss": 0.0007,
+      "step": 2087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510731262303741e-05,
+      "loss": 0.0005,
+      "step": 2088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510491189321554e-05,
+      "loss": 0.0007,
+      "step": 2089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.510251116339367e-05,
+      "loss": 0.0002,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5100110433571804e-05,
+      "loss": 0.0003,
+      "step": 2091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509770970374994e-05,
+      "loss": 0.0004,
+      "step": 2092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5095308973928075e-05,
+      "loss": 0.0003,
+      "step": 2093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.509290824410621e-05,
+      "loss": 0.0004,
+      "step": 2094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5090507514284345e-05,
+      "loss": 0.0004,
+      "step": 2095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5088106784462484e-05,
+      "loss": 0.0002,
+      "step": 2096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5085706054640616e-05,
+      "loss": 0.0002,
+      "step": 2097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508330532481875e-05,
+      "loss": 0.0003,
+      "step": 2098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.508090459499688e-05,
+      "loss": 0.0004,
+      "step": 2099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507850386517502e-05,
+      "loss": 0.0002,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507610313535315e-05,
+      "loss": 0.0002,
+      "step": 2101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507370240553128e-05,
+      "loss": 0.0004,
+      "step": 2102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507130167570941e-05,
+      "loss": 0.0002,
+      "step": 2103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506890094588755e-05,
+      "loss": 0.0003,
+      "step": 2104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506650021606569e-05,
+      "loss": 0.0002,
+      "step": 2105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506409948624382e-05,
+      "loss": 0.0004,
+      "step": 2106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5061698756421954e-05,
+      "loss": 0.0003,
+      "step": 2107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5059298026600086e-05,
+      "loss": 0.0573,
+      "step": 2108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5056897296778224e-05,
+      "loss": 0.0004,
+      "step": 2109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5054496566956356e-05,
+      "loss": 0.0002,
+      "step": 2110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505209583713449e-05,
+      "loss": 0.0002,
+      "step": 2111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504969510731262e-05,
+      "loss": 0.0002,
+      "step": 2112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5047294377490765e-05,
+      "loss": 0.0002,
+      "step": 2113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50448936476689e-05,
+      "loss": 0.7943,
+      "step": 2114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504249291784703e-05,
+      "loss": 0.0002,
+      "step": 2115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.504009218802516e-05,
+      "loss": 0.6177,
+      "step": 2116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.50376914582033e-05,
+      "loss": 0.0003,
+      "step": 2117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503529072838143e-05,
+      "loss": 0.0002,
+      "step": 2118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.503288999855956e-05,
+      "loss": 0.0004,
+      "step": 2119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5030489268737694e-05,
+      "loss": 0.0002,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502808853891583e-05,
+      "loss": 0.0003,
+      "step": 2121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502568780909397e-05,
+      "loss": 0.0013,
+      "step": 2122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.50232870792721e-05,
+      "loss": 0.0003,
+      "step": 2123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5020886349450235e-05,
+      "loss": 0.0002,
+      "step": 2124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.501848561962837e-05,
+      "loss": 0.6068,
+      "step": 2125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.5016084889806506e-05,
+      "loss": 0.0004,
+      "step": 2126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.501368415998464e-05,
+      "loss": 0.0002,
+      "step": 2127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.501128343016277e-05,
+      "loss": 0.0006,
+      "step": 2128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.50088827003409e-05,
+      "loss": 0.0004,
+      "step": 2129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.500648197051904e-05,
+      "loss": 0.0004,
+      "step": 2130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.500408124069718e-05,
+      "loss": 0.0007,
+      "step": 2131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.500168051087531e-05,
+      "loss": 0.0008,
+      "step": 2132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.499927978105344e-05,
+      "loss": 0.0019,
+      "step": 2133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.499687905123158e-05,
+      "loss": 0.0006,
+      "step": 2134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.499447832140971e-05,
+      "loss": 0.0003,
+      "step": 2135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4992077591587844e-05,
+      "loss": 0.0005,
+      "step": 2136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4989676861765976e-05,
+      "loss": 0.0002,
+      "step": 2137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4987276131944114e-05,
+      "loss": 0.0002,
+      "step": 2138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4984875402122246e-05,
+      "loss": 0.0003,
+      "step": 2139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.498247467230038e-05,
+      "loss": 0.0001,
+      "step": 2140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4980073942478516e-05,
+      "loss": 0.0003,
+      "step": 2141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.497767321265665e-05,
+      "loss": 0.0001,
+      "step": 2142
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.497527248283479e-05,
+      "loss": 0.0002,
+      "step": 2143
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.497287175301292e-05,
+      "loss": 0.0002,
+      "step": 2144
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.497047102319105e-05,
+      "loss": 0.0003,
+      "step": 2145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.496807029336918e-05,
+      "loss": 0.0002,
+      "step": 2146
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.496566956354732e-05,
+      "loss": 0.0002,
+      "step": 2147
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.496326883372545e-05,
+      "loss": 0.0007,
+      "step": 2148
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4960868103903584e-05,
+      "loss": 0.0001,
+      "step": 2149
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.495846737408172e-05,
+      "loss": 0.0002,
+      "step": 2150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.495606664425986e-05,
+      "loss": 0.0002,
+      "step": 2151
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.495366591443799e-05,
+      "loss": 0.0001,
+      "step": 2152
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4951265184616125e-05,
+      "loss": 0.5298,
+      "step": 2153
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494886445479426e-05,
+      "loss": 0.001,
+      "step": 2154
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4946463724972396e-05,
+      "loss": 0.0001,
+      "step": 2155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494406299515053e-05,
+      "loss": 0.001,
+      "step": 2156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.494166226532866e-05,
+      "loss": 0.0004,
+      "step": 2157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493926153550679e-05,
+      "loss": 0.0001,
+      "step": 2158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493686080568493e-05,
+      "loss": 0.0002,
+      "step": 2159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.493446007586307e-05,
+      "loss": 0.0002,
+      "step": 2160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49320593460412e-05,
+      "loss": 0.0003,
+      "step": 2161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492965861621933e-05,
+      "loss": 0.0001,
+      "step": 2162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.492725788639747e-05,
+      "loss": 0.0003,
+      "step": 2163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.49248571565756e-05,
+      "loss": 0.0003,
+      "step": 2164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4922456426753734e-05,
+      "loss": 0.0006,
+      "step": 2165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4920055696931866e-05,
+      "loss": 0.0002,
+      "step": 2166
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4917654967110004e-05,
+      "loss": 0.0019,
+      "step": 2167
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.491525423728814e-05,
+      "loss": 0.0001,
+      "step": 2168
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4912853507466275e-05,
+      "loss": 0.0002,
+      "step": 2169
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4910452777644406e-05,
+      "loss": 0.6723,
+      "step": 2170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490805204782254e-05,
+      "loss": 0.0004,
+      "step": 2171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490565131800068e-05,
+      "loss": 0.0002,
+      "step": 2172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490325058817881e-05,
+      "loss": 0.0002,
+      "step": 2173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.490084985835694e-05,
+      "loss": 0.8532,
+      "step": 2174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489844912853507e-05,
+      "loss": 0.0002,
+      "step": 2175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489604839871321e-05,
+      "loss": 0.0006,
+      "step": 2176
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489364766889135e-05,
+      "loss": 0.0007,
+      "step": 2177
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.489124693906948e-05,
+      "loss": 0.0003,
+      "step": 2178
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488884620924761e-05,
+      "loss": 0.0005,
+      "step": 2179
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488644547942575e-05,
+      "loss": 0.0043,
+      "step": 2180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488404474960388e-05,
+      "loss": 0.0005,
+      "step": 2181
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4881644019782015e-05,
+      "loss": 0.0005,
+      "step": 2182
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487924328996015e-05,
+      "loss": 0.0006,
+      "step": 2183
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4876842560138285e-05,
+      "loss": 0.0003,
+      "step": 2184
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487444183031642e-05,
+      "loss": 0.0005,
+      "step": 2185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4872041100494556e-05,
+      "loss": 0.0008,
+      "step": 2186
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486964037067269e-05,
+      "loss": 0.0005,
+      "step": 2187
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486723964085082e-05,
+      "loss": 0.0002,
+      "step": 2188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486483891102896e-05,
+      "loss": 0.0002,
+      "step": 2189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486243818120709e-05,
+      "loss": 0.0002,
+      "step": 2190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.486003745138522e-05,
+      "loss": 0.002,
+      "step": 2191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4857636721563353e-05,
+      "loss": 0.0002,
+      "step": 2192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485523599174149e-05,
+      "loss": 0.0001,
+      "step": 2193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4852835261919624e-05,
+      "loss": 0.0002,
+      "step": 2194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.485043453209776e-05,
+      "loss": 0.0009,
+      "step": 2195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4848033802275894e-05,
+      "loss": 0.0023,
+      "step": 2196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.484563307245403e-05,
+      "loss": 0.9691,
+      "step": 2197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4843232342632165e-05,
+      "loss": 0.0021,
+      "step": 2198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4840831612810296e-05,
+      "loss": 0.0002,
+      "step": 2199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483843088298843e-05,
+      "loss": 0.0117,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483603015316657e-05,
+      "loss": 0.0018,
+      "step": 2201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.48336294233447e-05,
+      "loss": 0.5259,
+      "step": 2202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.483122869352283e-05,
+      "loss": 0.0009,
+      "step": 2203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.482882796370096e-05,
+      "loss": 0.004,
+      "step": 2204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.48264272338791e-05,
+      "loss": 0.0021,
+      "step": 2205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.482402650405724e-05,
+      "loss": 0.0058,
+      "step": 2206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.482162577423537e-05,
+      "loss": 0.0022,
+      "step": 2207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.48192250444135e-05,
+      "loss": 0.0007,
+      "step": 2208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4816824314591635e-05,
+      "loss": 0.0016,
+      "step": 2209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.481442358476977e-05,
+      "loss": 0.0014,
+      "step": 2210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4812022854947905e-05,
+      "loss": 0.002,
+      "step": 2211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.480962212512604e-05,
+      "loss": 0.0006,
+      "step": 2212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.480722139530417e-05,
+      "loss": 0.0017,
+      "step": 2213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4804820665482314e-05,
+      "loss": 0.0006,
+      "step": 2214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4802419935660446e-05,
+      "loss": 0.0002,
+      "step": 2215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.480001920583858e-05,
+      "loss": 0.0821,
+      "step": 2216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479761847601671e-05,
+      "loss": 0.0041,
+      "step": 2217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479521774619485e-05,
+      "loss": 0.004,
+      "step": 2218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479281701637298e-05,
+      "loss": 0.0021,
+      "step": 2219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.479041628655111e-05,
+      "loss": 0.0006,
+      "step": 2220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4788015556729243e-05,
+      "loss": 0.001,
+      "step": 2221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.478561482690738e-05,
+      "loss": 0.0014,
+      "step": 2222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.478321409708552e-05,
+      "loss": 0.0013,
+      "step": 2223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.478081336726365e-05,
+      "loss": 0.0006,
+      "step": 2224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4778412637441784e-05,
+      "loss": 0.0002,
+      "step": 2225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4776011907619916e-05,
+      "loss": 0.0002,
+      "step": 2226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4773611177798055e-05,
+      "loss": 0.0003,
+      "step": 2227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4771210447976186e-05,
+      "loss": 0.0004,
+      "step": 2228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476880971815432e-05,
+      "loss": 0.001,
+      "step": 2229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476640898833246e-05,
+      "loss": 0.001,
+      "step": 2230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476400825851059e-05,
+      "loss": 0.0013,
+      "step": 2231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.476160752868873e-05,
+      "loss": 0.0002,
+      "step": 2232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475920679886686e-05,
+      "loss": 0.0002,
+      "step": 2233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475680606904499e-05,
+      "loss": 0.0001,
+      "step": 2234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475440533922313e-05,
+      "loss": 0.0002,
+      "step": 2235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.475200460940126e-05,
+      "loss": 0.0002,
+      "step": 2236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474960387957939e-05,
+      "loss": 0.0002,
+      "step": 2237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4747203149757525e-05,
+      "loss": 0.0004,
+      "step": 2238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.474480241993566e-05,
+      "loss": 0.0001,
+      "step": 2239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4742401690113795e-05,
+      "loss": 0.5358,
+      "step": 2240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4740000960291934e-05,
+      "loss": 0.0002,
+      "step": 2241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4737600230470065e-05,
+      "loss": 0.3494,
+      "step": 2242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47351995006482e-05,
+      "loss": 0.0002,
+      "step": 2243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4732798770826336e-05,
+      "loss": 0.0016,
+      "step": 2244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473039804100447e-05,
+      "loss": 0.0002,
+      "step": 2245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47279973111826e-05,
+      "loss": 0.0006,
+      "step": 2246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472559658136074e-05,
+      "loss": 0.0001,
+      "step": 2247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.472319585153887e-05,
+      "loss": 0.0002,
+      "step": 2248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4720795121717e-05,
+      "loss": 0.6216,
+      "step": 2249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471839439189514e-05,
+      "loss": 0.0031,
+      "step": 2250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471599366207327e-05,
+      "loss": 0.0007,
+      "step": 2251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471359293225141e-05,
+      "loss": 0.0068,
+      "step": 2252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.471119220242954e-05,
+      "loss": 0.0028,
+      "step": 2253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4708791472607674e-05,
+      "loss": 0.0002,
+      "step": 2254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4706390742785806e-05,
+      "loss": 0.716,
+      "step": 2255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4703990012963945e-05,
+      "loss": 0.0001,
+      "step": 2256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4701589283142076e-05,
+      "loss": 0.8238,
+      "step": 2257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469918855332021e-05,
+      "loss": 0.4007,
+      "step": 2258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469678782349835e-05,
+      "loss": 0.0001,
+      "step": 2259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4694387093676485e-05,
+      "loss": 0.0011,
+      "step": 2260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469198636385462e-05,
+      "loss": 0.001,
+      "step": 2261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468958563403275e-05,
+      "loss": 0.0001,
+      "step": 2262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468718490421088e-05,
+      "loss": 0.0002,
+      "step": 2263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468478417438902e-05,
+      "loss": 0.0002,
+      "step": 2264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.468238344456715e-05,
+      "loss": 0.0005,
+      "step": 2265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.467998271474528e-05,
+      "loss": 0.0001,
+      "step": 2266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4677581984923415e-05,
+      "loss": 0.0001,
+      "step": 2267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.467518125510155e-05,
+      "loss": 0.0003,
+      "step": 2268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.467278052527969e-05,
+      "loss": 0.6444,
+      "step": 2269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4670379795457824e-05,
+      "loss": 0.0001,
+      "step": 2270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4667979065635955e-05,
+      "loss": 0.001,
+      "step": 2271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.466557833581409e-05,
+      "loss": 0.0001,
+      "step": 2272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4663177605992226e-05,
+      "loss": 0.0003,
+      "step": 2273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.466077687617036e-05,
+      "loss": 0.3003,
+      "step": 2274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.465837614634849e-05,
+      "loss": 0.0013,
+      "step": 2275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.465597541652662e-05,
+      "loss": 0.0006,
+      "step": 2276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.465357468670476e-05,
+      "loss": 0.0002,
+      "step": 2277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46511739568829e-05,
+      "loss": 0.0013,
+      "step": 2278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464877322706103e-05,
+      "loss": 0.0002,
+      "step": 2279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464637249723916e-05,
+      "loss": 0.0003,
+      "step": 2280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46439717674173e-05,
+      "loss": 0.0009,
+      "step": 2281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.464157103759543e-05,
+      "loss": 0.0041,
+      "step": 2282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4639170307773564e-05,
+      "loss": 0.0003,
+      "step": 2283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4636769577951696e-05,
+      "loss": 0.0003,
+      "step": 2284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4634368848129834e-05,
+      "loss": 0.0001,
+      "step": 2285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4631968118307966e-05,
+      "loss": 0.0002,
+      "step": 2286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4629567388486105e-05,
+      "loss": 0.0007,
+      "step": 2287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462716665866424e-05,
+      "loss": 0.0008,
+      "step": 2288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462476592884237e-05,
+      "loss": 0.0004,
+      "step": 2289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.462236519902051e-05,
+      "loss": 0.0026,
+      "step": 2290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461996446919864e-05,
+      "loss": 0.0002,
+      "step": 2291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461756373937677e-05,
+      "loss": 0.0003,
+      "step": 2292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.46151630095549e-05,
+      "loss": 0.0009,
+      "step": 2293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461276227973304e-05,
+      "loss": 0.0004,
+      "step": 2294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.461036154991117e-05,
+      "loss": 0.0002,
+      "step": 2295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460796082008931e-05,
+      "loss": 0.0004,
+      "step": 2296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460556009026744e-05,
+      "loss": 0.002,
+      "step": 2297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.460315936044558e-05,
+      "loss": 0.0001,
+      "step": 2298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4600758630623714e-05,
+      "loss": 0.0002,
+      "step": 2299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4598357900801845e-05,
+      "loss": 0.0002,
+      "step": 2300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459595717097998e-05,
+      "loss": 0.0555,
+      "step": 2301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4593556441158116e-05,
+      "loss": 0.0019,
+      "step": 2302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.459115571133625e-05,
+      "loss": 0.0001,
+      "step": 2303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458875498151438e-05,
+      "loss": 0.0307,
+      "step": 2304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458635425169252e-05,
+      "loss": 0.0001,
+      "step": 2305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458395352187065e-05,
+      "loss": 0.0014,
+      "step": 2306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.458155279204879e-05,
+      "loss": 0.0001,
+      "step": 2307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457915206222692e-05,
+      "loss": 0.0001,
+      "step": 2308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457675133240505e-05,
+      "loss": 0.8632,
+      "step": 2309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4574350602583184e-05,
+      "loss": 0.0001,
+      "step": 2310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457194987276132e-05,
+      "loss": 0.0001,
+      "step": 2311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4569549142939454e-05,
+      "loss": 0.0001,
+      "step": 2312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4567148413117586e-05,
+      "loss": 0.0003,
+      "step": 2313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4564747683295724e-05,
+      "loss": 0.0001,
+      "step": 2314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.456234695347386e-05,
+      "loss": 0.4709,
+      "step": 2315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4559946223651995e-05,
+      "loss": 0.0033,
+      "step": 2316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455754549383013e-05,
+      "loss": 0.0001,
+      "step": 2317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455514476400826e-05,
+      "loss": 0.519,
+      "step": 2318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45527440341864e-05,
+      "loss": 0.0002,
+      "step": 2319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.455034330436453e-05,
+      "loss": 0.0001,
+      "step": 2320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454794257454266e-05,
+      "loss": 0.0002,
+      "step": 2321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454554184472079e-05,
+      "loss": 0.0001,
+      "step": 2322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454314111489893e-05,
+      "loss": 0.0005,
+      "step": 2323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454074038507707e-05,
+      "loss": 0.0013,
+      "step": 2324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45383396552552e-05,
+      "loss": 0.0022,
+      "step": 2325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.453593892543333e-05,
+      "loss": 0.0002,
+      "step": 2326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4533538195611465e-05,
+      "loss": 0.0009,
+      "step": 2327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4531137465789604e-05,
+      "loss": 0.0001,
+      "step": 2328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4528736735967735e-05,
+      "loss": 0.0006,
+      "step": 2329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.452633600614587e-05,
+      "loss": 0.9253,
+      "step": 2330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4523935276324006e-05,
+      "loss": 0.0017,
+      "step": 2331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.452153454650214e-05,
+      "loss": 0.001,
+      "step": 2332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4519133816680276e-05,
+      "loss": 0.0001,
+      "step": 2333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.451673308685841e-05,
+      "loss": 0.0008,
+      "step": 2334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.451433235703654e-05,
+      "loss": 0.0013,
+      "step": 2335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.451193162721468e-05,
+      "loss": 0.0035,
+      "step": 2336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.450953089739281e-05,
+      "loss": 0.0033,
+      "step": 2337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.450713016757094e-05,
+      "loss": 0.0031,
+      "step": 2338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4504729437749074e-05,
+      "loss": 0.0002,
+      "step": 2339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.450232870792721e-05,
+      "loss": 0.0012,
+      "step": 2340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4499927978105344e-05,
+      "loss": 0.0008,
+      "step": 2341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.449752724828348e-05,
+      "loss": 0.0001,
+      "step": 2342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4495126518461614e-05,
+      "loss": 0.0001,
+      "step": 2343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.449272578863975e-05,
+      "loss": 0.0005,
+      "step": 2344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4490325058817885e-05,
+      "loss": 0.0042,
+      "step": 2345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4487924328996017e-05,
+      "loss": 0.0014,
+      "step": 2346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.448552359917415e-05,
+      "loss": 0.0001,
+      "step": 2347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.448312286935229e-05,
+      "loss": 0.001,
+      "step": 2348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.448072213953042e-05,
+      "loss": 0.0001,
+      "step": 2349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447832140970855e-05,
+      "loss": 0.0011,
+      "step": 2350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447592067988669e-05,
+      "loss": 0.5061,
+      "step": 2351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447351995006482e-05,
+      "loss": 0.0001,
+      "step": 2352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.447111922024296e-05,
+      "loss": 0.8662,
+      "step": 2353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.446871849042109e-05,
+      "loss": 0.0001,
+      "step": 2354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.446631776059922e-05,
+      "loss": 0.0241,
+      "step": 2355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4463917030777355e-05,
+      "loss": 0.0002,
+      "step": 2356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4461516300955494e-05,
+      "loss": 0.0001,
+      "step": 2357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4459115571133625e-05,
+      "loss": 0.0004,
+      "step": 2358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.445671484131176e-05,
+      "loss": 0.0007,
+      "step": 2359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4454314111489896e-05,
+      "loss": 0.7418,
+      "step": 2360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4451913381668034e-05,
+      "loss": 0.0001,
+      "step": 2361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4449512651846166e-05,
+      "loss": 0.0231,
+      "step": 2362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44471119220243e-05,
+      "loss": 0.0005,
+      "step": 2363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444471119220243e-05,
+      "loss": 0.0008,
+      "step": 2364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.444231046238057e-05,
+      "loss": 0.0003,
+      "step": 2365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44399097325587e-05,
+      "loss": 0.0015,
+      "step": 2366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443750900273683e-05,
+      "loss": 0.0037,
+      "step": 2367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4435108272914964e-05,
+      "loss": 0.0002,
+      "step": 2368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.44327075430931e-05,
+      "loss": 0.0003,
+      "step": 2369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.443030681327124e-05,
+      "loss": 0.0035,
+      "step": 2370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.442790608344937e-05,
+      "loss": 0.0008,
+      "step": 2371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4425505353627504e-05,
+      "loss": 0.0001,
+      "step": 2372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4423104623805636e-05,
+      "loss": 0.0027,
+      "step": 2373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4420703893983775e-05,
+      "loss": 0.516,
+      "step": 2374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4418303164161907e-05,
+      "loss": 0.0002,
+      "step": 2375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441590243434004e-05,
+      "loss": 0.001,
+      "step": 2376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.441350170451817e-05,
+      "loss": 0.0003,
+      "step": 2377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4411100974696316e-05,
+      "loss": 0.6486,
+      "step": 2378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440870024487445e-05,
+      "loss": 0.0001,
+      "step": 2379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440629951505258e-05,
+      "loss": 0.003,
+      "step": 2380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440389878523071e-05,
+      "loss": 0.0002,
+      "step": 2381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440149805540885e-05,
+      "loss": 0.0005,
+      "step": 2382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439909732558698e-05,
+      "loss": 0.0021,
+      "step": 2383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.439669659576511e-05,
+      "loss": 0.0017,
+      "step": 2384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4394295865943245e-05,
+      "loss": 0.0003,
+      "step": 2385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4391895136121383e-05,
+      "loss": 0.0024,
+      "step": 2386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438949440629952e-05,
+      "loss": 0.0021,
+      "step": 2387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4387093676477654e-05,
+      "loss": 0.0024,
+      "step": 2388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4384692946655786e-05,
+      "loss": 0.0005,
+      "step": 2389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438229221683392e-05,
+      "loss": 0.0017,
+      "step": 2390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4379891487012056e-05,
+      "loss": 0.0015,
+      "step": 2391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437749075719019e-05,
+      "loss": 0.0002,
+      "step": 2392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437509002736832e-05,
+      "loss": 0.0001,
+      "step": 2393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437268929754645e-05,
+      "loss": 0.003,
+      "step": 2394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.437028856772459e-05,
+      "loss": 0.0003,
+      "step": 2395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436788783790272e-05,
+      "loss": 0.0002,
+      "step": 2396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436548710808086e-05,
+      "loss": 0.0002,
+      "step": 2397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436308637825899e-05,
+      "loss": 0.0001,
+      "step": 2398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436068564843713e-05,
+      "loss": 0.0001,
+      "step": 2399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435828491861526e-05,
+      "loss": 0.001,
+      "step": 2400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4355884188793394e-05,
+      "loss": 0.0034,
+      "step": 2401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4353483458971526e-05,
+      "loss": 0.0001,
+      "step": 2402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4351082729149665e-05,
+      "loss": 0.0001,
+      "step": 2403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4348681999327797e-05,
+      "loss": 0.0001,
+      "step": 2404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.434628126950593e-05,
+      "loss": 0.0001,
+      "step": 2405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.434388053968407e-05,
+      "loss": 0.0001,
+      "step": 2406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.43414798098622e-05,
+      "loss": 0.0002,
+      "step": 2407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.433907908004034e-05,
+      "loss": 0.001,
+      "step": 2408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.433667835021847e-05,
+      "loss": 0.0001,
+      "step": 2409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.43342776203966e-05,
+      "loss": 0.0003,
+      "step": 2410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.433187689057474e-05,
+      "loss": 0.0001,
+      "step": 2411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.432947616075287e-05,
+      "loss": 0.0001,
+      "step": 2412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4327075430931e-05,
+      "loss": 0.0003,
+      "step": 2413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4324674701109135e-05,
+      "loss": 0.0001,
+      "step": 2414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4322273971287273e-05,
+      "loss": 0.0001,
+      "step": 2415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.431987324146541e-05,
+      "loss": 0.0016,
+      "step": 2416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4317472511643544e-05,
+      "loss": 0.0008,
+      "step": 2417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4315071781821676e-05,
+      "loss": 0.0001,
+      "step": 2418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.431267105199981e-05,
+      "loss": 0.0001,
+      "step": 2419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4310270322177946e-05,
+      "loss": 0.001,
+      "step": 2420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430786959235608e-05,
+      "loss": 0.8052,
+      "step": 2421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430546886253421e-05,
+      "loss": 0.0055,
+      "step": 2422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430306813271234e-05,
+      "loss": 0.0001,
+      "step": 2423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.430066740289048e-05,
+      "loss": 0.5723,
+      "step": 2424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429826667306862e-05,
+      "loss": 0.0014,
+      "step": 2425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429586594324675e-05,
+      "loss": 0.0001,
+      "step": 2426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429346521342488e-05,
+      "loss": 0.0001,
+      "step": 2427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.429106448360302e-05,
+      "loss": 0.0001,
+      "step": 2428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.428866375378115e-05,
+      "loss": 0.0027,
+      "step": 2429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4286263023959284e-05,
+      "loss": 0.0001,
+      "step": 2430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4283862294137416e-05,
+      "loss": 0.0001,
+      "step": 2431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4281461564315555e-05,
+      "loss": 0.0001,
+      "step": 2432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427906083449369e-05,
+      "loss": 0.0001,
+      "step": 2433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4276660104671825e-05,
+      "loss": 0.0001,
+      "step": 2434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427425937484996e-05,
+      "loss": 0.0063,
+      "step": 2435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.427185864502809e-05,
+      "loss": 0.0001,
+      "step": 2436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426945791520623e-05,
+      "loss": 0.0022,
+      "step": 2437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426705718538436e-05,
+      "loss": 0.0001,
+      "step": 2438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426465645556249e-05,
+      "loss": 0.0002,
+      "step": 2439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.426225572574062e-05,
+      "loss": 0.0024,
+      "step": 2440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425985499591876e-05,
+      "loss": 0.0023,
+      "step": 2441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42574542660969e-05,
+      "loss": 0.0004,
+      "step": 2442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.425505353627503e-05,
+      "loss": 0.0001,
+      "step": 2443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4252652806453163e-05,
+      "loss": 0.0025,
+      "step": 2444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42502520766313e-05,
+      "loss": 0.0001,
+      "step": 2445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4247851346809434e-05,
+      "loss": 0.0029,
+      "step": 2446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4245450616987566e-05,
+      "loss": 0.0001,
+      "step": 2447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42430498871657e-05,
+      "loss": 0.0001,
+      "step": 2448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4240649157343836e-05,
+      "loss": 0.0015,
+      "step": 2449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423824842752197e-05,
+      "loss": 0.0001,
+      "step": 2450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.42358476977001e-05,
+      "loss": 0.0006,
+      "step": 2451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423344696787824e-05,
+      "loss": 0.0036,
+      "step": 2452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423104623805637e-05,
+      "loss": 0.9484,
+      "step": 2453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422864550823451e-05,
+      "loss": 0.0001,
+      "step": 2454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422624477841264e-05,
+      "loss": 0.0001,
+      "step": 2455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.422384404859077e-05,
+      "loss": 0.0014,
+      "step": 2456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4221443318768904e-05,
+      "loss": 0.0048,
+      "step": 2457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421904258894704e-05,
+      "loss": 0.5124,
+      "step": 2458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4216641859125174e-05,
+      "loss": 0.0005,
+      "step": 2459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4214241129303306e-05,
+      "loss": 0.51,
+      "step": 2460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4211840399481445e-05,
+      "loss": 0.0017,
+      "step": 2461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420943966965958e-05,
+      "loss": 0.0002,
+      "step": 2462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4207038939837715e-05,
+      "loss": 0.0011,
+      "step": 2463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420463821001585e-05,
+      "loss": 0.0012,
+      "step": 2464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.420223748019398e-05,
+      "loss": 0.0001,
+      "step": 2465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419983675037212e-05,
+      "loss": 0.0001,
+      "step": 2466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419743602055025e-05,
+      "loss": 0.0026,
+      "step": 2467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419503529072838e-05,
+      "loss": 0.0003,
+      "step": 2468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419263456090651e-05,
+      "loss": 0.0023,
+      "step": 2469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419023383108465e-05,
+      "loss": 0.0006,
+      "step": 2470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.418783310126279e-05,
+      "loss": 0.0025,
+      "step": 2471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.418543237144092e-05,
+      "loss": 0.0001,
+      "step": 2472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4183031641619053e-05,
+      "loss": 0.0001,
+      "step": 2473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4180630911797185e-05,
+      "loss": 0.0003,
+      "step": 2474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4178230181975324e-05,
+      "loss": 0.0007,
+      "step": 2475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4175829452153456e-05,
+      "loss": 0.0006,
+      "step": 2476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.417342872233159e-05,
+      "loss": 0.2844,
+      "step": 2477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.417102799250972e-05,
+      "loss": 0.001,
+      "step": 2478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4168627262687865e-05,
+      "loss": 0.0001,
+      "step": 2479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4166226532865996e-05,
+      "loss": 0.0001,
+      "step": 2480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.416382580304413e-05,
+      "loss": 0.0084,
+      "step": 2481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.416142507322226e-05,
+      "loss": 0.0001,
+      "step": 2482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.41590243434004e-05,
+      "loss": 0.0002,
+      "step": 2483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.415662361357853e-05,
+      "loss": 0.0024,
+      "step": 2484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.415422288375666e-05,
+      "loss": 0.0013,
+      "step": 2485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4151822153934794e-05,
+      "loss": 0.0001,
+      "step": 2486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.414942142411293e-05,
+      "loss": 0.0001,
+      "step": 2487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.414702069429107e-05,
+      "loss": 0.0001,
+      "step": 2488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.41446199644692e-05,
+      "loss": 0.0001,
+      "step": 2489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4142219234647335e-05,
+      "loss": 0.0002,
+      "step": 2490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4139818504825466e-05,
+      "loss": 0.0001,
+      "step": 2491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4137417775003605e-05,
+      "loss": 0.0001,
+      "step": 2492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.413501704518174e-05,
+      "loss": 0.0006,
+      "step": 2493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.413261631535987e-05,
+      "loss": 0.0001,
+      "step": 2494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.413021558553801e-05,
+      "loss": 0.0,
+      "step": 2495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412781485571614e-05,
+      "loss": 0.0001,
+      "step": 2496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412541412589428e-05,
+      "loss": 0.0005,
+      "step": 2497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412301339607241e-05,
+      "loss": 0.0008,
+      "step": 2498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.412061266625054e-05,
+      "loss": 0.0001,
+      "step": 2499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411821193642868e-05,
+      "loss": 0.0007,
+      "step": 2500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411581120660681e-05,
+      "loss": 0.0191,
+      "step": 2501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.411341047678494e-05,
+      "loss": 0.0,
+      "step": 2502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4111009746963075e-05,
+      "loss": 0.0005,
+      "step": 2503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4108609017141214e-05,
+      "loss": 0.0001,
+      "step": 2504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4106208287319346e-05,
+      "loss": 0.0001,
+      "step": 2505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4103807557497484e-05,
+      "loss": 0.0001,
+      "step": 2506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4101406827675616e-05,
+      "loss": 0.0001,
+      "step": 2507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4099006097853755e-05,
+      "loss": 0.001,
+      "step": 2508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4096605368031886e-05,
+      "loss": 0.0019,
+      "step": 2509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409420463821002e-05,
+      "loss": 0.0,
+      "step": 2510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.409180390838815e-05,
+      "loss": 0.0001,
+      "step": 2511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408940317856629e-05,
+      "loss": 0.0005,
+      "step": 2512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408700244874442e-05,
+      "loss": 0.0001,
+      "step": 2513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.408460171892255e-05,
+      "loss": 0.0001,
+      "step": 2514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4082200989100684e-05,
+      "loss": 0.0,
+      "step": 2515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407980025927882e-05,
+      "loss": 0.0004,
+      "step": 2516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407739952945696e-05,
+      "loss": 1.2402,
+      "step": 2517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407499879963509e-05,
+      "loss": 0.0001,
+      "step": 2518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4072598069813225e-05,
+      "loss": 0.0004,
+      "step": 2519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4070197339991356e-05,
+      "loss": 0.0006,
+      "step": 2520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4067796610169495e-05,
+      "loss": 0.0019,
+      "step": 2521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406539588034763e-05,
+      "loss": 0.0013,
+      "step": 2522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406299515052576e-05,
+      "loss": 0.0013,
+      "step": 2523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.406059442070389e-05,
+      "loss": 0.0009,
+      "step": 2524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4058193690882036e-05,
+      "loss": 0.7367,
+      "step": 2525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405579296106017e-05,
+      "loss": 0.7818,
+      "step": 2526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40533922312383e-05,
+      "loss": 0.0005,
+      "step": 2527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.405099150141643e-05,
+      "loss": 0.0007,
+      "step": 2528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404859077159457e-05,
+      "loss": 0.001,
+      "step": 2529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40461900417727e-05,
+      "loss": 0.0017,
+      "step": 2530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404378931195083e-05,
+      "loss": 0.007,
+      "step": 2531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4041388582128965e-05,
+      "loss": 0.0014,
+      "step": 2532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4038987852307104e-05,
+      "loss": 0.0027,
+      "step": 2533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403658712248524e-05,
+      "loss": 0.0036,
+      "step": 2534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4034186392663374e-05,
+      "loss": 0.0036,
+      "step": 2535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4031785662841506e-05,
+      "loss": 0.0016,
+      "step": 2536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402938493301964e-05,
+      "loss": 0.0009,
+      "step": 2537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4026984203197776e-05,
+      "loss": 0.0009,
+      "step": 2538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402458347337591e-05,
+      "loss": 0.0009,
+      "step": 2539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.402218274355404e-05,
+      "loss": 0.0016,
+      "step": 2540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401978201373217e-05,
+      "loss": 0.0018,
+      "step": 2541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401738128391031e-05,
+      "loss": 0.0021,
+      "step": 2542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401498055408845e-05,
+      "loss": 0.0009,
+      "step": 2543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401257982426658e-05,
+      "loss": 0.0004,
+      "step": 2544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.401017909444471e-05,
+      "loss": 0.0008,
+      "step": 2545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.400777836462285e-05,
+      "loss": 0.0005,
+      "step": 2546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.400537763480098e-05,
+      "loss": 0.0077,
+      "step": 2547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4002976904979115e-05,
+      "loss": 0.0006,
+      "step": 2548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4000576175157246e-05,
+      "loss": 0.0006,
+      "step": 2549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3998175445335385e-05,
+      "loss": 0.0005,
+      "step": 2550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.399577471551352e-05,
+      "loss": 0.0006,
+      "step": 2551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3993373985691655e-05,
+      "loss": 0.0005,
+      "step": 2552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.399097325586979e-05,
+      "loss": 0.0005,
+      "step": 2553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398857252604792e-05,
+      "loss": 0.0041,
+      "step": 2554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398617179622606e-05,
+      "loss": 0.0003,
+      "step": 2555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398377106640419e-05,
+      "loss": 0.7006,
+      "step": 2556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.398137033658232e-05,
+      "loss": 0.0003,
+      "step": 2557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397896960676045e-05,
+      "loss": 0.7151,
+      "step": 2558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397656887693859e-05,
+      "loss": 0.0003,
+      "step": 2559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397416814711672e-05,
+      "loss": 0.0004,
+      "step": 2560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.397176741729486e-05,
+      "loss": 0.002,
+      "step": 2561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3969366687472994e-05,
+      "loss": 0.0015,
+      "step": 2562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.396696595765113e-05,
+      "loss": 0.0002,
+      "step": 2563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3964565227829264e-05,
+      "loss": 0.0003,
+      "step": 2564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3962164498007396e-05,
+      "loss": 0.9411,
+      "step": 2565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.395976376818553e-05,
+      "loss": 0.0005,
+      "step": 2566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3957363038363666e-05,
+      "loss": 0.001,
+      "step": 2567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39549623085418e-05,
+      "loss": 0.0067,
+      "step": 2568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.395256157871993e-05,
+      "loss": 0.002,
+      "step": 2569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.395016084889807e-05,
+      "loss": 0.0009,
+      "step": 2570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39477601190762e-05,
+      "loss": 0.0017,
+      "step": 2571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394535938925434e-05,
+      "loss": 0.461,
+      "step": 2572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.394295865943247e-05,
+      "loss": 0.0015,
+      "step": 2573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.39405579296106e-05,
+      "loss": 0.0013,
+      "step": 2574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3938157199788734e-05,
+      "loss": 0.004,
+      "step": 2575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.393575646996687e-05,
+      "loss": 0.0036,
+      "step": 2576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3933355740145005e-05,
+      "loss": 0.0032,
+      "step": 2577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3930955010323136e-05,
+      "loss": 0.0027,
+      "step": 2578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3928554280501275e-05,
+      "loss": 0.0031,
+      "step": 2579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3926153550679414e-05,
+      "loss": 0.5834,
+      "step": 2580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3923752820857545e-05,
+      "loss": 0.0023,
+      "step": 2581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.392135209103568e-05,
+      "loss": 0.0138,
+      "step": 2582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391895136121381e-05,
+      "loss": 0.0017,
+      "step": 2583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391655063139195e-05,
+      "loss": 0.0028,
+      "step": 2584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391414990157008e-05,
+      "loss": 0.0014,
+      "step": 2585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.391174917174821e-05,
+      "loss": 0.0008,
+      "step": 2586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390934844192634e-05,
+      "loss": 0.0012,
+      "step": 2587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390694771210448e-05,
+      "loss": 0.5162,
+      "step": 2588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390454698228262e-05,
+      "loss": 0.0015,
+      "step": 2589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.390214625246075e-05,
+      "loss": 0.0026,
+      "step": 2590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3899745522638884e-05,
+      "loss": 0.0011,
+      "step": 2591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389734479281702e-05,
+      "loss": 0.0007,
+      "step": 2592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3894944062995154e-05,
+      "loss": 0.0022,
+      "step": 2593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3892543333173286e-05,
+      "loss": 0.0045,
+      "step": 2594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389014260335142e-05,
+      "loss": 0.0024,
+      "step": 2595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3887741873529556e-05,
+      "loss": 0.5886,
+      "step": 2596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388534114370769e-05,
+      "loss": 0.0037,
+      "step": 2597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3882940413885827e-05,
+      "loss": 0.0021,
+      "step": 2598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388053968406396e-05,
+      "loss": 0.0025,
+      "step": 2599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387813895424209e-05,
+      "loss": 0.0022,
+      "step": 2600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387573822442023e-05,
+      "loss": 0.0049,
+      "step": 2601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387333749459836e-05,
+      "loss": 0.0022,
+      "step": 2602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387093676477649e-05,
+      "loss": 0.0032,
+      "step": 2603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3868536034954624e-05,
+      "loss": 0.0057,
+      "step": 2604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386613530513276e-05,
+      "loss": 0.0055,
+      "step": 2605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3863734575310895e-05,
+      "loss": 0.0011,
+      "step": 2606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.386133384548903e-05,
+      "loss": 0.0023,
+      "step": 2607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3858933115667165e-05,
+      "loss": 0.0034,
+      "step": 2608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3856532385845304e-05,
+      "loss": 0.4628,
+      "step": 2609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3854131656023435e-05,
+      "loss": 0.0027,
+      "step": 2610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.385173092620157e-05,
+      "loss": 0.0006,
+      "step": 2611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.38493301963797e-05,
+      "loss": 0.0027,
+      "step": 2612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.384692946655784e-05,
+      "loss": 0.0034,
+      "step": 2613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.384452873673597e-05,
+      "loss": 0.0005,
+      "step": 2614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.38421280069141e-05,
+      "loss": 0.0034,
+      "step": 2615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.383972727709224e-05,
+      "loss": 0.0045,
+      "step": 2616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.383732654727037e-05,
+      "loss": 0.0005,
+      "step": 2617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.383492581744851e-05,
+      "loss": 0.0017,
+      "step": 2618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.383252508762664e-05,
+      "loss": 0.001,
+      "step": 2619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3830124357804774e-05,
+      "loss": 0.0011,
+      "step": 2620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3827723627982905e-05,
+      "loss": 0.0007,
+      "step": 2621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3825322898161044e-05,
+      "loss": 0.0004,
+      "step": 2622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3822922168339176e-05,
+      "loss": 0.0002,
+      "step": 2623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.382052143851731e-05,
+      "loss": 0.0004,
+      "step": 2624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3818120708695446e-05,
+      "loss": 0.0012,
+      "step": 2625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3815719978873585e-05,
+      "loss": 0.0059,
+      "step": 2626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3813319249051717e-05,
+      "loss": 0.325,
+      "step": 2627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.381091851922985e-05,
+      "loss": 0.0009,
+      "step": 2628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380851778940798e-05,
+      "loss": 0.0006,
+      "step": 2629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380611705958612e-05,
+      "loss": 0.0007,
+      "step": 2630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380371632976425e-05,
+      "loss": 0.0038,
+      "step": 2631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.380131559994238e-05,
+      "loss": 0.3104,
+      "step": 2632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3798914870120514e-05,
+      "loss": 0.0018,
+      "step": 2633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.379651414029865e-05,
+      "loss": 0.0005,
+      "step": 2634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.379411341047679e-05,
+      "loss": 0.0015,
+      "step": 2635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.379171268065492e-05,
+      "loss": 0.0003,
+      "step": 2636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3789311950833055e-05,
+      "loss": 0.0003,
+      "step": 2637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.378691122101119e-05,
+      "loss": 0.0006,
+      "step": 2638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3784510491189325e-05,
+      "loss": 0.0006,
+      "step": 2639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.378210976136746e-05,
+      "loss": 0.0005,
+      "step": 2640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377970903154559e-05,
+      "loss": 0.0026,
+      "step": 2641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377730830172372e-05,
+      "loss": 0.0002,
+      "step": 2642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377490757190186e-05,
+      "loss": 0.0002,
+      "step": 2643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377250684208e-05,
+      "loss": 0.0003,
+      "step": 2644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.377010611225813e-05,
+      "loss": 0.0011,
+      "step": 2645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376770538243626e-05,
+      "loss": 0.0002,
+      "step": 2646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.37653046526144e-05,
+      "loss": 0.0002,
+      "step": 2647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.376290392279253e-05,
+      "loss": 0.0002,
+      "step": 2648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3760503192970664e-05,
+      "loss": 0.0011,
+      "step": 2649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3758102463148795e-05,
+      "loss": 0.0005,
+      "step": 2650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3755701733326934e-05,
+      "loss": 0.0009,
+      "step": 2651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3753301003505066e-05,
+      "loss": 1.0307,
+      "step": 2652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3750900273683204e-05,
+      "loss": 0.3299,
+      "step": 2653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3748499543861336e-05,
+      "loss": 0.0012,
+      "step": 2654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374609881403947e-05,
+      "loss": 0.0002,
+      "step": 2655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3743698084217607e-05,
+      "loss": 0.0002,
+      "step": 2656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374129735439574e-05,
+      "loss": 0.0003,
+      "step": 2657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373889662457387e-05,
+      "loss": 0.0003,
+      "step": 2658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373649589475201e-05,
+      "loss": 0.0006,
+      "step": 2659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373409516493014e-05,
+      "loss": 0.0007,
+      "step": 2660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.373169443510827e-05,
+      "loss": 0.0005,
+      "step": 2661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372929370528641e-05,
+      "loss": 0.0006,
+      "step": 2662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372689297546454e-05,
+      "loss": 0.0026,
+      "step": 2663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372449224564268e-05,
+      "loss": 0.0004,
+      "step": 2664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.372209151582081e-05,
+      "loss": 0.0003,
+      "step": 2665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3719690785998945e-05,
+      "loss": 0.0006,
+      "step": 2666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371729005617708e-05,
+      "loss": 0.0005,
+      "step": 2667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3714889326355215e-05,
+      "loss": 0.001,
+      "step": 2668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371248859653335e-05,
+      "loss": 0.863,
+      "step": 2669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.371008786671148e-05,
+      "loss": 0.0007,
+      "step": 2670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370768713688962e-05,
+      "loss": 0.0006,
+      "step": 2671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370528640706775e-05,
+      "loss": 0.0006,
+      "step": 2672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370288567724589e-05,
+      "loss": 0.0038,
+      "step": 2673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370048494742402e-05,
+      "loss": 0.0005,
+      "step": 2674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369808421760215e-05,
+      "loss": 0.0007,
+      "step": 2675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369568348778029e-05,
+      "loss": 0.0067,
+      "step": 2676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.369328275795842e-05,
+      "loss": 0.0006,
+      "step": 2677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3690882028136554e-05,
+      "loss": 0.0006,
+      "step": 2678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3688481298314685e-05,
+      "loss": 0.8125,
+      "step": 2679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3686080568492824e-05,
+      "loss": 0.0007,
+      "step": 2680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.368367983867096e-05,
+      "loss": 0.0005,
+      "step": 2681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3681279108849094e-05,
+      "loss": 0.0009,
+      "step": 2682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3678878379027226e-05,
+      "loss": 0.0013,
+      "step": 2683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.367647764920536e-05,
+      "loss": 0.002,
+      "step": 2684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3674076919383497e-05,
+      "loss": 0.0013,
+      "step": 2685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.367167618956163e-05,
+      "loss": 0.001,
+      "step": 2686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.366927545973976e-05,
+      "loss": 0.0005,
+      "step": 2687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.366687472991789e-05,
+      "loss": 0.0006,
+      "step": 2688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.366447400009604e-05,
+      "loss": 0.0007,
+      "step": 2689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.366207327027417e-05,
+      "loss": 0.0012,
+      "step": 2690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36596725404523e-05,
+      "loss": 0.0018,
+      "step": 2691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365727181063043e-05,
+      "loss": 0.714,
+      "step": 2692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365487108080857e-05,
+      "loss": 0.0017,
+      "step": 2693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.36524703509867e-05,
+      "loss": 0.0012,
+      "step": 2694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3650069621164835e-05,
+      "loss": 0.0021,
+      "step": 2695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.364766889134297e-05,
+      "loss": 0.0017,
+      "step": 2696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3645268161521105e-05,
+      "loss": 0.0012,
+      "step": 2697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3642867431699244e-05,
+      "loss": 0.0008,
+      "step": 2698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3640466701877376e-05,
+      "loss": 0.0014,
+      "step": 2699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363806597205551e-05,
+      "loss": 0.0008,
+      "step": 2700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363566524223364e-05,
+      "loss": 0.001,
+      "step": 2701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363326451241178e-05,
+      "loss": 0.0013,
+      "step": 2702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363086378258991e-05,
+      "loss": 0.0009,
+      "step": 2703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.362846305276804e-05,
+      "loss": 0.0008,
+      "step": 2704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.362606232294617e-05,
+      "loss": 0.0007,
+      "step": 2705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.362366159312431e-05,
+      "loss": 0.0006,
+      "step": 2706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3621260863302444e-05,
+      "loss": 0.0004,
+      "step": 2707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361886013348058e-05,
+      "loss": 0.0011,
+      "step": 2708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3616459403658714e-05,
+      "loss": 0.0004,
+      "step": 2709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.361405867383685e-05,
+      "loss": 0.0014,
+      "step": 2710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3611657944014984e-05,
+      "loss": 0.0012,
+      "step": 2711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3609257214193116e-05,
+      "loss": 0.0007,
+      "step": 2712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360685648437125e-05,
+      "loss": 0.0006,
+      "step": 2713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3604455754549386e-05,
+      "loss": 0.0004,
+      "step": 2714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360205502472752e-05,
+      "loss": 0.0007,
+      "step": 2715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359965429490565e-05,
+      "loss": 0.0009,
+      "step": 2716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359725356508379e-05,
+      "loss": 0.0013,
+      "step": 2717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359485283526192e-05,
+      "loss": 0.0006,
+      "step": 2718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359245210544006e-05,
+      "loss": 0.0024,
+      "step": 2719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359005137561819e-05,
+      "loss": 0.0007,
+      "step": 2720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358765064579632e-05,
+      "loss": 0.0008,
+      "step": 2721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3585249915974454e-05,
+      "loss": 0.0006,
+      "step": 2722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358284918615259e-05,
+      "loss": 0.001,
+      "step": 2723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3580448456330725e-05,
+      "loss": 0.0009,
+      "step": 2724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.357804772650886e-05,
+      "loss": 0.0004,
+      "step": 2725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3575646996686995e-05,
+      "loss": 0.0012,
+      "step": 2726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3573246266865134e-05,
+      "loss": 0.0003,
+      "step": 2727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3570845537043266e-05,
+      "loss": 0.0007,
+      "step": 2728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35684448072214e-05,
+      "loss": 0.0005,
+      "step": 2729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356604407739953e-05,
+      "loss": 0.0003,
+      "step": 2730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356364334757767e-05,
+      "loss": 0.0005,
+      "step": 2731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35612426177558e-05,
+      "loss": 0.0008,
+      "step": 2732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355884188793393e-05,
+      "loss": 0.5298,
+      "step": 2733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355644115811206e-05,
+      "loss": 0.0007,
+      "step": 2734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35540404282902e-05,
+      "loss": 0.0005,
+      "step": 2735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355163969846834e-05,
+      "loss": 0.0005,
+      "step": 2736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.354923896864647e-05,
+      "loss": 0.0007,
+      "step": 2737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3546838238824604e-05,
+      "loss": 0.0005,
+      "step": 2738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3544437509002736e-05,
+      "loss": 0.0005,
+      "step": 2739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3542036779180874e-05,
+      "loss": 0.0005,
+      "step": 2740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3539636049359006e-05,
+      "loss": 0.0004,
+      "step": 2741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353723531953714e-05,
+      "loss": 0.0004,
+      "step": 2742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3534834589715276e-05,
+      "loss": 0.0004,
+      "step": 2743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3532433859893415e-05,
+      "loss": 0.0003,
+      "step": 2744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353003313007155e-05,
+      "loss": 0.0005,
+      "step": 2745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352763240024968e-05,
+      "loss": 0.0007,
+      "step": 2746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352523167042781e-05,
+      "loss": 0.0005,
+      "step": 2747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352283094060595e-05,
+      "loss": 0.9793,
+      "step": 2748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.352043021078408e-05,
+      "loss": 0.0004,
+      "step": 2749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351802948096221e-05,
+      "loss": 0.0004,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3515628751140344e-05,
+      "loss": 0.0016,
+      "step": 2751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351322802131848e-05,
+      "loss": 0.0013,
+      "step": 2752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.351082729149662e-05,
+      "loss": 0.0007,
+      "step": 2753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.350842656167475e-05,
+      "loss": 0.0034,
+      "step": 2754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3506025831852885e-05,
+      "loss": 0.0015,
+      "step": 2755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3503625102031024e-05,
+      "loss": 0.001,
+      "step": 2756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3501224372209156e-05,
+      "loss": 0.0015,
+      "step": 2757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349882364238729e-05,
+      "loss": 0.0004,
+      "step": 2758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349642291256542e-05,
+      "loss": 0.0011,
+      "step": 2759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349402218274356e-05,
+      "loss": 0.0007,
+      "step": 2760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.349162145292169e-05,
+      "loss": 0.001,
+      "step": 2761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348922072309983e-05,
+      "loss": 0.0012,
+      "step": 2762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348681999327796e-05,
+      "loss": 0.0003,
+      "step": 2763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348441926345609e-05,
+      "loss": 0.4771,
+      "step": 2764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.348201853363423e-05,
+      "loss": 0.0006,
+      "step": 2765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.347961780381236e-05,
+      "loss": 0.0006,
+      "step": 2766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3477217073990494e-05,
+      "loss": 0.0009,
+      "step": 2767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3474816344168626e-05,
+      "loss": 0.0005,
+      "step": 2768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3472415614346764e-05,
+      "loss": 0.0006,
+      "step": 2769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3470014884524896e-05,
+      "loss": 0.0008,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.346761415470303e-05,
+      "loss": 0.0005,
+      "step": 2771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3465213424881166e-05,
+      "loss": 0.7473,
+      "step": 2772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3462812695059305e-05,
+      "loss": 0.001,
+      "step": 2773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.346041196523744e-05,
+      "loss": 0.4878,
+      "step": 2774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.345801123541557e-05,
+      "loss": 0.0008,
+      "step": 2775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34556105055937e-05,
+      "loss": 0.001,
+      "step": 2776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.345320977577184e-05,
+      "loss": 0.0005,
+      "step": 2777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.345080904594997e-05,
+      "loss": 0.0008,
+      "step": 2778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34484083161281e-05,
+      "loss": 0.0013,
+      "step": 2779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3446007586306234e-05,
+      "loss": 0.0005,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344360685648437e-05,
+      "loss": 0.0006,
+      "step": 2781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.344120612666251e-05,
+      "loss": 0.0011,
+      "step": 2782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343880539684064e-05,
+      "loss": 0.002,
+      "step": 2783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3436404667018775e-05,
+      "loss": 0.3299,
+      "step": 2784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.343400393719691e-05,
+      "loss": 0.0015,
+      "step": 2785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3431603207375046e-05,
+      "loss": 0.0016,
+      "step": 2786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342920247755318e-05,
+      "loss": 0.0012,
+      "step": 2787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342680174773131e-05,
+      "loss": 0.0014,
+      "step": 2788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.342440101790944e-05,
+      "loss": 0.0003,
+      "step": 2789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3422000288087586e-05,
+      "loss": 0.0017,
+      "step": 2790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341959955826572e-05,
+      "loss": 0.001,
+      "step": 2791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341719882844385e-05,
+      "loss": 0.0028,
+      "step": 2792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341479809862198e-05,
+      "loss": 0.005,
+      "step": 2793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.341239736880012e-05,
+      "loss": 0.0008,
+      "step": 2794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340999663897825e-05,
+      "loss": 0.0008,
+      "step": 2795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3407595909156384e-05,
+      "loss": 0.0004,
+      "step": 2796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3405195179334516e-05,
+      "loss": 0.0088,
+      "step": 2797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3402794449512654e-05,
+      "loss": 0.0022,
+      "step": 2798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340039371969079e-05,
+      "loss": 0.0006,
+      "step": 2799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3397992989868925e-05,
+      "loss": 0.0005,
+      "step": 2800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3395592260047056e-05,
+      "loss": 0.2432,
+      "step": 2801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339319153022519e-05,
+      "loss": 0.0009,
+      "step": 2802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339079080040333e-05,
+      "loss": 0.0017,
+      "step": 2803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338839007058146e-05,
+      "loss": 0.0007,
+      "step": 2804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338598934075959e-05,
+      "loss": 0.0007,
+      "step": 2805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338358861093772e-05,
+      "loss": 0.2462,
+      "step": 2806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.338118788111586e-05,
+      "loss": 0.0005,
+      "step": 2807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3378787151294e-05,
+      "loss": 0.0005,
+      "step": 2808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337638642147213e-05,
+      "loss": 0.0004,
+      "step": 2809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337398569165026e-05,
+      "loss": 0.0006,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33715849618284e-05,
+      "loss": 0.0008,
+      "step": 2811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.336918423200653e-05,
+      "loss": 0.0004,
+      "step": 2812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3366783502184665e-05,
+      "loss": 0.0005,
+      "step": 2813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.33643827723628e-05,
+      "loss": 0.806,
+      "step": 2814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3361982042540935e-05,
+      "loss": 0.8994,
+      "step": 2815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335958131271907e-05,
+      "loss": 0.0005,
+      "step": 2816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3357180582897206e-05,
+      "loss": 0.0009,
+      "step": 2817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335477985307534e-05,
+      "loss": 0.2991,
+      "step": 2818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.335237912325347e-05,
+      "loss": 0.0006,
+      "step": 2819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334997839343161e-05,
+      "loss": 0.0007,
+      "step": 2820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334757766360974e-05,
+      "loss": 0.0006,
+      "step": 2821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334517693378787e-05,
+      "loss": 0.0005,
+      "step": 2822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3342776203966003e-05,
+      "loss": 0.0011,
+      "step": 2823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.334037547414414e-05,
+      "loss": 0.001,
+      "step": 2824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3337974744322274e-05,
+      "loss": 0.0012,
+      "step": 2825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3335574014500406e-05,
+      "loss": 0.0021,
+      "step": 2826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3333173284678544e-05,
+      "loss": 0.0011,
+      "step": 2827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.333077255485668e-05,
+      "loss": 0.0008,
+      "step": 2828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3328371825034815e-05,
+      "loss": 0.0033,
+      "step": 2829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3325971095212946e-05,
+      "loss": 0.0006,
+      "step": 2830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.332357036539108e-05,
+      "loss": 0.0079,
+      "step": 2831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.332116963556922e-05,
+      "loss": 0.001,
+      "step": 2832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.331876890574735e-05,
+      "loss": 0.0003,
+      "step": 2833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.331636817592548e-05,
+      "loss": 0.0008,
+      "step": 2834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.331396744610361e-05,
+      "loss": 0.0009,
+      "step": 2835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.331156671628175e-05,
+      "loss": 0.001,
+      "step": 2836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.330916598645989e-05,
+      "loss": 0.0006,
+      "step": 2837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.330676525663802e-05,
+      "loss": 0.0009,
+      "step": 2838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.330436452681615e-05,
+      "loss": 0.0005,
+      "step": 2839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.330196379699429e-05,
+      "loss": 0.001,
+      "step": 2840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.329956306717242e-05,
+      "loss": 0.0007,
+      "step": 2841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3297162337350555e-05,
+      "loss": 0.0009,
+      "step": 2842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.329476160752869e-05,
+      "loss": 0.0005,
+      "step": 2843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3292360877706825e-05,
+      "loss": 0.0005,
+      "step": 2844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3289960147884964e-05,
+      "loss": 0.0002,
+      "step": 2845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3287559418063096e-05,
+      "loss": 0.0004,
+      "step": 2846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328515868824123e-05,
+      "loss": 0.0004,
+      "step": 2847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.328275795841936e-05,
+      "loss": 0.0001,
+      "step": 2848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32803572285975e-05,
+      "loss": 0.0007,
+      "step": 2849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327795649877563e-05,
+      "loss": 0.0004,
+      "step": 2850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327555576895376e-05,
+      "loss": 0.0004,
+      "step": 2851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3273155039131893e-05,
+      "loss": 0.0002,
+      "step": 2852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.327075430931003e-05,
+      "loss": 0.0007,
+      "step": 2853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326835357948817e-05,
+      "loss": 0.0004,
+      "step": 2854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.32659528496663e-05,
+      "loss": 0.0004,
+      "step": 2855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3263552119844434e-05,
+      "loss": 0.0003,
+      "step": 2856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.326115139002257e-05,
+      "loss": 0.0002,
+      "step": 2857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3258750660200705e-05,
+      "loss": 0.0003,
+      "step": 2858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3256349930378836e-05,
+      "loss": 0.0004,
+      "step": 2859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325394920055697e-05,
+      "loss": 0.8098,
+      "step": 2860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.325154847073511e-05,
+      "loss": 0.0005,
+      "step": 2861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324914774091324e-05,
+      "loss": 0.0002,
+      "step": 2862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324674701109138e-05,
+      "loss": 0.0004,
+      "step": 2863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324434628126951e-05,
+      "loss": 0.0001,
+      "step": 2864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324194555144764e-05,
+      "loss": 0.0007,
+      "step": 2865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323954482162578e-05,
+      "loss": 0.0004,
+      "step": 2866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323714409180391e-05,
+      "loss": 0.0004,
+      "step": 2867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323474336198204e-05,
+      "loss": 0.0003,
+      "step": 2868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3232342632160175e-05,
+      "loss": 0.0006,
+      "step": 2869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.322994190233831e-05,
+      "loss": 0.0002,
+      "step": 2870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3227541172516445e-05,
+      "loss": 0.0001,
+      "step": 2871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3225140442694584e-05,
+      "loss": 0.0009,
+      "step": 2872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3222739712872715e-05,
+      "loss": 0.0408,
+      "step": 2873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3220338983050854e-05,
+      "loss": 0.0005,
+      "step": 2874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3217938253228986e-05,
+      "loss": 0.5606,
+      "step": 2875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321553752340712e-05,
+      "loss": 0.0005,
+      "step": 2876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321313679358525e-05,
+      "loss": 0.001,
+      "step": 2877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.321073606376339e-05,
+      "loss": 0.0005,
+      "step": 2878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320833533394152e-05,
+      "loss": 0.9181,
+      "step": 2879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320593460411965e-05,
+      "loss": 0.0004,
+      "step": 2880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320353387429779e-05,
+      "loss": 0.0004,
+      "step": 2881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320113314447592e-05,
+      "loss": 0.0007,
+      "step": 2882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319873241465406e-05,
+      "loss": 0.8418,
+      "step": 2883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319633168483219e-05,
+      "loss": 0.818,
+      "step": 2884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3193930955010324e-05,
+      "loss": 0.7701,
+      "step": 2885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3191530225188456e-05,
+      "loss": 0.002,
+      "step": 2886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3189129495366595e-05,
+      "loss": 0.0023,
+      "step": 2887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3186728765544726e-05,
+      "loss": 0.0033,
+      "step": 2888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.318432803572286e-05,
+      "loss": 0.5094,
+      "step": 2889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.318192730590099e-05,
+      "loss": 0.023,
+      "step": 2890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3179526576079135e-05,
+      "loss": 0.0243,
+      "step": 2891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.317712584625727e-05,
+      "loss": 0.0156,
+      "step": 2892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.31747251164354e-05,
+      "loss": 0.0094,
+      "step": 2893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.317232438661353e-05,
+      "loss": 0.0078,
+      "step": 2894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.316992365679167e-05,
+      "loss": 0.0059,
+      "step": 2895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.31675229269698e-05,
+      "loss": 0.5595,
+      "step": 2896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.316512219714793e-05,
+      "loss": 0.0085,
+      "step": 2897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3162721467326065e-05,
+      "loss": 0.0057,
+      "step": 2898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.31603207375042e-05,
+      "loss": 0.0048,
+      "step": 2899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.315792000768234e-05,
+      "loss": 0.0079,
+      "step": 2900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3155519277860474e-05,
+      "loss": 0.2639,
+      "step": 2901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3153118548038605e-05,
+      "loss": 0.0059,
+      "step": 2902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.315071781821674e-05,
+      "loss": 0.0036,
+      "step": 2903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3148317088394876e-05,
+      "loss": 0.005,
+      "step": 2904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.314591635857301e-05,
+      "loss": 0.0055,
+      "step": 2905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.314351562875114e-05,
+      "loss": 0.0055,
+      "step": 2906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.314111489892928e-05,
+      "loss": 0.0042,
+      "step": 2907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313871416910741e-05,
+      "loss": 0.0008,
+      "step": 2908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313631343928555e-05,
+      "loss": 0.0029,
+      "step": 2909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313391270946368e-05,
+      "loss": 0.0038,
+      "step": 2910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.313151197964181e-05,
+      "loss": 0.0026,
+      "step": 2911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.312911124981995e-05,
+      "loss": 0.0027,
+      "step": 2912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.312671051999808e-05,
+      "loss": 0.0023,
+      "step": 2913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3124309790176214e-05,
+      "loss": 0.002,
+      "step": 2914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3121909060354346e-05,
+      "loss": 0.0014,
+      "step": 2915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3119508330532484e-05,
+      "loss": 0.0028,
+      "step": 2916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3117107600710616e-05,
+      "loss": 0.0017,
+      "step": 2917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3114706870888755e-05,
+      "loss": 0.002,
+      "step": 2918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.311230614106689e-05,
+      "loss": 0.0052,
+      "step": 2919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310990541124502e-05,
+      "loss": 0.002,
+      "step": 2920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310750468142316e-05,
+      "loss": 0.0015,
+      "step": 2921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310510395160129e-05,
+      "loss": 0.427,
+      "step": 2922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310270322177942e-05,
+      "loss": 0.0018,
+      "step": 2923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.310030249195756e-05,
+      "loss": 0.001,
+      "step": 2924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309790176213569e-05,
+      "loss": 0.0013,
+      "step": 2925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309550103231382e-05,
+      "loss": 0.0009,
+      "step": 2926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309310030249196e-05,
+      "loss": 0.0017,
+      "step": 2927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.309069957267009e-05,
+      "loss": 0.0014,
+      "step": 2928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308829884284823e-05,
+      "loss": 0.6501,
+      "step": 2929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3085898113026364e-05,
+      "loss": 0.0013,
+      "step": 2930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3083497383204495e-05,
+      "loss": 0.6997,
+      "step": 2931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.308109665338263e-05,
+      "loss": 0.0013,
+      "step": 2932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3078695923560766e-05,
+      "loss": 0.6618,
+      "step": 2933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30762951937389e-05,
+      "loss": 0.001,
+      "step": 2934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307389446391703e-05,
+      "loss": 0.0012,
+      "step": 2935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.307149373409517e-05,
+      "loss": 0.0017,
+      "step": 2936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3069093004273307e-05,
+      "loss": 0.0287,
+      "step": 2937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306669227445144e-05,
+      "loss": 0.0013,
+      "step": 2938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.306429154462957e-05,
+      "loss": 0.0012,
+      "step": 2939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30618908148077e-05,
+      "loss": 0.0006,
+      "step": 2940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305949008498584e-05,
+      "loss": 0.004,
+      "step": 2941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.305708935516397e-05,
+      "loss": 0.0006,
+      "step": 2942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3054688625342104e-05,
+      "loss": 0.0006,
+      "step": 2943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3052287895520236e-05,
+      "loss": 0.0047,
+      "step": 2944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3049887165698374e-05,
+      "loss": 0.0042,
+      "step": 2945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304748643587651e-05,
+      "loss": 0.0285,
+      "step": 2946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3045085706054645e-05,
+      "loss": 0.0003,
+      "step": 2947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304268497623278e-05,
+      "loss": 0.0006,
+      "step": 2948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304028424641091e-05,
+      "loss": 0.815,
+      "step": 2949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303788351658905e-05,
+      "loss": 0.0013,
+      "step": 2950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303548278676718e-05,
+      "loss": 0.0041,
+      "step": 2951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303308205694531e-05,
+      "loss": 0.0007,
+      "step": 2952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.303068132712344e-05,
+      "loss": 0.0008,
+      "step": 2953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302828059730158e-05,
+      "loss": 0.0009,
+      "step": 2954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302587986747972e-05,
+      "loss": 0.0028,
+      "step": 2955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302347913765785e-05,
+      "loss": 0.0006,
+      "step": 2956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.302107840783598e-05,
+      "loss": 0.0009,
+      "step": 2957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.301867767801412e-05,
+      "loss": 0.002,
+      "step": 2958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3016276948192254e-05,
+      "loss": 0.0028,
+      "step": 2959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3013876218370385e-05,
+      "loss": 0.0007,
+      "step": 2960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.301147548854852e-05,
+      "loss": 0.0012,
+      "step": 2961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3009074758726656e-05,
+      "loss": 0.0005,
+      "step": 2962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300667402890479e-05,
+      "loss": 0.0005,
+      "step": 2963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3004273299082926e-05,
+      "loss": 0.0029,
+      "step": 2964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.300187256926106e-05,
+      "loss": 0.0037,
+      "step": 2965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.299947183943919e-05,
+      "loss": 0.0009,
+      "step": 2966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.299707110961733e-05,
+      "loss": 0.0032,
+      "step": 2967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.299467037979546e-05,
+      "loss": 0.0005,
+      "step": 2968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.299226964997359e-05,
+      "loss": 0.1242,
+      "step": 2969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2989868920151724e-05,
+      "loss": 0.0005,
+      "step": 2970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298746819032986e-05,
+      "loss": 0.0006,
+      "step": 2971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2985067460507994e-05,
+      "loss": 0.0011,
+      "step": 2972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.298266673068613e-05,
+      "loss": 0.0008,
+      "step": 2973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2980266000864264e-05,
+      "loss": 0.0007,
+      "step": 2974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29778652710424e-05,
+      "loss": 0.0008,
+      "step": 2975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2975464541220535e-05,
+      "loss": 0.0006,
+      "step": 2976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2973063811398667e-05,
+      "loss": 0.0004,
+      "step": 2977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29706630815768e-05,
+      "loss": 0.0018,
+      "step": 2978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296826235175494e-05,
+      "loss": 0.0035,
+      "step": 2979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296586162193307e-05,
+      "loss": 0.0007,
+      "step": 2980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.29634608921112e-05,
+      "loss": 0.0005,
+      "step": 2981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.296106016228934e-05,
+      "loss": 0.0004,
+      "step": 2982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295865943246747e-05,
+      "loss": 0.0012,
+      "step": 2983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295625870264561e-05,
+      "loss": 0.0017,
+      "step": 2984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295385797282374e-05,
+      "loss": 0.0025,
+      "step": 2985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.295145724300187e-05,
+      "loss": 0.0005,
+      "step": 2986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2949056513180005e-05,
+      "loss": 0.0002,
+      "step": 2987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2946655783358144e-05,
+      "loss": 0.0003,
+      "step": 2988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2944255053536275e-05,
+      "loss": 0.0005,
+      "step": 2989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.294185432371441e-05,
+      "loss": 0.0002,
+      "step": 2990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2939453593892546e-05,
+      "loss": 0.0004,
+      "step": 2991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2937052864070684e-05,
+      "loss": 0.0004,
+      "step": 2992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2934652134248816e-05,
+      "loss": 0.0013,
+      "step": 2993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.293225140442695e-05,
+      "loss": 0.6101,
+      "step": 2994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292985067460508e-05,
+      "loss": 0.0003,
+      "step": 2995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292744994478322e-05,
+      "loss": 0.7079,
+      "step": 2996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292504921496135e-05,
+      "loss": 0.0004,
+      "step": 2997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.292264848513948e-05,
+      "loss": 0.0004,
+      "step": 2998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2920247755317614e-05,
+      "loss": 0.0032,
+      "step": 2999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291784702549575e-05,
+      "loss": 0.0005,
+      "step": 3000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291544629567389e-05,
+      "loss": 0.0003,
+      "step": 3001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.291304556585202e-05,
+      "loss": 0.0006,
+      "step": 3002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2910644836030154e-05,
+      "loss": 0.0019,
+      "step": 3003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290824410620829e-05,
+      "loss": 0.0003,
+      "step": 3004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2905843376386425e-05,
+      "loss": 0.0031,
+      "step": 3005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2903442646564557e-05,
+      "loss": 0.0003,
+      "step": 3006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290104191674269e-05,
+      "loss": 0.077,
+      "step": 3007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289864118692083e-05,
+      "loss": 0.0016,
+      "step": 3008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2896240457098966e-05,
+      "loss": 0.0015,
+      "step": 3009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28938397272771e-05,
+      "loss": 0.0024,
+      "step": 3010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.289143899745523e-05,
+      "loss": 0.0003,
+      "step": 3011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288903826763336e-05,
+      "loss": 0.001,
+      "step": 3012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.28866375378115e-05,
+      "loss": 0.0003,
+      "step": 3013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288423680798963e-05,
+      "loss": 0.0002,
+      "step": 3014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.288183607816776e-05,
+      "loss": 0.0002,
+      "step": 3015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2879435348345895e-05,
+      "loss": 0.0084,
+      "step": 3016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2877034618524033e-05,
+      "loss": 0.0004,
+      "step": 3017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2874633888702165e-05,
+      "loss": 0.0003,
+      "step": 3018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2872233158880304e-05,
+      "loss": 0.0003,
+      "step": 3019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2869832429058436e-05,
+      "loss": 0.004,
+      "step": 3020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2867431699236574e-05,
+      "loss": 0.0003,
+      "step": 3021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2865030969414706e-05,
+      "loss": 0.0002,
+      "step": 3022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286263023959284e-05,
+      "loss": 0.0021,
+      "step": 3023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286022950977097e-05,
+      "loss": 0.0008,
+      "step": 3024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285782877994911e-05,
+      "loss": 0.0004,
+      "step": 3025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285542805012724e-05,
+      "loss": 0.0013,
+      "step": 3026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.285302732030537e-05,
+      "loss": 0.0029,
+      "step": 3027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.285062659048351e-05,
+      "loss": 0.0002,
+      "step": 3028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.284822586066164e-05,
+      "loss": 0.0016,
+      "step": 3029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.284582513083978e-05,
+      "loss": 0.0001,
+      "step": 3030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.284342440101791e-05,
+      "loss": 0.0002,
+      "step": 3031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2841023671196044e-05,
+      "loss": 0.0009,
+      "step": 3032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2838622941374176e-05,
+      "loss": 0.0002,
+      "step": 3033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2836222211552315e-05,
+      "loss": 0.0002,
+      "step": 3034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2833821481730447e-05,
+      "loss": 0.0001,
+      "step": 3035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.283142075190858e-05,
+      "loss": 0.0002,
+      "step": 3036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.282902002208672e-05,
+      "loss": 0.0009,
+      "step": 3037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2826619292264856e-05,
+      "loss": 0.0002,
+      "step": 3038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.282421856244299e-05,
+      "loss": 0.0009,
+      "step": 3039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.282181783262112e-05,
+      "loss": 0.0003,
+      "step": 3040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281941710279925e-05,
+      "loss": 0.0001,
+      "step": 3041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281701637297739e-05,
+      "loss": 0.0006,
+      "step": 3042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281461564315552e-05,
+      "loss": 0.0004,
+      "step": 3043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.281221491333365e-05,
+      "loss": 0.0003,
+      "step": 3044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2809814183511785e-05,
+      "loss": 0.0002,
+      "step": 3045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2807413453689923e-05,
+      "loss": 0.0002,
+      "step": 3046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.280501272386806e-05,
+      "loss": 0.0023,
+      "step": 3047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2802611994046194e-05,
+      "loss": 0.0014,
+      "step": 3048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2800211264224326e-05,
+      "loss": 0.0003,
+      "step": 3049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.279781053440246e-05,
+      "loss": 0.0004,
+      "step": 3050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2795409804580596e-05,
+      "loss": 0.0001,
+      "step": 3051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.279300907475873e-05,
+      "loss": 0.0001,
+      "step": 3052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.279060834493686e-05,
+      "loss": 0.0002,
+      "step": 3053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.278820761511499e-05,
+      "loss": 0.0006,
+      "step": 3054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.278580688529314e-05,
+      "loss": 0.0001,
+      "step": 3055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.278340615547127e-05,
+      "loss": 0.0006,
+      "step": 3056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27810054256494e-05,
+      "loss": 0.0007,
+      "step": 3057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277860469582753e-05,
+      "loss": 0.0004,
+      "step": 3058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.277620396600567e-05,
+      "loss": 0.0001,
+      "step": 3059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.27738032361838e-05,
+      "loss": 0.0002,
+      "step": 3060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2771402506361934e-05,
+      "loss": 0.0004,
+      "step": 3061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2769001776540066e-05,
+      "loss": 0.0001,
+      "step": 3062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2766601046718205e-05,
+      "loss": 0.814,
+      "step": 3063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.276420031689634e-05,
+      "loss": 0.268,
+      "step": 3064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2761799587074475e-05,
+      "loss": 0.0001,
+      "step": 3065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275939885725261e-05,
+      "loss": 0.0009,
+      "step": 3066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275699812743074e-05,
+      "loss": 0.0002,
+      "step": 3067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275459739760888e-05,
+      "loss": 0.0007,
+      "step": 3068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.275219666778701e-05,
+      "loss": 0.0002,
+      "step": 3069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274979593796514e-05,
+      "loss": 0.0002,
+      "step": 3070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274739520814327e-05,
+      "loss": 0.0007,
+      "step": 3071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274499447832141e-05,
+      "loss": 0.0001,
+      "step": 3072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274259374849955e-05,
+      "loss": 0.0002,
+      "step": 3073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274019301867768e-05,
+      "loss": 0.0001,
+      "step": 3074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2737792288855813e-05,
+      "loss": 0.0005,
+      "step": 3075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273539155903395e-05,
+      "loss": 0.0003,
+      "step": 3076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2732990829212084e-05,
+      "loss": 0.0089,
+      "step": 3077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2730590099390216e-05,
+      "loss": 0.0012,
+      "step": 3078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272818936956835e-05,
+      "loss": 0.001,
+      "step": 3079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2725788639746486e-05,
+      "loss": 0.0005,
+      "step": 3080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272338790992462e-05,
+      "loss": 0.0001,
+      "step": 3081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.272098718010275e-05,
+      "loss": 0.0001,
+      "step": 3082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271858645028089e-05,
+      "loss": 0.0004,
+      "step": 3083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271618572045902e-05,
+      "loss": 0.0005,
+      "step": 3084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271378499063716e-05,
+      "loss": 0.0001,
+      "step": 3085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271138426081529e-05,
+      "loss": 0.769,
+      "step": 3086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270898353099342e-05,
+      "loss": 0.0004,
+      "step": 3087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270658280117156e-05,
+      "loss": 0.0001,
+      "step": 3088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270418207134969e-05,
+      "loss": 0.0011,
+      "step": 3089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2701781341527824e-05,
+      "loss": 0.0006,
+      "step": 3090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2699380611705956e-05,
+      "loss": 0.0006,
+      "step": 3091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2696979881884095e-05,
+      "loss": 0.0006,
+      "step": 3092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.269457915206223e-05,
+      "loss": 0.0006,
+      "step": 3093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2692178422240365e-05,
+      "loss": 0.0002,
+      "step": 3094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26897776924185e-05,
+      "loss": 0.0001,
+      "step": 3095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268737696259663e-05,
+      "loss": 0.0001,
+      "step": 3096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.268497623277477e-05,
+      "loss": 0.001,
+      "step": 3097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.26825755029529e-05,
+      "loss": 0.0001,
+      "step": 3098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.268017477313103e-05,
+      "loss": 0.0001,
+      "step": 3099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.267777404330916e-05,
+      "loss": 0.0001,
+      "step": 3100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.26753733134873e-05,
+      "loss": 0.0001,
+      "step": 3101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.267297258366544e-05,
+      "loss": 0.0006,
+      "step": 3102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.267057185384357e-05,
+      "loss": 0.0001,
+      "step": 3103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2668171124021703e-05,
+      "loss": 0.0022,
+      "step": 3104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.266577039419984e-05,
+      "loss": 0.0001,
+      "step": 3105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2663369664377974e-05,
+      "loss": 0.0001,
+      "step": 3106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2660968934556106e-05,
+      "loss": 0.001,
+      "step": 3107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.265856820473424e-05,
+      "loss": 0.0007,
+      "step": 3108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2656167474912376e-05,
+      "loss": 0.0001,
+      "step": 3109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2653766745090515e-05,
+      "loss": 0.0011,
+      "step": 3110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2651366015268646e-05,
+      "loss": 0.0001,
+      "step": 3111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264896528544678e-05,
+      "loss": 0.0001,
+      "step": 3112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264656455562491e-05,
+      "loss": 0.0001,
+      "step": 3113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264416382580305e-05,
+      "loss": 0.576,
+      "step": 3114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.264176309598118e-05,
+      "loss": 0.0005,
+      "step": 3115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.263936236615931e-05,
+      "loss": 0.0001,
+      "step": 3116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2636961636337444e-05,
+      "loss": 0.0004,
+      "step": 3117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.263456090651558e-05,
+      "loss": 0.0001,
+      "step": 3118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.263216017669372e-05,
+      "loss": 0.0001,
+      "step": 3119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.262975944687185e-05,
+      "loss": 0.0001,
+      "step": 3120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2627358717049985e-05,
+      "loss": 0.0005,
+      "step": 3121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.262495798722812e-05,
+      "loss": 0.6215,
+      "step": 3122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2622557257406255e-05,
+      "loss": 0.0003,
+      "step": 3123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.262015652758439e-05,
+      "loss": 0.0001,
+      "step": 3124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261775579776252e-05,
+      "loss": 0.0002,
+      "step": 3125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261535506794066e-05,
+      "loss": 0.0005,
+      "step": 3126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261295433811879e-05,
+      "loss": 0.0001,
+      "step": 3127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.261055360829693e-05,
+      "loss": 0.0001,
+      "step": 3128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260815287847506e-05,
+      "loss": 0.0001,
+      "step": 3129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260575214865319e-05,
+      "loss": 0.0013,
+      "step": 3130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260335141883133e-05,
+      "loss": 0.0008,
+      "step": 3131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.260095068900946e-05,
+      "loss": 0.0002,
+      "step": 3132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.259854995918759e-05,
+      "loss": 0.0001,
+      "step": 3133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2596149229365725e-05,
+      "loss": 0.0001,
+      "step": 3134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2593748499543864e-05,
+      "loss": 0.0002,
+      "step": 3135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2591347769721996e-05,
+      "loss": 0.0008,
+      "step": 3136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.258894703990013e-05,
+      "loss": 0.0002,
+      "step": 3137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2586546310078266e-05,
+      "loss": 0.0001,
+      "step": 3138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2584145580256405e-05,
+      "loss": 0.0001,
+      "step": 3139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2581744850434536e-05,
+      "loss": 0.0001,
+      "step": 3140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257934412061267e-05,
+      "loss": 0.0015,
+      "step": 3141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25769433907908e-05,
+      "loss": 0.0001,
+      "step": 3142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257454266096894e-05,
+      "loss": 0.0001,
+      "step": 3143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.257214193114707e-05,
+      "loss": 0.0001,
+      "step": 3144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.25697412013252e-05,
+      "loss": 0.0013,
+      "step": 3145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2567340471503334e-05,
+      "loss": 0.0001,
+      "step": 3146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256493974168147e-05,
+      "loss": 0.0001,
+      "step": 3147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256253901185961e-05,
+      "loss": 0.0001,
+      "step": 3148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.256013828203774e-05,
+      "loss": 0.0001,
+      "step": 3149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2557737552215875e-05,
+      "loss": 0.0011,
+      "step": 3150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2555336822394006e-05,
+      "loss": 0.001,
+      "step": 3151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2552936092572145e-05,
+      "loss": 0.0129,
+      "step": 3152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.255053536275028e-05,
+      "loss": 0.0001,
+      "step": 3153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254813463292841e-05,
+      "loss": 0.0079,
+      "step": 3154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254573390310655e-05,
+      "loss": 0.0001,
+      "step": 3155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2543333173284686e-05,
+      "loss": 0.0001,
+      "step": 3156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.254093244346282e-05,
+      "loss": 0.0001,
+      "step": 3157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253853171364095e-05,
+      "loss": 1.5359,
+      "step": 3158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253613098381908e-05,
+      "loss": 0.0006,
+      "step": 3159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253373025399722e-05,
+      "loss": 0.0008,
+      "step": 3160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253132952417535e-05,
+      "loss": 0.0009,
+      "step": 3161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252892879435348e-05,
+      "loss": 0.0001,
+      "step": 3162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2526528064531615e-05,
+      "loss": 0.0001,
+      "step": 3163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2524127334709754e-05,
+      "loss": 0.0001,
+      "step": 3164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.252172660488789e-05,
+      "loss": 0.0007,
+      "step": 3165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2519325875066024e-05,
+      "loss": 0.0001,
+      "step": 3166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2516925145244156e-05,
+      "loss": 0.0007,
+      "step": 3167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.251452441542229e-05,
+      "loss": 0.0007,
+      "step": 3168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2512123685600426e-05,
+      "loss": 0.001,
+      "step": 3169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250972295577856e-05,
+      "loss": 0.7301,
+      "step": 3170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250732222595669e-05,
+      "loss": 0.0001,
+      "step": 3171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250492149613483e-05,
+      "loss": 0.0084,
+      "step": 3172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.250252076631296e-05,
+      "loss": 0.0002,
+      "step": 3173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.25001200364911e-05,
+      "loss": 0.0003,
+      "step": 3174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.249771930666923e-05,
+      "loss": 0.0069,
+      "step": 3175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.249531857684736e-05,
+      "loss": 0.0006,
+      "step": 3176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24929178470255e-05,
+      "loss": 0.0002,
+      "step": 3177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.249051711720363e-05,
+      "loss": 0.0005,
+      "step": 3178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2488116387381765e-05,
+      "loss": 0.0001,
+      "step": 3179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2485715657559896e-05,
+      "loss": 0.6275,
+      "step": 3180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2483314927738035e-05,
+      "loss": 0.0001,
+      "step": 3181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.248091419791617e-05,
+      "loss": 0.0001,
+      "step": 3182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2478513468094305e-05,
+      "loss": 0.0001,
+      "step": 3183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.247611273827244e-05,
+      "loss": 0.0001,
+      "step": 3184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2473712008450576e-05,
+      "loss": 0.0001,
+      "step": 3185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.247131127862871e-05,
+      "loss": 0.0001,
+      "step": 3186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246891054880684e-05,
+      "loss": 0.0028,
+      "step": 3187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246650981898497e-05,
+      "loss": 0.0004,
+      "step": 3188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246410908916311e-05,
+      "loss": 0.0,
+      "step": 3189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.246170835934124e-05,
+      "loss": 0.0001,
+      "step": 3190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.245930762951937e-05,
+      "loss": 0.0001,
+      "step": 3191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.245690689969751e-05,
+      "loss": 0.0001,
+      "step": 3192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2454506169875644e-05,
+      "loss": 0.7207,
+      "step": 3193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.245210544005378e-05,
+      "loss": 0.0001,
+      "step": 3194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2449704710231914e-05,
+      "loss": 0.523,
+      "step": 3195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2447303980410046e-05,
+      "loss": 0.017,
+      "step": 3196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244490325058818e-05,
+      "loss": 0.002,
+      "step": 3197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2442502520766316e-05,
+      "loss": 0.0001,
+      "step": 3198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.244010179094445e-05,
+      "loss": 0.0013,
+      "step": 3199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243770106112258e-05,
+      "loss": 0.0001,
+      "step": 3200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243530033130071e-05,
+      "loss": 0.0002,
+      "step": 3201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243289960147886e-05,
+      "loss": 0.0001,
+      "step": 3202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.243049887165699e-05,
+      "loss": 0.5907,
+      "step": 3203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242809814183512e-05,
+      "loss": 0.0016,
+      "step": 3204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242569741201325e-05,
+      "loss": 0.0016,
+      "step": 3205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242329668219139e-05,
+      "loss": 0.4452,
+      "step": 3206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.242089595236952e-05,
+      "loss": 0.0001,
+      "step": 3207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2418495222547655e-05,
+      "loss": 0.0006,
+      "step": 3208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2416094492725786e-05,
+      "loss": 0.0031,
+      "step": 3209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2413693762903925e-05,
+      "loss": 0.0007,
+      "step": 3210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2411293033082064e-05,
+      "loss": 0.0001,
+      "step": 3211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2408892303260195e-05,
+      "loss": 0.0076,
+      "step": 3212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240649157343833e-05,
+      "loss": 0.0001,
+      "step": 3213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240409084361646e-05,
+      "loss": 0.0194,
+      "step": 3214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.24016901137946e-05,
+      "loss": 0.0054,
+      "step": 3215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239928938397273e-05,
+      "loss": 0.0001,
+      "step": 3216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239688865415086e-05,
+      "loss": 0.4041,
+      "step": 3217
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239448792432899e-05,
+      "loss": 0.0016,
+      "step": 3218
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.239208719450713e-05,
+      "loss": 0.091,
+      "step": 3219
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238968646468527e-05,
+      "loss": 0.0001,
+      "step": 3220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23872857348634e-05,
+      "loss": 0.0009,
+      "step": 3221
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2384885005041534e-05,
+      "loss": 0.0001,
+      "step": 3222
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.238248427521967e-05,
+      "loss": 0.0073,
+      "step": 3223
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2380083545397804e-05,
+      "loss": 0.0001,
+      "step": 3224
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2377682815575936e-05,
+      "loss": 0.0028,
+      "step": 3225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237528208575407e-05,
+      "loss": 0.0033,
+      "step": 3226
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2372881355932206e-05,
+      "loss": 0.0001,
+      "step": 3227
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.237048062611034e-05,
+      "loss": 0.0001,
+      "step": 3228
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2368079896288477e-05,
+      "loss": 0.0001,
+      "step": 3229
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236567916646661e-05,
+      "loss": 0.0177,
+      "step": 3230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236327843664474e-05,
+      "loss": 0.0023,
+      "step": 3231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.236087770682288e-05,
+      "loss": 1.2166,
+      "step": 3232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235847697700101e-05,
+      "loss": 0.0137,
+      "step": 3233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235607624717914e-05,
+      "loss": 0.396,
+      "step": 3234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2353675517357274e-05,
+      "loss": 1.1375,
+      "step": 3235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.235127478753541e-05,
+      "loss": 0.0463,
+      "step": 3236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2348874057713545e-05,
+      "loss": 0.0151,
+      "step": 3237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.234647332789168e-05,
+      "loss": 0.0187,
+      "step": 3238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2344072598069815e-05,
+      "loss": 0.0097,
+      "step": 3239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2341671868247954e-05,
+      "loss": 0.0024,
+      "step": 3240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2339271138426085e-05,
+      "loss": 0.7908,
+      "step": 3241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.233687040860422e-05,
+      "loss": 0.0021,
+      "step": 3242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.233446967878235e-05,
+      "loss": 0.0022,
+      "step": 3243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.233206894896049e-05,
+      "loss": 0.0028,
+      "step": 3244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232966821913862e-05,
+      "loss": 0.0023,
+      "step": 3245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232726748931675e-05,
+      "loss": 0.0014,
+      "step": 3246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232486675949489e-05,
+      "loss": 0.003,
+      "step": 3247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232246602967302e-05,
+      "loss": 0.0024,
+      "step": 3248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.232006529985116e-05,
+      "loss": 0.0035,
+      "step": 3249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.231766457002929e-05,
+      "loss": 0.0013,
+      "step": 3250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2315263840207424e-05,
+      "loss": 0.0081,
+      "step": 3251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.231286311038556e-05,
+      "loss": 0.0025,
+      "step": 3252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2310462380563694e-05,
+      "loss": 0.0021,
+      "step": 3253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2308061650741826e-05,
+      "loss": 0.0384,
+      "step": 3254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.230566092091996e-05,
+      "loss": 0.0029,
+      "step": 3255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2303260191098096e-05,
+      "loss": 0.0034,
+      "step": 3256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2300859461276235e-05,
+      "loss": 0.0028,
+      "step": 3257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2298458731454367e-05,
+      "loss": 0.0025,
+      "step": 3258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22960580016325e-05,
+      "loss": 0.0023,
+      "step": 3259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.229365727181063e-05,
+      "loss": 0.6653,
+      "step": 3260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.229125654198877e-05,
+      "loss": 0.0022,
+      "step": 3261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22888558121669e-05,
+      "loss": 0.0026,
+      "step": 3262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.228645508234503e-05,
+      "loss": 0.0021,
+      "step": 3263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2284054352523164e-05,
+      "loss": 0.002,
+      "step": 3264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22816536227013e-05,
+      "loss": 0.0014,
+      "step": 3265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227925289287944e-05,
+      "loss": 0.0035,
+      "step": 3266
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.227685216305757e-05,
+      "loss": 0.003,
+      "step": 3267
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2274451433235705e-05,
+      "loss": 0.004,
+      "step": 3268
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2272050703413843e-05,
+      "loss": 0.6824,
+      "step": 3269
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2269649973591975e-05,
+      "loss": 0.0024,
+      "step": 3270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226724924377011e-05,
+      "loss": 0.0045,
+      "step": 3271
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226484851394824e-05,
+      "loss": 0.0016,
+      "step": 3272
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226244778412638e-05,
+      "loss": 0.003,
+      "step": 3273
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226004705430451e-05,
+      "loss": 0.0027,
+      "step": 3274
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225764632448265e-05,
+      "loss": 0.6708,
+      "step": 3275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225524559466078e-05,
+      "loss": 0.0033,
+      "step": 3276
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225284486483891e-05,
+      "loss": 0.6432,
+      "step": 3277
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225044413501705e-05,
+      "loss": 0.0034,
+      "step": 3278
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.224804340519518e-05,
+      "loss": 0.0058,
+      "step": 3279
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2245642675373314e-05,
+      "loss": 0.0049,
+      "step": 3280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2243241945551445e-05,
+      "loss": 0.5702,
+      "step": 3281
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2240841215729584e-05,
+      "loss": 0.0092,
+      "step": 3282
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2238440485907716e-05,
+      "loss": 0.4832,
+      "step": 3283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2236039756085854e-05,
+      "loss": 0.0064,
+      "step": 3284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2233639026263986e-05,
+      "loss": 0.0109,
+      "step": 3285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2231238296442125e-05,
+      "loss": 0.5296,
+      "step": 3286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2228837566620257e-05,
+      "loss": 0.0108,
+      "step": 3287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222643683679839e-05,
+      "loss": 0.4839,
+      "step": 3288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222403610697652e-05,
+      "loss": 0.0196,
+      "step": 3289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.222163537715466e-05,
+      "loss": 0.0226,
+      "step": 3290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221923464733279e-05,
+      "loss": 0.0344,
+      "step": 3291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221683391751092e-05,
+      "loss": 0.0281,
+      "step": 3292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221443318768906e-05,
+      "loss": 0.4073,
+      "step": 3293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221203245786719e-05,
+      "loss": 0.0203,
+      "step": 3294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220963172804533e-05,
+      "loss": 0.0067,
+      "step": 3295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220723099822346e-05,
+      "loss": 0.0052,
+      "step": 3296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2204830268401595e-05,
+      "loss": 0.0059,
+      "step": 3297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220242953857973e-05,
+      "loss": 0.0035,
+      "step": 3298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2200028808757865e-05,
+      "loss": 0.0041,
+      "step": 3299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2197628078936e-05,
+      "loss": 0.0025,
+      "step": 3300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219522734911413e-05,
+      "loss": 0.0035,
+      "step": 3301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.219282661929227e-05,
+      "loss": 0.0031,
+      "step": 3302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2190425889470406e-05,
+      "loss": 0.004,
+      "step": 3303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218802515964854e-05,
+      "loss": 0.0028,
+      "step": 3304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.218562442982667e-05,
+      "loss": 0.0039,
+      "step": 3305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.21832237000048e-05,
+      "loss": 0.6563,
+      "step": 3306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.218082297018294e-05,
+      "loss": 0.002,
+      "step": 3307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.217842224036107e-05,
+      "loss": 0.0033,
+      "step": 3308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2176021510539204e-05,
+      "loss": 1.0284,
+      "step": 3309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2173620780717335e-05,
+      "loss": 0.0031,
+      "step": 3310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2171220050895474e-05,
+      "loss": 0.003,
+      "step": 3311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.216881932107361e-05,
+      "loss": 0.002,
+      "step": 3312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2166418591251744e-05,
+      "loss": 0.0032,
+      "step": 3313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2164017861429876e-05,
+      "loss": 0.0037,
+      "step": 3314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.216161713160801e-05,
+      "loss": 0.0033,
+      "step": 3315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2159216401786147e-05,
+      "loss": 0.0022,
+      "step": 3316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.215681567196428e-05,
+      "loss": 0.0032,
+      "step": 3317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.215441494214241e-05,
+      "loss": 0.0024,
+      "step": 3318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.215201421232054e-05,
+      "loss": 0.0027,
+      "step": 3319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214961348249869e-05,
+      "loss": 0.0029,
+      "step": 3320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214721275267682e-05,
+      "loss": 0.0025,
+      "step": 3321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214481202285495e-05,
+      "loss": 0.0025,
+      "step": 3322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214241129303308e-05,
+      "loss": 0.0026,
+      "step": 3323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.214001056321122e-05,
+      "loss": 0.0028,
+      "step": 3324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.213760983338935e-05,
+      "loss": 0.002,
+      "step": 3325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2135209103567485e-05,
+      "loss": 0.0023,
+      "step": 3326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.213280837374562e-05,
+      "loss": 0.0023,
+      "step": 3327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2130407643923755e-05,
+      "loss": 0.8462,
+      "step": 3328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212800691410189e-05,
+      "loss": 0.0021,
+      "step": 3329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2125606184280026e-05,
+      "loss": 0.0023,
+      "step": 3330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212320545445816e-05,
+      "loss": 0.0027,
+      "step": 3331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.212080472463629e-05,
+      "loss": 0.0024,
+      "step": 3332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211840399481443e-05,
+      "loss": 0.0026,
+      "step": 3333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211600326499256e-05,
+      "loss": 0.0026,
+      "step": 3334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211360253517069e-05,
+      "loss": 0.0027,
+      "step": 3335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.211120180534883e-05,
+      "loss": 0.0025,
+      "step": 3336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210880107552696e-05,
+      "loss": 0.0023,
+      "step": 3337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2106400345705094e-05,
+      "loss": 0.0032,
+      "step": 3338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.210399961588323e-05,
+      "loss": 0.0021,
+      "step": 3339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2101598886061364e-05,
+      "loss": 0.7851,
+      "step": 3340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20991981562395e-05,
+      "loss": 0.7173,
+      "step": 3341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2096797426417634e-05,
+      "loss": 0.7102,
+      "step": 3342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2094396696595766e-05,
+      "loss": 0.0041,
+      "step": 3343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20919959667739e-05,
+      "loss": 0.0072,
+      "step": 3344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2089595236952036e-05,
+      "loss": 0.0053,
+      "step": 3345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208719450713017e-05,
+      "loss": 0.4778,
+      "step": 3346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.20847937773083e-05,
+      "loss": 0.5177,
+      "step": 3347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.208239304748644e-05,
+      "loss": 0.0205,
+      "step": 3348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207999231766457e-05,
+      "loss": 0.2962,
+      "step": 3349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207759158784271e-05,
+      "loss": 0.0422,
+      "step": 3350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207519085802084e-05,
+      "loss": 0.074,
+      "step": 3351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207279012819897e-05,
+      "loss": 0.2735,
+      "step": 3352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.207038939837711e-05,
+      "loss": 0.3014,
+      "step": 3353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206798866855524e-05,
+      "loss": 0.272,
+      "step": 3354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2065587938733375e-05,
+      "loss": 0.0834,
+      "step": 3355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.206318720891151e-05,
+      "loss": 0.289,
+      "step": 3356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2060786479089645e-05,
+      "loss": 0.0905,
+      "step": 3357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2058385749267784e-05,
+      "loss": 0.1164,
+      "step": 3358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2055985019445916e-05,
+      "loss": 0.3909,
+      "step": 3359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205358428962405e-05,
+      "loss": 0.1153,
+      "step": 3360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.205118355980218e-05,
+      "loss": 0.0745,
+      "step": 3361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204878282998032e-05,
+      "loss": 0.0715,
+      "step": 3362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204638210015845e-05,
+      "loss": 0.0457,
+      "step": 3363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204398137033658e-05,
+      "loss": 0.0284,
+      "step": 3364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.204158064051471e-05,
+      "loss": 0.027,
+      "step": 3365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203917991069286e-05,
+      "loss": 0.0236,
+      "step": 3366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203677918087099e-05,
+      "loss": 0.0127,
+      "step": 3367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203437845104912e-05,
+      "loss": 0.0176,
+      "step": 3368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2031977721227254e-05,
+      "loss": 0.0136,
+      "step": 3369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202957699140539e-05,
+      "loss": 0.0068,
+      "step": 3370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2027176261583524e-05,
+      "loss": 0.005,
+      "step": 3371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2024775531761656e-05,
+      "loss": 0.0044,
+      "step": 3372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202237480193979e-05,
+      "loss": 0.0041,
+      "step": 3373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2019974072117926e-05,
+      "loss": 0.0039,
+      "step": 3374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2017573342296065e-05,
+      "loss": 0.0031,
+      "step": 3375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.20151726124742e-05,
+      "loss": 0.7388,
+      "step": 3376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.201277188265233e-05,
+      "loss": 0.0017,
+      "step": 3377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.201037115283046e-05,
+      "loss": 0.002,
+      "step": 3378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.20079704230086e-05,
+      "loss": 0.002,
+      "step": 3379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.200556969318673e-05,
+      "loss": 0.0014,
+      "step": 3380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.200316896336486e-05,
+      "loss": 0.8325,
+      "step": 3381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2000768233542994e-05,
+      "loss": 0.0016,
+      "step": 3382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.199836750372113e-05,
+      "loss": 0.0017,
+      "step": 3383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.199596677389927e-05,
+      "loss": 0.002,
+      "step": 3384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19935660440774e-05,
+      "loss": 0.0018,
+      "step": 3385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1991165314255535e-05,
+      "loss": 0.0017,
+      "step": 3386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1988764584433674e-05,
+      "loss": 0.0018,
+      "step": 3387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1986363854611806e-05,
+      "loss": 0.8439,
+      "step": 3388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.198396312478994e-05,
+      "loss": 0.0016,
+      "step": 3389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.198156239496807e-05,
+      "loss": 0.0019,
+      "step": 3390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.197916166514621e-05,
+      "loss": 0.0014,
+      "step": 3391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.197676093532434e-05,
+      "loss": 0.0015,
+      "step": 3392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.197436020550247e-05,
+      "loss": 0.0016,
+      "step": 3393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.197195947568061e-05,
+      "loss": 0.0018,
+      "step": 3394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.196955874585874e-05,
+      "loss": 0.0019,
+      "step": 3395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.196715801603688e-05,
+      "loss": 0.0016,
+      "step": 3396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.196475728621501e-05,
+      "loss": 0.0017,
+      "step": 3397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1962356556393144e-05,
+      "loss": 0.0018,
+      "step": 3398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1959955826571276e-05,
+      "loss": 0.7974,
+      "step": 3399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1957555096749414e-05,
+      "loss": 0.0017,
+      "step": 3400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1955154366927546e-05,
+      "loss": 0.0016,
+      "step": 3401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.195275363710568e-05,
+      "loss": 0.0014,
+      "step": 3402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1950352907283816e-05,
+      "loss": 0.0016,
+      "step": 3403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1947952177461955e-05,
+      "loss": 0.0018,
+      "step": 3404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194555144764009e-05,
+      "loss": 0.0017,
+      "step": 3405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194315071781822e-05,
+      "loss": 0.0018,
+      "step": 3406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.194074998799635e-05,
+      "loss": 0.0016,
+      "step": 3407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193834925817449e-05,
+      "loss": 0.0016,
+      "step": 3408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193594852835262e-05,
+      "loss": 0.0015,
+      "step": 3409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.193354779853075e-05,
+      "loss": 0.002,
+      "step": 3410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1931147068708884e-05,
+      "loss": 0.7724,
+      "step": 3411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192874633888702e-05,
+      "loss": 0.7866,
+      "step": 3412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192634560906516e-05,
+      "loss": 0.0017,
+      "step": 3413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.192394487924329e-05,
+      "loss": 0.0015,
+      "step": 3414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1921544149421425e-05,
+      "loss": 0.002,
+      "step": 3415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191914341959956e-05,
+      "loss": 0.0019,
+      "step": 3416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1916742689777696e-05,
+      "loss": 0.0019,
+      "step": 3417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191434195995583e-05,
+      "loss": 0.002,
+      "step": 3418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.191194123013396e-05,
+      "loss": 0.0018,
+      "step": 3419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19095405003121e-05,
+      "loss": 0.002,
+      "step": 3420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1907139770490236e-05,
+      "loss": 0.0018,
+      "step": 3421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190473904066837e-05,
+      "loss": 0.0018,
+      "step": 3422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.19023383108465e-05,
+      "loss": 0.0015,
+      "step": 3423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189993758102463e-05,
+      "loss": 0.8274,
+      "step": 3424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.189753685120277e-05,
+      "loss": 0.0018,
+      "step": 3425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.18951361213809e-05,
+      "loss": 0.7624,
+      "step": 3426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1892735391559034e-05,
+      "loss": 0.0018,
+      "step": 3427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1890334661737166e-05,
+      "loss": 0.002,
+      "step": 3428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1887933931915304e-05,
+      "loss": 0.0016,
+      "step": 3429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.188553320209344e-05,
+      "loss": 0.0023,
+      "step": 3430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1883132472271575e-05,
+      "loss": 0.0059,
+      "step": 3431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1880731742449706e-05,
+      "loss": 0.0021,
+      "step": 3432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1878331012627845e-05,
+      "loss": 0.002,
+      "step": 3433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187593028280598e-05,
+      "loss": 0.0023,
+      "step": 3434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187352955298411e-05,
+      "loss": 0.0019,
+      "step": 3435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187112882316224e-05,
+      "loss": 0.002,
+      "step": 3436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186872809334038e-05,
+      "loss": 0.0022,
+      "step": 3437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186632736351851e-05,
+      "loss": 0.0021,
+      "step": 3438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186392663369665e-05,
+      "loss": 0.002,
+      "step": 3439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186152590387478e-05,
+      "loss": 0.0019,
+      "step": 3440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185912517405291e-05,
+      "loss": 0.0031,
+      "step": 3441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185672444423105e-05,
+      "loss": 0.0016,
+      "step": 3442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.185432371440918e-05,
+      "loss": 0.0015,
+      "step": 3443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1851922984587315e-05,
+      "loss": 0.0023,
+      "step": 3444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.184952225476545e-05,
+      "loss": 0.0015,
+      "step": 3445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1847121524943585e-05,
+      "loss": 0.8957,
+      "step": 3446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.184472079512172e-05,
+      "loss": 0.0017,
+      "step": 3447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1842320065299856e-05,
+      "loss": 0.0014,
+      "step": 3448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.183991933547799e-05,
+      "loss": 0.0021,
+      "step": 3449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1837518605656126e-05,
+      "loss": 0.0018,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.183511787583426e-05,
+      "loss": 0.8008,
+      "step": 3451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.183271714601239e-05,
+      "loss": 0.0016,
+      "step": 3452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.183031641619052e-05,
+      "loss": 0.787,
+      "step": 3453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.182791568636866e-05,
+      "loss": 0.002,
+      "step": 3454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.182551495654679e-05,
+      "loss": 0.0018,
+      "step": 3455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1823114226724924e-05,
+      "loss": 0.0017,
+      "step": 3456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1820713496903056e-05,
+      "loss": 0.0018,
+      "step": 3457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1818312767081194e-05,
+      "loss": 0.0022,
+      "step": 3458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.181591203725933e-05,
+      "loss": 0.0022,
+      "step": 3459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1813511307437465e-05,
+      "loss": 0.0019,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1811110577615596e-05,
+      "loss": 0.7879,
+      "step": 3461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.180870984779373e-05,
+      "loss": 0.0017,
+      "step": 3462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.180630911797187e-05,
+      "loss": 0.0021,
+      "step": 3463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.180390838815e-05,
+      "loss": 0.0021,
+      "step": 3464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.180150765832813e-05,
+      "loss": 0.0024,
+      "step": 3465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179910692850626e-05,
+      "loss": 0.0023,
+      "step": 3466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179670619868441e-05,
+      "loss": 0.8026,
+      "step": 3467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179430546886254e-05,
+      "loss": 0.0022,
+      "step": 3468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.179190473904067e-05,
+      "loss": 0.6813,
+      "step": 3469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.17895040092188e-05,
+      "loss": 0.0026,
+      "step": 3470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.178710327939694e-05,
+      "loss": 0.0024,
+      "step": 3471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.178470254957507e-05,
+      "loss": 0.7068,
+      "step": 3472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1782301819753205e-05,
+      "loss": 0.0025,
+      "step": 3473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177990108993134e-05,
+      "loss": 0.2276,
+      "step": 3474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1777500360109475e-05,
+      "loss": 0.0024,
+      "step": 3475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1775099630287614e-05,
+      "loss": 0.0039,
+      "step": 3476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1772698900465746e-05,
+      "loss": 0.0039,
+      "step": 3477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.177029817064388e-05,
+      "loss": 0.0051,
+      "step": 3478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176789744082201e-05,
+      "loss": 0.0039,
+      "step": 3479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176549671100015e-05,
+      "loss": 0.0066,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176309598117828e-05,
+      "loss": 0.0043,
+      "step": 3481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.176069525135641e-05,
+      "loss": 0.6559,
+      "step": 3482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1758294521534543e-05,
+      "loss": 0.003,
+      "step": 3483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175589379171268e-05,
+      "loss": 0.0035,
+      "step": 3484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175349306189082e-05,
+      "loss": 0.0061,
+      "step": 3485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.175109233206895e-05,
+      "loss": 0.0061,
+      "step": 3486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1748691602247084e-05,
+      "loss": 0.004,
+      "step": 3487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.174629087242522e-05,
+      "loss": 0.0052,
+      "step": 3488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1743890142603355e-05,
+      "loss": 0.005,
+      "step": 3489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1741489412781486e-05,
+      "loss": 0.0049,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173908868295962e-05,
+      "loss": 0.7718,
+      "step": 3491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173668795313776e-05,
+      "loss": 0.7727,
+      "step": 3492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173428722331589e-05,
+      "loss": 0.0028,
+      "step": 3493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173188649349403e-05,
+      "loss": 0.0038,
+      "step": 3494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172948576367216e-05,
+      "loss": 0.7462,
+      "step": 3495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172708503385029e-05,
+      "loss": 0.0027,
+      "step": 3496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172468430402843e-05,
+      "loss": 0.0044,
+      "step": 3497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.172228357420656e-05,
+      "loss": 0.0068,
+      "step": 3498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171988284438469e-05,
+      "loss": 0.0048,
+      "step": 3499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1717482114562825e-05,
+      "loss": 0.0053,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171508138474096e-05,
+      "loss": 0.7192,
+      "step": 3501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1712680654919095e-05,
+      "loss": 0.0044,
+      "step": 3502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1710279925097234e-05,
+      "loss": 0.0048,
+      "step": 3503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1707879195275365e-05,
+      "loss": 0.0047,
+      "step": 3504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1705478465453504e-05,
+      "loss": 0.0047,
+      "step": 3505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1703077735631636e-05,
+      "loss": 0.0044,
+      "step": 3506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.170067700580977e-05,
+      "loss": 0.1749,
+      "step": 3507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16982762759879e-05,
+      "loss": 0.0056,
+      "step": 3508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169587554616604e-05,
+      "loss": 0.0042,
+      "step": 3509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.169347481634417e-05,
+      "loss": 0.0034,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16910740865223e-05,
+      "loss": 0.004,
+      "step": 3511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1688673356700433e-05,
+      "loss": 0.0046,
+      "step": 3512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168627262687857e-05,
+      "loss": 1.4848,
+      "step": 3513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.168387189705671e-05,
+      "loss": 0.0049,
+      "step": 3514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.168147116723484e-05,
+      "loss": 0.0036,
+      "step": 3515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1679070437412974e-05,
+      "loss": 0.6574,
+      "step": 3516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.167666970759111e-05,
+      "loss": 1.3221,
+      "step": 3517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1674268977769245e-05,
+      "loss": 0.0053,
+      "step": 3518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1671868247947376e-05,
+      "loss": 0.6638,
+      "step": 3519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.166946751812551e-05,
+      "loss": 0.6521,
+      "step": 3520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.166706678830365e-05,
+      "loss": 0.6418,
+      "step": 3521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1664666058481785e-05,
+      "loss": 0.0055,
+      "step": 3522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.166226532865992e-05,
+      "loss": 0.0074,
+      "step": 3523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165986459883805e-05,
+      "loss": 0.5805,
+      "step": 3524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165746386901618e-05,
+      "loss": 0.5993,
+      "step": 3525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165506313919432e-05,
+      "loss": 1.0875,
+      "step": 3526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165266240937245e-05,
+      "loss": 0.0108,
+      "step": 3527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.165026167955058e-05,
+      "loss": 1.0069,
+      "step": 3528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1647860949728715e-05,
+      "loss": 0.9996,
+      "step": 3529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.164546021990685e-05,
+      "loss": 0.4075,
+      "step": 3530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.164305949008499e-05,
+      "loss": 0.4015,
+      "step": 3531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1640658760263124e-05,
+      "loss": 0.045,
+      "step": 3532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1638258030441255e-05,
+      "loss": 0.4339,
+      "step": 3533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1635857300619394e-05,
+      "loss": 0.6639,
+      "step": 3534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1633456570797526e-05,
+      "loss": 0.09,
+      "step": 3535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.163105584097566e-05,
+      "loss": 0.1131,
+      "step": 3536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.162865511115379e-05,
+      "loss": 0.3375,
+      "step": 3537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.162625438133193e-05,
+      "loss": 0.3508,
+      "step": 3538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.162385365151006e-05,
+      "loss": 0.0678,
+      "step": 3539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.16214529216882e-05,
+      "loss": 0.1653,
+      "step": 3540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.161905219186633e-05,
+      "loss": 0.433,
+      "step": 3541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.161665146204446e-05,
+      "loss": 0.0726,
+      "step": 3542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.16142507322226e-05,
+      "loss": 0.0825,
+      "step": 3543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.161185000240073e-05,
+      "loss": 0.0891,
+      "step": 3544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1609449272578864e-05,
+      "loss": 0.3583,
+      "step": 3545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1607048542756996e-05,
+      "loss": 0.1224,
+      "step": 3546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1604647812935134e-05,
+      "loss": 0.081,
+      "step": 3547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1602247083113266e-05,
+      "loss": 0.0617,
+      "step": 3548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1599846353291405e-05,
+      "loss": 0.0676,
+      "step": 3549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159744562346954e-05,
+      "loss": 0.032,
+      "step": 3550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1595044893647675e-05,
+      "loss": 0.0308,
+      "step": 3551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159264416382581e-05,
+      "loss": 0.0176,
+      "step": 3552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.159024343400394e-05,
+      "loss": 0.0268,
+      "step": 3553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158784270418207e-05,
+      "loss": 0.0206,
+      "step": 3554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158544197436021e-05,
+      "loss": 0.0096,
+      "step": 3555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158304124453834e-05,
+      "loss": 0.0183,
+      "step": 3556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.158064051471647e-05,
+      "loss": 0.008,
+      "step": 3557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157823978489461e-05,
+      "loss": 0.0099,
+      "step": 3558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157583905507274e-05,
+      "loss": 0.0092,
+      "step": 3559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.157343832525088e-05,
+      "loss": 0.0121,
+      "step": 3560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1571037595429014e-05,
+      "loss": 0.0049,
+      "step": 3561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1568636865607145e-05,
+      "loss": 0.0093,
+      "step": 3562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156623613578528e-05,
+      "loss": 0.0052,
+      "step": 3563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1563835405963416e-05,
+      "loss": 0.0052,
+      "step": 3564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156143467614155e-05,
+      "loss": 0.7759,
+      "step": 3565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155903394631968e-05,
+      "loss": 0.0035,
+      "step": 3566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155663321649782e-05,
+      "loss": 0.0047,
+      "step": 3567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1554232486675957e-05,
+      "loss": 0.0031,
+      "step": 3568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.155183175685409e-05,
+      "loss": 0.7724,
+      "step": 3569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154943102703222e-05,
+      "loss": 0.0046,
+      "step": 3570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154703029721035e-05,
+      "loss": 0.0043,
+      "step": 3571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154462956738849e-05,
+      "loss": 0.0042,
+      "step": 3572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.154222883756662e-05,
+      "loss": 0.7265,
+      "step": 3573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1539828107744754e-05,
+      "loss": 0.0033,
+      "step": 3574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1537427377922886e-05,
+      "loss": 0.0041,
+      "step": 3575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1535026648101024e-05,
+      "loss": 0.0038,
+      "step": 3576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.153262591827916e-05,
+      "loss": 0.0037,
+      "step": 3577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1530225188457295e-05,
+      "loss": 0.6996,
+      "step": 3578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152782445863543e-05,
+      "loss": 0.0043,
+      "step": 3579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152542372881356e-05,
+      "loss": 0.0038,
+      "step": 3580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.15230229989917e-05,
+      "loss": 0.0039,
+      "step": 3581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152062226916983e-05,
+      "loss": 0.7061,
+      "step": 3582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151822153934796e-05,
+      "loss": 0.0041,
+      "step": 3583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.15158208095261e-05,
+      "loss": 0.0043,
+      "step": 3584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.151342007970423e-05,
+      "loss": 0.0048,
+      "step": 3585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.151101934988237e-05,
+      "loss": 0.0037,
+      "step": 3586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.15086186200605e-05,
+      "loss": 0.6384,
+      "step": 3587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.150621789023863e-05,
+      "loss": 0.0049,
+      "step": 3588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.150381716041677e-05,
+      "loss": 0.0047,
+      "step": 3589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1501416430594904e-05,
+      "loss": 0.0047,
+      "step": 3590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1499015700773035e-05,
+      "loss": 0.0033,
+      "step": 3591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.149661497095117e-05,
+      "loss": 0.0038,
+      "step": 3592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1494214241129306e-05,
+      "loss": 0.0044,
+      "step": 3593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.149181351130744e-05,
+      "loss": 0.006,
+      "step": 3594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1489412781485576e-05,
+      "loss": 0.006,
+      "step": 3595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.148701205166371e-05,
+      "loss": 0.0083,
+      "step": 3596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.148461132184184e-05,
+      "loss": 0.0042,
+      "step": 3597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.148221059201998e-05,
+      "loss": 0.0038,
+      "step": 3598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147980986219811e-05,
+      "loss": 0.0033,
+      "step": 3599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147740913237624e-05,
+      "loss": 0.0031,
+      "step": 3600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147500840255438e-05,
+      "loss": 0.0044,
+      "step": 3601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.147260767273251e-05,
+      "loss": 0.0035,
+      "step": 3602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1470206942910644e-05,
+      "loss": 0.0043,
+      "step": 3603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.146780621308878e-05,
+      "loss": 0.0043,
+      "step": 3604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1465405483266914e-05,
+      "loss": 0.0036,
+      "step": 3605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.146300475344505e-05,
+      "loss": 0.0034,
+      "step": 3606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1460604023623185e-05,
+      "loss": 0.0034,
+      "step": 3607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145820329380132e-05,
+      "loss": 0.0038,
+      "step": 3608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145580256397945e-05,
+      "loss": 0.5311,
+      "step": 3609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145340183415759e-05,
+      "loss": 0.0028,
+      "step": 3610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.145100110433572e-05,
+      "loss": 0.003,
+      "step": 3611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144860037451385e-05,
+      "loss": 0.004,
+      "step": 3612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144619964469199e-05,
+      "loss": 0.0035,
+      "step": 3613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144379891487013e-05,
+      "loss": 0.0031,
+      "step": 3614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.144139818504826e-05,
+      "loss": 0.0035,
+      "step": 3615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143899745522639e-05,
+      "loss": 0.0034,
+      "step": 3616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143659672540452e-05,
+      "loss": 0.0032,
+      "step": 3617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.143419599558266e-05,
+      "loss": 0.0027,
+      "step": 3618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1431795265760794e-05,
+      "loss": 0.0028,
+      "step": 3619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1429394535938925e-05,
+      "loss": 0.0025,
+      "step": 3620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.142699380611706e-05,
+      "loss": 0.5769,
+      "step": 3621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1424593076295196e-05,
+      "loss": 0.0032,
+      "step": 3622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1422192346473334e-05,
+      "loss": 0.0039,
+      "step": 3623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1419791616651466e-05,
+      "loss": 0.0027,
+      "step": 3624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.14173908868296e-05,
+      "loss": 0.0033,
+      "step": 3625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141499015700773e-05,
+      "loss": 0.5208,
+      "step": 3626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141258942718587e-05,
+      "loss": 0.0028,
+      "step": 3627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1410188697364e-05,
+      "loss": 0.0035,
+      "step": 3628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140778796754213e-05,
+      "loss": 0.0032,
+      "step": 3629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1405387237720264e-05,
+      "loss": 0.0049,
+      "step": 3630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140298650789841e-05,
+      "loss": 0.0039,
+      "step": 3631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.140058577807654e-05,
+      "loss": 0.0024,
+      "step": 3632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139818504825467e-05,
+      "loss": 0.7885,
+      "step": 3633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1395784318432804e-05,
+      "loss": 0.0029,
+      "step": 3634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.139338358861094e-05,
+      "loss": 0.0032,
+      "step": 3635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1390982858789075e-05,
+      "loss": 0.8143,
+      "step": 3636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1388582128967207e-05,
+      "loss": 0.0029,
+      "step": 3637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138618139914534e-05,
+      "loss": 0.0026,
+      "step": 3638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138378066932348e-05,
+      "loss": 0.003,
+      "step": 3639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138137993950161e-05,
+      "loss": 0.002,
+      "step": 3640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137897920967975e-05,
+      "loss": 0.0019,
+      "step": 3641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137657847985788e-05,
+      "loss": 0.0025,
+      "step": 3642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137417775003601e-05,
+      "loss": 0.0101,
+      "step": 3643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.137177702021415e-05,
+      "loss": 0.0027,
+      "step": 3644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136937629039228e-05,
+      "loss": 0.0027,
+      "step": 3645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.136697556057041e-05,
+      "loss": 0.003,
+      "step": 3646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1364574830748545e-05,
+      "loss": 0.0025,
+      "step": 3647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1362174100926683e-05,
+      "loss": 0.0024,
+      "step": 3648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1359773371104815e-05,
+      "loss": 0.0023,
+      "step": 3649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1357372641282954e-05,
+      "loss": 0.0022,
+      "step": 3650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1354971911461086e-05,
+      "loss": 0.0025,
+      "step": 3651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1352571181639224e-05,
+      "loss": 0.0026,
+      "step": 3652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1350170451817356e-05,
+      "loss": 0.0022,
+      "step": 3653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134776972199549e-05,
+      "loss": 0.0025,
+      "step": 3654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134536899217362e-05,
+      "loss": 0.0025,
+      "step": 3655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134296826235176e-05,
+      "loss": 0.0025,
+      "step": 3656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.134056753252989e-05,
+      "loss": 0.002,
+      "step": 3657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133816680270802e-05,
+      "loss": 0.0029,
+      "step": 3658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133576607288616e-05,
+      "loss": 0.0025,
+      "step": 3659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133336534306429e-05,
+      "loss": 0.0024,
+      "step": 3660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.133096461324243e-05,
+      "loss": 0.0022,
+      "step": 3661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.132856388342056e-05,
+      "loss": 0.0023,
+      "step": 3662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1326163153598694e-05,
+      "loss": 0.0021,
+      "step": 3663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1323762423776826e-05,
+      "loss": 0.5612,
+      "step": 3664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1321361693954965e-05,
+      "loss": 0.0023,
+      "step": 3665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1318960964133097e-05,
+      "loss": 0.002,
+      "step": 3666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.131656023431123e-05,
+      "loss": 0.0018,
+      "step": 3667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.131415950448937e-05,
+      "loss": 0.0019,
+      "step": 3668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1311758774667506e-05,
+      "loss": 0.7206,
+      "step": 3669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130935804484564e-05,
+      "loss": 0.0019,
+      "step": 3670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130695731502377e-05,
+      "loss": 0.0021,
+      "step": 3671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.13045565852019e-05,
+      "loss": 0.0019,
+      "step": 3672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.130215585538004e-05,
+      "loss": 0.002,
+      "step": 3673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.129975512555817e-05,
+      "loss": 0.0049,
+      "step": 3674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12973543957363e-05,
+      "loss": 0.002,
+      "step": 3675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1294953665914435e-05,
+      "loss": 0.0021,
+      "step": 3676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1292552936092573e-05,
+      "loss": 0.0016,
+      "step": 3677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.129015220627071e-05,
+      "loss": 0.0019,
+      "step": 3678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1287751476448844e-05,
+      "loss": 0.0018,
+      "step": 3679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1285350746626976e-05,
+      "loss": 0.0042,
+      "step": 3680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1282950016805114e-05,
+      "loss": 0.002,
+      "step": 3681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1280549286983246e-05,
+      "loss": 0.0017,
+      "step": 3682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127814855716138e-05,
+      "loss": 0.7886,
+      "step": 3683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127574782733951e-05,
+      "loss": 0.0021,
+      "step": 3684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127334709751765e-05,
+      "loss": 0.7807,
+      "step": 3685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.127094636769579e-05,
+      "loss": 0.0021,
+      "step": 3686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126854563787392e-05,
+      "loss": 0.0018,
+      "step": 3687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126614490805205e-05,
+      "loss": 0.0019,
+      "step": 3688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126374417823018e-05,
+      "loss": 0.6091,
+      "step": 3689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.126134344840832e-05,
+      "loss": 0.002,
+      "step": 3690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.125894271858645e-05,
+      "loss": 0.0019,
+      "step": 3691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1256541988764584e-05,
+      "loss": 0.7737,
+      "step": 3692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1254141258942716e-05,
+      "loss": 0.0022,
+      "step": 3693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1251740529120855e-05,
+      "loss": 0.0022,
+      "step": 3694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124933979929899e-05,
+      "loss": 0.0022,
+      "step": 3695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1246939069477125e-05,
+      "loss": 0.0024,
+      "step": 3696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.124453833965526e-05,
+      "loss": 0.0023,
+      "step": 3697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1242137609833395e-05,
+      "loss": 0.0023,
+      "step": 3698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123973688001153e-05,
+      "loss": 0.0027,
+      "step": 3699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123733615018966e-05,
+      "loss": 0.0024,
+      "step": 3700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123493542036779e-05,
+      "loss": 0.0024,
+      "step": 3701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123253469054593e-05,
+      "loss": 0.0019,
+      "step": 3702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.123013396072406e-05,
+      "loss": 0.0026,
+      "step": 3703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122773323090219e-05,
+      "loss": 0.0024,
+      "step": 3704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.122533250108033e-05,
+      "loss": 1.5737,
+      "step": 3705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1222931771258463e-05,
+      "loss": 0.0021,
+      "step": 3706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12205310414366e-05,
+      "loss": 0.0021,
+      "step": 3707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1218130311614734e-05,
+      "loss": 0.0025,
+      "step": 3708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1215729581792866e-05,
+      "loss": 0.0028,
+      "step": 3709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1213328851971e-05,
+      "loss": 0.0024,
+      "step": 3710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1210928122149136e-05,
+      "loss": 0.7562,
+      "step": 3711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120852739232727e-05,
+      "loss": 0.0029,
+      "step": 3712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12061266625054e-05,
+      "loss": 0.0025,
+      "step": 3713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120372593268354e-05,
+      "loss": 0.0024,
+      "step": 3714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120132520286168e-05,
+      "loss": 0.007,
+      "step": 3715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119892447303981e-05,
+      "loss": 0.002,
+      "step": 3716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119652374321794e-05,
+      "loss": 0.0022,
+      "step": 3717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119412301339607e-05,
+      "loss": 0.0025,
+      "step": 3718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.119172228357421e-05,
+      "loss": 0.0024,
+      "step": 3719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118932155375234e-05,
+      "loss": 0.0033,
+      "step": 3720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1186920823930474e-05,
+      "loss": 0.0019,
+      "step": 3721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1184520094108606e-05,
+      "loss": 0.0027,
+      "step": 3722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1182119364286745e-05,
+      "loss": 0.0025,
+      "step": 3723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117971863446488e-05,
+      "loss": 0.0024,
+      "step": 3724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1177317904643015e-05,
+      "loss": 0.0021,
+      "step": 3725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117491717482115e-05,
+      "loss": 1.6091,
+      "step": 3726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117251644499928e-05,
+      "loss": 0.0023,
+      "step": 3727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.117011571517742e-05,
+      "loss": 0.811,
+      "step": 3728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116771498535555e-05,
+      "loss": 0.002,
+      "step": 3729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116531425553368e-05,
+      "loss": 0.0021,
+      "step": 3730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116291352571181e-05,
+      "loss": 0.0023,
+      "step": 3731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.116051279588996e-05,
+      "loss": 0.0024,
+      "step": 3732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.115811206606809e-05,
+      "loss": 0.0023,
+      "step": 3733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.115571133624622e-05,
+      "loss": 0.0024,
+      "step": 3734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1153310606424353e-05,
+      "loss": 0.0024,
+      "step": 3735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.115090987660249e-05,
+      "loss": 0.0024,
+      "step": 3736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1148509146780624e-05,
+      "loss": 0.0023,
+      "step": 3737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1146108416958756e-05,
+      "loss": 0.0025,
+      "step": 3738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.114370768713689e-05,
+      "loss": 0.0023,
+      "step": 3739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1141306957315026e-05,
+      "loss": 0.0025,
+      "step": 3740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1138906227493165e-05,
+      "loss": 0.0025,
+      "step": 3741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1136505497671296e-05,
+      "loss": 0.0022,
+      "step": 3742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.113410476784943e-05,
+      "loss": 0.0024,
+      "step": 3743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.113170403802756e-05,
+      "loss": 0.0023,
+      "step": 3744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.11293033082057e-05,
+      "loss": 0.0033,
+      "step": 3745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.112690257838383e-05,
+      "loss": 0.0022,
+      "step": 3746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.112450184856196e-05,
+      "loss": 0.0027,
+      "step": 3747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1122101118740094e-05,
+      "loss": 0.002,
+      "step": 3748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111970038891823e-05,
+      "loss": 0.0021,
+      "step": 3749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111729965909637e-05,
+      "loss": 0.0022,
+      "step": 3750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.11148989292745e-05,
+      "loss": 0.002,
+      "step": 3751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1112498199452635e-05,
+      "loss": 0.0021,
+      "step": 3752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.111009746963077e-05,
+      "loss": 0.0022,
+      "step": 3753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1107696739808905e-05,
+      "loss": 0.0022,
+      "step": 3754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110529600998704e-05,
+      "loss": 0.0022,
+      "step": 3755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110289528016517e-05,
+      "loss": 0.0021,
+      "step": 3756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.110049455034331e-05,
+      "loss": 0.0019,
+      "step": 3757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109809382052144e-05,
+      "loss": 0.002,
+      "step": 3758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109569309069958e-05,
+      "loss": 0.0019,
+      "step": 3759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109329236087771e-05,
+      "loss": 0.0015,
+      "step": 3760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.109089163105584e-05,
+      "loss": 0.0016,
+      "step": 3761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108849090123398e-05,
+      "loss": 0.0018,
+      "step": 3762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108609017141211e-05,
+      "loss": 0.0016,
+      "step": 3763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108368944159024e-05,
+      "loss": 0.0019,
+      "step": 3764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.108128871176838e-05,
+      "loss": 0.0017,
+      "step": 3765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1078887981946514e-05,
+      "loss": 0.0018,
+      "step": 3766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1076487252124646e-05,
+      "loss": 0.0019,
+      "step": 3767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107408652230278e-05,
+      "loss": 0.0024,
+      "step": 3768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1071685792480916e-05,
+      "loss": 0.0015,
+      "step": 3769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1069285062659055e-05,
+      "loss": 0.002,
+      "step": 3770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1066884332837186e-05,
+      "loss": 0.0016,
+      "step": 3771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106448360301532e-05,
+      "loss": 0.0015,
+      "step": 3772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.106208287319345e-05,
+      "loss": 0.0018,
+      "step": 3773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105968214337159e-05,
+      "loss": 0.0015,
+      "step": 3774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105728141354972e-05,
+      "loss": 0.0017,
+      "step": 3775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105488068372785e-05,
+      "loss": 0.772,
+      "step": 3776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1052479953905984e-05,
+      "loss": 0.0014,
+      "step": 3777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.105007922408413e-05,
+      "loss": 0.0014,
+      "step": 3778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104767849426226e-05,
+      "loss": 0.0014,
+      "step": 3779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104527776444039e-05,
+      "loss": 0.0014,
+      "step": 3780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1042877034618525e-05,
+      "loss": 0.0015,
+      "step": 3781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.104047630479666e-05,
+      "loss": 0.0016,
+      "step": 3782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1038075574974795e-05,
+      "loss": 0.0015,
+      "step": 3783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103567484515293e-05,
+      "loss": 0.0014,
+      "step": 3784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103327411533106e-05,
+      "loss": 0.0015,
+      "step": 3785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10308733855092e-05,
+      "loss": 0.7877,
+      "step": 3786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1028472655687336e-05,
+      "loss": 0.0015,
+      "step": 3787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102607192586547e-05,
+      "loss": 0.0024,
+      "step": 3788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.10236711960436e-05,
+      "loss": 0.0016,
+      "step": 3789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102127046622173e-05,
+      "loss": 0.0014,
+      "step": 3790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.101886973639987e-05,
+      "loss": 0.0013,
+      "step": 3791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1016469006578e-05,
+      "loss": 0.0017,
+      "step": 3792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.101406827675613e-05,
+      "loss": 0.0014,
+      "step": 3793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1011667546934265e-05,
+      "loss": 0.0013,
+      "step": 3794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1009266817112404e-05,
+      "loss": 0.0015,
+      "step": 3795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.100686608729054e-05,
+      "loss": 0.0039,
+      "step": 3796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1004465357468674e-05,
+      "loss": 0.0016,
+      "step": 3797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1002064627646806e-05,
+      "loss": 0.0019,
+      "step": 3798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0999663897824944e-05,
+      "loss": 0.0016,
+      "step": 3799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0997263168003076e-05,
+      "loss": 0.0017,
+      "step": 3800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099486243818121e-05,
+      "loss": 0.0015,
+      "step": 3801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099246170835934e-05,
+      "loss": 0.0019,
+      "step": 3802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.099006097853748e-05,
+      "loss": 0.0017,
+      "step": 3803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.098766024871561e-05,
+      "loss": 0.0014,
+      "step": 3804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.098525951889375e-05,
+      "loss": 0.7295,
+      "step": 3805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.098285878907188e-05,
+      "loss": 0.0013,
+      "step": 3806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.098045805925001e-05,
+      "loss": 0.0016,
+      "step": 3807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.097805732942815e-05,
+      "loss": 0.0015,
+      "step": 3808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.097565659960628e-05,
+      "loss": 0.0016,
+      "step": 3809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0973255869784415e-05,
+      "loss": 0.5173,
+      "step": 3810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0970855139962546e-05,
+      "loss": 0.0014,
+      "step": 3811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0968454410140685e-05,
+      "loss": 0.0014,
+      "step": 3812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.096605368031882e-05,
+      "loss": 0.0014,
+      "step": 3813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0963652950496955e-05,
+      "loss": 0.0014,
+      "step": 3814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.096125222067509e-05,
+      "loss": 0.0179,
+      "step": 3815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0958851490853226e-05,
+      "loss": 0.4898,
+      "step": 3816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.095645076103136e-05,
+      "loss": 0.0013,
+      "step": 3817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.095405003120949e-05,
+      "loss": 0.0016,
+      "step": 3818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.095164930138762e-05,
+      "loss": 0.0015,
+      "step": 3819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094924857156576e-05,
+      "loss": 0.0011,
+      "step": 3820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094684784174389e-05,
+      "loss": 0.0029,
+      "step": 3821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.094444711192202e-05,
+      "loss": 0.0026,
+      "step": 3822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0942046382100155e-05,
+      "loss": 0.0015,
+      "step": 3823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0939645652278294e-05,
+      "loss": 0.0012,
+      "step": 3824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093724492245643e-05,
+      "loss": 0.0015,
+      "step": 3825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0934844192634564e-05,
+      "loss": 0.8585,
+      "step": 3826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0932443462812696e-05,
+      "loss": 0.0014,
+      "step": 3827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.093004273299083e-05,
+      "loss": 0.0017,
+      "step": 3828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0927642003168966e-05,
+      "loss": 0.0018,
+      "step": 3829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09252412733471e-05,
+      "loss": 0.0023,
+      "step": 3830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092284054352523e-05,
+      "loss": 0.0013,
+      "step": 3831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.092043981370337e-05,
+      "loss": 0.0031,
+      "step": 3832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091803908388151e-05,
+      "loss": 0.0019,
+      "step": 3833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091563835405964e-05,
+      "loss": 0.0015,
+      "step": 3834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.091323762423777e-05,
+      "loss": 0.0017,
+      "step": 3835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.09108368944159e-05,
+      "loss": 0.0015,
+      "step": 3836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090843616459404e-05,
+      "loss": 0.0015,
+      "step": 3837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090603543477217e-05,
+      "loss": 0.0012,
+      "step": 3838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0903634704950305e-05,
+      "loss": 0.0013,
+      "step": 3839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0901233975128436e-05,
+      "loss": 0.0014,
+      "step": 3840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0898833245306575e-05,
+      "loss": 0.0012,
+      "step": 3841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0896432515484714e-05,
+      "loss": 0.0014,
+      "step": 3842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0894031785662845e-05,
+      "loss": 0.0826,
+      "step": 3843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.089163105584098e-05,
+      "loss": 0.002,
+      "step": 3844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088923032601911e-05,
+      "loss": 0.0027,
+      "step": 3845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088682959619725e-05,
+      "loss": 0.0014,
+      "step": 3846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088442886637538e-05,
+      "loss": 0.0014,
+      "step": 3847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.088202813655351e-05,
+      "loss": 0.0011,
+      "step": 3848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087962740673165e-05,
+      "loss": 0.7716,
+      "step": 3849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087722667690978e-05,
+      "loss": 0.0013,
+      "step": 3850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087482594708792e-05,
+      "loss": 0.0012,
+      "step": 3851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.087242521726605e-05,
+      "loss": 0.0011,
+      "step": 3852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0870024487444184e-05,
+      "loss": 0.0013,
+      "step": 3853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086762375762232e-05,
+      "loss": 0.6748,
+      "step": 3854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0865223027800454e-05,
+      "loss": 0.0014,
+      "step": 3855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0862822297978586e-05,
+      "loss": 0.0013,
+      "step": 3856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.086042156815672e-05,
+      "loss": 0.0012,
+      "step": 3857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0858020838334856e-05,
+      "loss": 0.0015,
+      "step": 3858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085562010851299e-05,
+      "loss": 0.0014,
+      "step": 3859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0853219378691127e-05,
+      "loss": 0.0017,
+      "step": 3860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.085081864886926e-05,
+      "loss": 0.0012,
+      "step": 3861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.08484179190474e-05,
+      "loss": 0.0015,
+      "step": 3862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084601718922553e-05,
+      "loss": 0.0013,
+      "step": 3863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084361645940366e-05,
+      "loss": 0.0012,
+      "step": 3864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.084121572958179e-05,
+      "loss": 0.064,
+      "step": 3865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.083881499975993e-05,
+      "loss": 0.0013,
+      "step": 3866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.083641426993806e-05,
+      "loss": 0.0012,
+      "step": 3867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0834013540116195e-05,
+      "loss": 0.6782,
+      "step": 3868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.083161281029433e-05,
+      "loss": 0.0013,
+      "step": 3869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0829212080472465e-05,
+      "loss": 0.0013,
+      "step": 3870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0826811350650604e-05,
+      "loss": 0.0011,
+      "step": 3871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0824410620828735e-05,
+      "loss": 0.0012,
+      "step": 3872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082200989100687e-05,
+      "loss": 0.0013,
+      "step": 3873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0819609161185e-05,
+      "loss": 0.0022,
+      "step": 3874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081720843136314e-05,
+      "loss": 0.0014,
+      "step": 3875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081480770154127e-05,
+      "loss": 0.0011,
+      "step": 3876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.08124069717194e-05,
+      "loss": 0.0012,
+      "step": 3877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.081000624189754e-05,
+      "loss": 0.0016,
+      "step": 3878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080760551207568e-05,
+      "loss": 0.0053,
+      "step": 3879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080520478225381e-05,
+      "loss": 0.0023,
+      "step": 3880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.080280405243194e-05,
+      "loss": 0.001,
+      "step": 3881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0800403322610074e-05,
+      "loss": 0.0012,
+      "step": 3882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079800259278821e-05,
+      "loss": 0.0011,
+      "step": 3883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0795601862966344e-05,
+      "loss": 0.001,
+      "step": 3884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0793201133144476e-05,
+      "loss": 0.0013,
+      "step": 3885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079080040332261e-05,
+      "loss": 0.0011,
+      "step": 3886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0788399673500746e-05,
+      "loss": 0.0012,
+      "step": 3887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0785998943678885e-05,
+      "loss": 1.7198,
+      "step": 3888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0783598213857017e-05,
+      "loss": 0.0009,
+      "step": 3889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.078119748403515e-05,
+      "loss": 0.8398,
+      "step": 3890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077879675421328e-05,
+      "loss": 1.6355,
+      "step": 3891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077639602439142e-05,
+      "loss": 0.001,
+      "step": 3892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077399529456955e-05,
+      "loss": 0.0012,
+      "step": 3893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.077159456474768e-05,
+      "loss": 0.0012,
+      "step": 3894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0769193834925814e-05,
+      "loss": 0.0012,
+      "step": 3895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076679310510395e-05,
+      "loss": 0.0012,
+      "step": 3896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076439237528209e-05,
+      "loss": 0.0202,
+      "step": 3897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076199164546022e-05,
+      "loss": 0.0014,
+      "step": 3898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0759590915638355e-05,
+      "loss": 0.8182,
+      "step": 3899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0757190185816493e-05,
+      "loss": 0.0013,
+      "step": 3900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0754789455994625e-05,
+      "loss": 0.0018,
+      "step": 3901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075238872617276e-05,
+      "loss": 0.7745,
+      "step": 3902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074998799635089e-05,
+      "loss": 0.0023,
+      "step": 3903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074758726652903e-05,
+      "loss": 0.0022,
+      "step": 3904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074518653670716e-05,
+      "loss": 0.0026,
+      "step": 3905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07427858068853e-05,
+      "loss": 0.7049,
+      "step": 3906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.074038507706343e-05,
+      "loss": 0.0039,
+      "step": 3907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073798434724156e-05,
+      "loss": 0.0027,
+      "step": 3908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07355836174197e-05,
+      "loss": 0.0037,
+      "step": 3909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073318288759783e-05,
+      "loss": 0.0022,
+      "step": 3910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0730782157775964e-05,
+      "loss": 0.0035,
+      "step": 3911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0728381427954095e-05,
+      "loss": 0.6673,
+      "step": 3912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0725980698132234e-05,
+      "loss": 0.0043,
+      "step": 3913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0723579968310366e-05,
+      "loss": 0.0055,
+      "step": 3914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0721179238488504e-05,
+      "loss": 0.0032,
+      "step": 3915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0718778508666636e-05,
+      "loss": 0.0042,
+      "step": 3916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0716377778844775e-05,
+      "loss": 0.6603,
+      "step": 3917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0713977049022907e-05,
+      "loss": 0.859,
+      "step": 3918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.071157631920104e-05,
+      "loss": 0.5993,
+      "step": 3919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070917558937917e-05,
+      "loss": 0.0069,
+      "step": 3920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070677485955731e-05,
+      "loss": 0.6244,
+      "step": 3921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070437412973544e-05,
+      "loss": 0.0066,
+      "step": 3922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.070197339991357e-05,
+      "loss": 0.0083,
+      "step": 3923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069957267009171e-05,
+      "loss": 0.0065,
+      "step": 3924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069717194026984e-05,
+      "loss": 0.0064,
+      "step": 3925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069477121044798e-05,
+      "loss": 0.5565,
+      "step": 3926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.069237048062611e-05,
+      "loss": 0.5503,
+      "step": 3927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0689969750804245e-05,
+      "loss": 0.0102,
+      "step": 3928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0687569020982383e-05,
+      "loss": 0.0089,
+      "step": 3929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0685168291160515e-05,
+      "loss": 0.012,
+      "step": 3930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068276756133865e-05,
+      "loss": 0.0175,
+      "step": 3931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068036683151678e-05,
+      "loss": 0.0113,
+      "step": 3932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067796610169492e-05,
+      "loss": 0.0135,
+      "step": 3933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0675565371873056e-05,
+      "loss": 0.5343,
+      "step": 3934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067316464205119e-05,
+      "loss": 1.0518,
+      "step": 3935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067076391222932e-05,
+      "loss": 0.5092,
+      "step": 3936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066836318240745e-05,
+      "loss": 0.0076,
+      "step": 3937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066596245258559e-05,
+      "loss": 0.557,
+      "step": 3938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066356172276372e-05,
+      "loss": 0.9502,
+      "step": 3939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0661160992941854e-05,
+      "loss": 1.8763,
+      "step": 3940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0658760263119985e-05,
+      "loss": 0.8797,
+      "step": 3941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0656359533298124e-05,
+      "loss": 0.8757,
+      "step": 3942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065395880347626e-05,
+      "loss": 0.8824,
+      "step": 3943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0651558073654394e-05,
+      "loss": 0.4225,
+      "step": 3944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0649157343832526e-05,
+      "loss": 0.7789,
+      "step": 3945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0646756614010665e-05,
+      "loss": 0.6435,
+      "step": 3946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0644355884188797e-05,
+      "loss": 0.7086,
+      "step": 3947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064195515436693e-05,
+      "loss": 0.3284,
+      "step": 3948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063955442454506e-05,
+      "loss": 0.8966,
+      "step": 3949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06371536947232e-05,
+      "loss": 1.0235,
+      "step": 3950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063475296490133e-05,
+      "loss": 0.3588,
+      "step": 3951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063235223507947e-05,
+      "loss": 0.5706,
+      "step": 3952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06299515052576e-05,
+      "loss": 0.5948,
+      "step": 3953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062755077543573e-05,
+      "loss": 0.6343,
+      "step": 3954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062515004561387e-05,
+      "loss": 0.4641,
+      "step": 3955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0622749315792e-05,
+      "loss": 0.3411,
+      "step": 3956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0620348585970135e-05,
+      "loss": 0.5753,
+      "step": 3957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061794785614827e-05,
+      "loss": 0.4651,
+      "step": 3958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0615547126326405e-05,
+      "loss": 0.3346,
+      "step": 3959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.061314639650454e-05,
+      "loss": 0.4775,
+      "step": 3960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0610745666682676e-05,
+      "loss": 0.5181,
+      "step": 3961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060834493686081e-05,
+      "loss": 0.2973,
+      "step": 3962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0605944207038946e-05,
+      "loss": 0.4455,
+      "step": 3963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060354347721708e-05,
+      "loss": 0.3096,
+      "step": 3964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.060114274739521e-05,
+      "loss": 0.4697,
+      "step": 3965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059874201757334e-05,
+      "loss": 0.2211,
+      "step": 3966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059634128775148e-05,
+      "loss": 0.388,
+      "step": 3967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059394055792961e-05,
+      "loss": 0.2594,
+      "step": 3968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0591539828107744e-05,
+      "loss": 0.2187,
+      "step": 3969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058913909828588e-05,
+      "loss": 0.18,
+      "step": 3970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0586738368464014e-05,
+      "loss": 0.4354,
+      "step": 3971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058433763864215e-05,
+      "loss": 0.348,
+      "step": 3972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0581936908820284e-05,
+      "loss": 0.1658,
+      "step": 3973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0579536178998416e-05,
+      "loss": 0.2318,
+      "step": 3974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057713544917655e-05,
+      "loss": 0.1192,
+      "step": 3975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0574734719354686e-05,
+      "loss": 0.0623,
+      "step": 3976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057233398953282e-05,
+      "loss": 0.1662,
+      "step": 3977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056993325971095e-05,
+      "loss": 0.0982,
+      "step": 3978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056753252988909e-05,
+      "loss": 0.0324,
+      "step": 3979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056513180006723e-05,
+      "loss": 0.0928,
+      "step": 3980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056273107024536e-05,
+      "loss": 0.0461,
+      "step": 3981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056033034042349e-05,
+      "loss": 0.0245,
+      "step": 3982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055792961060162e-05,
+      "loss": 0.0152,
+      "step": 3983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055552888077976e-05,
+      "loss": 0.0185,
+      "step": 3984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055312815095789e-05,
+      "loss": 0.8348,
+      "step": 3985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0550727421136025e-05,
+      "loss": 0.7987,
+      "step": 3986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054832669131416e-05,
+      "loss": 0.7646,
+      "step": 3987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0545925961492295e-05,
+      "loss": 0.0124,
+      "step": 3988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0543525231670434e-05,
+      "loss": 0.0234,
+      "step": 3989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0541124501848566e-05,
+      "loss": 0.0102,
+      "step": 3990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05387237720267e-05,
+      "loss": 0.0079,
+      "step": 3991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053632304220483e-05,
+      "loss": 0.3864,
+      "step": 3992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053392231238297e-05,
+      "loss": 0.0087,
+      "step": 3993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05315215825611e-05,
+      "loss": 0.0075,
+      "step": 3994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052912085273923e-05,
+      "loss": 0.0086,
+      "step": 3995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052672012291736e-05,
+      "loss": 0.0056,
+      "step": 3996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052431939309551e-05,
+      "loss": 0.0078,
+      "step": 3997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052191866327364e-05,
+      "loss": 0.0064,
+      "step": 3998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051951793345177e-05,
+      "loss": 0.0073,
+      "step": 3999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0517117203629904e-05,
+      "loss": 0.0061,
+      "step": 4000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051471647380804e-05,
+      "loss": 0.4332,
+      "step": 4001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0512315743986174e-05,
+      "loss": 0.6002,
+      "step": 4002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0509915014164306e-05,
+      "loss": 0.0122,
+      "step": 4003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.050751428434244e-05,
+      "loss": 0.0067,
+      "step": 4004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0505113554520576e-05,
+      "loss": 0.5839,
+      "step": 4005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0502712824698715e-05,
+      "loss": 0.0055,
+      "step": 4006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.050031209487685e-05,
+      "loss": 0.5857,
+      "step": 4007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049791136505498e-05,
+      "loss": 1.2315,
+      "step": 4008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049551063523311e-05,
+      "loss": 1.1873,
+      "step": 4009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049310990541125e-05,
+      "loss": 0.005,
+      "step": 4010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.049070917558938e-05,
+      "loss": 1.101,
+      "step": 4011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048830844576751e-05,
+      "loss": 1.0642,
+      "step": 4012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048590771594565e-05,
+      "loss": 0.0094,
+      "step": 4013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.048350698612378e-05,
+      "loss": 0.0072,
+      "step": 4014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0481106256301915e-05,
+      "loss": 0.0136,
+      "step": 4015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.047870552648005e-05,
+      "loss": 0.0106,
+      "step": 4016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0476304796658185e-05,
+      "loss": 0.0109,
+      "step": 4017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0473904066836324e-05,
+      "loss": 0.0107,
+      "step": 4018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0471503337014456e-05,
+      "loss": 0.0091,
+      "step": 4019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046910260719259e-05,
+      "loss": 0.5297,
+      "step": 4020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046670187737072e-05,
+      "loss": 0.0066,
+      "step": 4021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046430114754886e-05,
+      "loss": 0.758,
+      "step": 4022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046190041772699e-05,
+      "loss": 0.0081,
+      "step": 4023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045949968790512e-05,
+      "loss": 0.0075,
+      "step": 4024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045709895808326e-05,
+      "loss": 0.781,
+      "step": 4025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04546982282614e-05,
+      "loss": 0.0097,
+      "step": 4026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045229749843953e-05,
+      "loss": 0.0104,
+      "step": 4027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044989676861766e-05,
+      "loss": 0.5411,
+      "step": 4028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0447496038795794e-05,
+      "loss": 0.0027,
+      "step": 4029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044509530897393e-05,
+      "loss": 0.007,
+      "step": 4030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0442694579152064e-05,
+      "loss": 0.0144,
+      "step": 4031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0440293849330196e-05,
+      "loss": 0.0076,
+      "step": 4032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043789311950833e-05,
+      "loss": 0.7039,
+      "step": 4033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0435492389686466e-05,
+      "loss": 0.005,
+      "step": 4034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0433091659864605e-05,
+      "loss": 0.0088,
+      "step": 4035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043069093004274e-05,
+      "loss": 0.6274,
+      "step": 4036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042829020022087e-05,
+      "loss": 0.7084,
+      "step": 4037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0425889470399e-05,
+      "loss": 0.0081,
+      "step": 4038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042348874057714e-05,
+      "loss": 0.0084,
+      "step": 4039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042108801075527e-05,
+      "loss": 0.0058,
+      "step": 4040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04186872809334e-05,
+      "loss": 0.5487,
+      "step": 4041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0416286551111534e-05,
+      "loss": 0.4288,
+      "step": 4042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041388582128968e-05,
+      "loss": 0.008,
+      "step": 4043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.041148509146781e-05,
+      "loss": 0.0047,
+      "step": 4044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040908436164594e-05,
+      "loss": 0.0063,
+      "step": 4045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0406683631824075e-05,
+      "loss": 0.0075,
+      "step": 4046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0404282902002214e-05,
+      "loss": 0.0062,
+      "step": 4047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0401882172180346e-05,
+      "loss": 0.0075,
+      "step": 4048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039948144235848e-05,
+      "loss": 0.0056,
+      "step": 4049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039708071253661e-05,
+      "loss": 0.0068,
+      "step": 4050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039467998271475e-05,
+      "loss": 0.0068,
+      "step": 4051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0392279252892886e-05,
+      "loss": 0.6561,
+      "step": 4052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038987852307102e-05,
+      "loss": 0.0067,
+      "step": 4053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038747779324915e-05,
+      "loss": 0.0056,
+      "step": 4054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038507706342728e-05,
+      "loss": 1.3462,
+      "step": 4055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038267633360542e-05,
+      "loss": 0.005,
+      "step": 4056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038027560378355e-05,
+      "loss": 0.0062,
+      "step": 4057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0377874873961684e-05,
+      "loss": 0.0058,
+      "step": 4058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0375474144139816e-05,
+      "loss": 0.0055,
+      "step": 4059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0373073414317954e-05,
+      "loss": 0.6301,
+      "step": 4060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037067268449609e-05,
+      "loss": 0.0054,
+      "step": 4061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0368271954674225e-05,
+      "loss": 0.0044,
+      "step": 4062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0365871224852356e-05,
+      "loss": 0.0057,
+      "step": 4063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0363470495030495e-05,
+      "loss": 0.0055,
+      "step": 4064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.036106976520863e-05,
+      "loss": 0.0058,
+      "step": 4065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035866903538676e-05,
+      "loss": 0.0061,
+      "step": 4066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035626830556489e-05,
+      "loss": 0.0052,
+      "step": 4067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035386757574303e-05,
+      "loss": 0.0045,
+      "step": 4068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035146684592116e-05,
+      "loss": 0.004,
+      "step": 4069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.03490661160993e-05,
+      "loss": 0.0045,
+      "step": 4070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034666538627743e-05,
+      "loss": 0.0051,
+      "step": 4071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.034426465645556e-05,
+      "loss": 0.0039,
+      "step": 4072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03418639266337e-05,
+      "loss": 0.004,
+      "step": 4073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033946319681183e-05,
+      "loss": 0.0045,
+      "step": 4074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0337062466989965e-05,
+      "loss": 0.0053,
+      "step": 4075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03346617371681e-05,
+      "loss": 0.0047,
+      "step": 4076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0332261007346235e-05,
+      "loss": 0.6168,
+      "step": 4077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032986027752437e-05,
+      "loss": 0.6343,
+      "step": 4078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03274595477025e-05,
+      "loss": 0.6091,
+      "step": 4079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032505881788064e-05,
+      "loss": 0.0052,
+      "step": 4080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0322658088058776e-05,
+      "loss": 0.0049,
+      "step": 4081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032025735823691e-05,
+      "loss": 0.0037,
+      "step": 4082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031785662841504e-05,
+      "loss": 0.6801,
+      "step": 4083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031545589859317e-05,
+      "loss": 0.005,
+      "step": 4084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031305516877131e-05,
+      "loss": 0.0048,
+      "step": 4085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031065443894944e-05,
+      "loss": 0.0029,
+      "step": 4086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0308253709127574e-05,
+      "loss": 0.6211,
+      "step": 4087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0305852979305706e-05,
+      "loss": 0.0037,
+      "step": 4088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0303452249483844e-05,
+      "loss": 0.0067,
+      "step": 4089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.030105151966198e-05,
+      "loss": 0.0068,
+      "step": 4090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0298650789840115e-05,
+      "loss": 0.0058,
+      "step": 4091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0296250060018246e-05,
+      "loss": 0.0045,
+      "step": 4092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029384933019638e-05,
+      "loss": 0.0078,
+      "step": 4093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029144860037452e-05,
+      "loss": 0.5873,
+      "step": 4094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028904787055265e-05,
+      "loss": 0.5424,
+      "step": 4095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028664714073078e-05,
+      "loss": 0.006,
+      "step": 4096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028424641090892e-05,
+      "loss": 0.0073,
+      "step": 4097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028184568108706e-05,
+      "loss": 0.0065,
+      "step": 4098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027944495126519e-05,
+      "loss": 0.0111,
+      "step": 4099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027704422144332e-05,
+      "loss": 0.0066,
+      "step": 4100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027464349162145e-05,
+      "loss": 0.0055,
+      "step": 4101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027224276179959e-05,
+      "loss": 0.006,
+      "step": 4102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026984203197772e-05,
+      "loss": 0.0071,
+      "step": 4103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0267441302155855e-05,
+      "loss": 0.004,
+      "step": 4104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026504057233399e-05,
+      "loss": 0.0147,
+      "step": 4105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0262639842512125e-05,
+      "loss": 0.0039,
+      "step": 4106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0260239112690264e-05,
+      "loss": 0.0066,
+      "step": 4107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0257838382868396e-05,
+      "loss": 0.0061,
+      "step": 4108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025543765304653e-05,
+      "loss": 0.0064,
+      "step": 4109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0253036923224666e-05,
+      "loss": 0.006,
+      "step": 4110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02506361934028e-05,
+      "loss": 0.0071,
+      "step": 4111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024823546358093e-05,
+      "loss": 0.0033,
+      "step": 4112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024583473375906e-05,
+      "loss": 0.0073,
+      "step": 4113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02434340039372e-05,
+      "loss": 0.0048,
+      "step": 4114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.024103327411533e-05,
+      "loss": 0.0077,
+      "step": 4115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023863254429347e-05,
+      "loss": 0.0047,
+      "step": 4116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02362318144716e-05,
+      "loss": 0.023,
+      "step": 4117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0233831084649734e-05,
+      "loss": 0.0042,
+      "step": 4118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023143035482787e-05,
+      "loss": 0.005,
+      "step": 4119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0229029625006005e-05,
+      "loss": 1.2441,
+      "step": 4120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0226628895184136e-05,
+      "loss": 0.1493,
+      "step": 4121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022422816536227e-05,
+      "loss": 0.0046,
+      "step": 4122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022182743554041e-05,
+      "loss": 0.0051,
+      "step": 4123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021942670571854e-05,
+      "loss": 0.7958,
+      "step": 4124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021702597589668e-05,
+      "loss": 0.0031,
+      "step": 4125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021462524607481e-05,
+      "loss": 0.003,
+      "step": 4126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021222451625295e-05,
+      "loss": 0.0033,
+      "step": 4127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020982378643108e-05,
+      "loss": 0.0984,
+      "step": 4128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020742305660921e-05,
+      "loss": 0.0036,
+      "step": 4129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020502232678734e-05,
+      "loss": 0.8157,
+      "step": 4130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020262159696548e-05,
+      "loss": 0.0044,
+      "step": 4131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020022086714361e-05,
+      "loss": 0.0032,
+      "step": 4132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0197820137321745e-05,
+      "loss": 0.0039,
+      "step": 4133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0195419407499884e-05,
+      "loss": 0.0034,
+      "step": 4134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0193018677678015e-05,
+      "loss": 0.0037,
+      "step": 4135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0190617947856154e-05,
+      "loss": 0.0037,
+      "step": 4136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0188217218034286e-05,
+      "loss": 0.0028,
+      "step": 4137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018581648821242e-05,
+      "loss": 0.004,
+      "step": 4138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018341575839055e-05,
+      "loss": 0.0033,
+      "step": 4139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018101502856869e-05,
+      "loss": 0.0035,
+      "step": 4140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017861429874682e-05,
+      "loss": 0.0034,
+      "step": 4141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017621356892495e-05,
+      "loss": 0.003,
+      "step": 4142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0173812839103083e-05,
+      "loss": 0.0036,
+      "step": 4143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.017141210928123e-05,
+      "loss": 0.0038,
+      "step": 4144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.016901137945936e-05,
+      "loss": 0.6654,
+      "step": 4145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.016661064963749e-05,
+      "loss": 0.0026,
+      "step": 4146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0164209919815624e-05,
+      "loss": 0.0033,
+      "step": 4147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.016180918999376e-05,
+      "loss": 0.0032,
+      "step": 4148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0159408460171895e-05,
+      "loss": 0.0028,
+      "step": 4149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0157007730350026e-05,
+      "loss": 0.0027,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.015460700052816e-05,
+      "loss": 0.0067,
+      "step": 4151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01522062707063e-05,
+      "loss": 0.7898,
+      "step": 4152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0149805540884435e-05,
+      "loss": 0.0036,
+      "step": 4153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014740481106257e-05,
+      "loss": 0.004,
+      "step": 4154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01450040812407e-05,
+      "loss": 0.6563,
+      "step": 4155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014260335141883e-05,
+      "loss": 0.0038,
+      "step": 4156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.014020262159697e-05,
+      "loss": 0.0056,
+      "step": 4157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01378018917751e-05,
+      "loss": 0.0038,
+      "step": 4158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.013540116195323e-05,
+      "loss": 0.0032,
+      "step": 4159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0133000432131365e-05,
+      "loss": 0.0034,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.01305997023095e-05,
+      "loss": 0.0037,
+      "step": 4161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.012819897248764e-05,
+      "loss": 0.0029,
+      "step": 4162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0125798242665774e-05,
+      "loss": 0.0028,
+      "step": 4163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0123397512843905e-05,
+      "loss": 0.0031,
+      "step": 4164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0120996783022044e-05,
+      "loss": 0.0037,
+      "step": 4165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0118596053200176e-05,
+      "loss": 0.0031,
+      "step": 4166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.011619532337831e-05,
+      "loss": 0.0027,
+      "step": 4167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.011379459355644e-05,
+      "loss": 0.0032,
+      "step": 4168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.011139386373458e-05,
+      "loss": 0.0031,
+      "step": 4169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010899313391271e-05,
+      "loss": 0.6135,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010659240409085e-05,
+      "loss": 0.0036,
+      "step": 4171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010419167426898e-05,
+      "loss": 0.0028,
+      "step": 4172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.010179094444711e-05,
+      "loss": 0.0035,
+      "step": 4173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009939021462525e-05,
+      "loss": 0.0023,
+      "step": 4174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009698948480338e-05,
+      "loss": 0.0052,
+      "step": 4175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0094588754981514e-05,
+      "loss": 0.0035,
+      "step": 4176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.009218802515965e-05,
+      "loss": 0.0027,
+      "step": 4177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0089787295337784e-05,
+      "loss": 0.003,
+      "step": 4178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0087386565515916e-05,
+      "loss": 0.0024,
+      "step": 4179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0084985835694055e-05,
+      "loss": 0.0037,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.008258510587219e-05,
+      "loss": 0.0023,
+      "step": 4181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0080184376050325e-05,
+      "loss": 0.0028,
+      "step": 4182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007778364622846e-05,
+      "loss": 0.0024,
+      "step": 4183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007538291640659e-05,
+      "loss": 0.8173,
+      "step": 4184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007298218658472e-05,
+      "loss": 0.0022,
+      "step": 4185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007058145676286e-05,
+      "loss": 0.0023,
+      "step": 4186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006818072694099e-05,
+      "loss": 0.0028,
+      "step": 4187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006577999711912e-05,
+      "loss": 0.0022,
+      "step": 4188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006337926729726e-05,
+      "loss": 0.0027,
+      "step": 4189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.006097853747539e-05,
+      "loss": 0.6698,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.005857780765353e-05,
+      "loss": 0.0026,
+      "step": 4191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0056177077831664e-05,
+      "loss": 0.0027,
+      "step": 4192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0053776348009795e-05,
+      "loss": 0.0028,
+      "step": 4193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0051375618187934e-05,
+      "loss": 0.0027,
+      "step": 4194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0048974888366066e-05,
+      "loss": 0.0024,
+      "step": 4195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00465741585442e-05,
+      "loss": 0.0029,
+      "step": 4196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004417342872233e-05,
+      "loss": 0.0026,
+      "step": 4197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.004177269890047e-05,
+      "loss": 0.0023,
+      "step": 4198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0039371969078607e-05,
+      "loss": 0.0025,
+      "step": 4199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003697123925674e-05,
+      "loss": 0.0042,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003457050943487e-05,
+      "loss": 0.0027,
+      "step": 4201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0032169779613e-05,
+      "loss": 0.0026,
+      "step": 4202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002976904979114e-05,
+      "loss": 0.0022,
+      "step": 4203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.002736831996927e-05,
+      "loss": 0.0021,
+      "step": 4204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0024967590147404e-05,
+      "loss": 0.0024,
+      "step": 4205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0022566860325536e-05,
+      "loss": 0.0023,
+      "step": 4206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0020166130503674e-05,
+      "loss": 0.7942,
+      "step": 4207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001776540068181e-05,
+      "loss": 0.0024,
+      "step": 4208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0015364670859945e-05,
+      "loss": 0.0021,
+      "step": 4209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001296394103808e-05,
+      "loss": 0.0022,
+      "step": 4210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0010563211216215e-05,
+      "loss": 0.0022,
+      "step": 4211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000816248139435e-05,
+      "loss": 0.0023,
+      "step": 4212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000576175157248e-05,
+      "loss": 0.0023,
+      "step": 4213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000336102175061e-05,
+      "loss": 0.0023,
+      "step": 4214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.000096029192875e-05,
+      "loss": 0.0022,
+      "step": 4215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999855956210688e-05,
+      "loss": 0.002,
+      "step": 4216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999615883228502e-05,
+      "loss": 0.0032,
+      "step": 4217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999375810246315e-05,
+      "loss": 0.0025,
+      "step": 4218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.999135737264128e-05,
+      "loss": 0.0021,
+      "step": 4219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.998895664281942e-05,
+      "loss": 0.0021,
+      "step": 4220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9986555912997554e-05,
+      "loss": 0.0024,
+      "step": 4221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9984155183175685e-05,
+      "loss": 0.7252,
+      "step": 4222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.998175445335382e-05,
+      "loss": 0.0024,
+      "step": 4223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9979353723531956e-05,
+      "loss": 0.5703,
+      "step": 4224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.997695299371009e-05,
+      "loss": 0.0021,
+      "step": 4225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9974552263888226e-05,
+      "loss": 0.0025,
+      "step": 4226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.997215153406636e-05,
+      "loss": 0.0024,
+      "step": 4227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9969750804244496e-05,
+      "loss": 0.0022,
+      "step": 4228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996735007442263e-05,
+      "loss": 0.0027,
+      "step": 4229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996494934460076e-05,
+      "loss": 0.0023,
+      "step": 4230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996254861477889e-05,
+      "loss": 0.7492,
+      "step": 4231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.996014788495703e-05,
+      "loss": 0.0021,
+      "step": 4232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.995774715513516e-05,
+      "loss": 0.0026,
+      "step": 4233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9955346425313294e-05,
+      "loss": 0.5894,
+      "step": 4234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.995294569549143e-05,
+      "loss": 0.0023,
+      "step": 4235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9950544965669564e-05,
+      "loss": 0.0124,
+      "step": 4236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99481442358477e-05,
+      "loss": 0.0022,
+      "step": 4237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9945743506025835e-05,
+      "loss": 0.0023,
+      "step": 4238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994334277620397e-05,
+      "loss": 0.0024,
+      "step": 4239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99409420463821e-05,
+      "loss": 0.0022,
+      "step": 4240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993854131656024e-05,
+      "loss": 0.002,
+      "step": 4241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993614058673837e-05,
+      "loss": 0.002,
+      "step": 4242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99337398569165e-05,
+      "loss": 0.0034,
+      "step": 4243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.993133912709464e-05,
+      "loss": 0.0021,
+      "step": 4244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992893839727278e-05,
+      "loss": 0.0025,
+      "step": 4245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992653766745091e-05,
+      "loss": 0.002,
+      "step": 4246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992413693762904e-05,
+      "loss": 0.0023,
+      "step": 4247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.992173620780717e-05,
+      "loss": 0.0023,
+      "step": 4248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991933547798531e-05,
+      "loss": 0.0017,
+      "step": 4249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9916934748163444e-05,
+      "loss": 0.0016,
+      "step": 4250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9914534018341575e-05,
+      "loss": 0.0019,
+      "step": 4251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991213328851971e-05,
+      "loss": 0.0017,
+      "step": 4252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9909732558697846e-05,
+      "loss": 0.4555,
+      "step": 4253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9907331828875984e-05,
+      "loss": 0.002,
+      "step": 4254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9904931099054116e-05,
+      "loss": 0.002,
+      "step": 4255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990253036923225e-05,
+      "loss": 0.7547,
+      "step": 4256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990012963941038e-05,
+      "loss": 0.8194,
+      "step": 4257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989772890958852e-05,
+      "loss": 0.0121,
+      "step": 4258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989532817976665e-05,
+      "loss": 0.0029,
+      "step": 4259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989292744994478e-05,
+      "loss": 0.0017,
+      "step": 4260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989052672012292e-05,
+      "loss": 0.0016,
+      "step": 4261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988812599030105e-05,
+      "loss": 0.002,
+      "step": 4262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988572526047919e-05,
+      "loss": 0.0034,
+      "step": 4263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988332453065732e-05,
+      "loss": 0.0024,
+      "step": 4264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9880923800835454e-05,
+      "loss": 0.0018,
+      "step": 4265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987852307101359e-05,
+      "loss": 0.0048,
+      "step": 4266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9876122341191725e-05,
+      "loss": 0.7785,
+      "step": 4267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9873721611369857e-05,
+      "loss": 0.7878,
+      "step": 4268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.987132088154799e-05,
+      "loss": 0.4421,
+      "step": 4269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986892015172613e-05,
+      "loss": 0.0022,
+      "step": 4270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986651942190426e-05,
+      "loss": 0.002,
+      "step": 4271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98641186920824e-05,
+      "loss": 0.0023,
+      "step": 4272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.986171796226053e-05,
+      "loss": 0.0022,
+      "step": 4273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985931723243867e-05,
+      "loss": 0.0019,
+      "step": 4274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98569165026168e-05,
+      "loss": 0.0026,
+      "step": 4275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985451577279493e-05,
+      "loss": 0.002,
+      "step": 4276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985211504297306e-05,
+      "loss": 0.0021,
+      "step": 4277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98497143131512e-05,
+      "loss": 0.0022,
+      "step": 4278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9847313583329333e-05,
+      "loss": 0.7604,
+      "step": 4279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9844912853507465e-05,
+      "loss": 0.0026,
+      "step": 4280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9842512123685604e-05,
+      "loss": 0.0017,
+      "step": 4281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9840111393863736e-05,
+      "loss": 0.0017,
+      "step": 4282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9837710664041874e-05,
+      "loss": 0.0024,
+      "step": 4283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9835309934220006e-05,
+      "loss": 0.0019,
+      "step": 4284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.983290920439814e-05,
+      "loss": 0.0111,
+      "step": 4285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.983050847457627e-05,
+      "loss": 0.002,
+      "step": 4286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982810774475441e-05,
+      "loss": 0.0022,
+      "step": 4287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982570701493254e-05,
+      "loss": 0.0024,
+      "step": 4288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982330628511067e-05,
+      "loss": 0.0078,
+      "step": 4289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.982090555528881e-05,
+      "loss": 0.0017,
+      "step": 4290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.981850482546695e-05,
+      "loss": 0.0017,
+      "step": 4291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.981610409564508e-05,
+      "loss": 0.0034,
+      "step": 4292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.981370336582321e-05,
+      "loss": 0.0024,
+      "step": 4293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9811302636001344e-05,
+      "loss": 0.4454,
+      "step": 4294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.980890190617948e-05,
+      "loss": 0.0023,
+      "step": 4295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9806501176357615e-05,
+      "loss": 0.0048,
+      "step": 4296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9804100446535747e-05,
+      "loss": 0.0021,
+      "step": 4297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.980169971671388e-05,
+      "loss": 0.0021,
+      "step": 4298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979929898689202e-05,
+      "loss": 0.0021,
+      "step": 4299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9796898257070156e-05,
+      "loss": 0.0025,
+      "step": 4300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979449752724829e-05,
+      "loss": 0.002,
+      "step": 4301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.979209679742642e-05,
+      "loss": 0.0021,
+      "step": 4302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.978969606760455e-05,
+      "loss": 0.0017,
+      "step": 4303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.978729533778269e-05,
+      "loss": 0.0019,
+      "step": 4304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.978489460796082e-05,
+      "loss": 0.0025,
+      "step": 4305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.978249387813895e-05,
+      "loss": 0.0019,
+      "step": 4306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9780093148317085e-05,
+      "loss": 0.7375,
+      "step": 4307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977769241849523e-05,
+      "loss": 0.0022,
+      "step": 4308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.977529168867336e-05,
+      "loss": 0.002,
+      "step": 4309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9772890958851494e-05,
+      "loss": 0.0018,
+      "step": 4310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9770490229029626e-05,
+      "loss": 0.0021,
+      "step": 4311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9768089499207764e-05,
+      "loss": 0.0021,
+      "step": 4312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9765688769385896e-05,
+      "loss": 0.003,
+      "step": 4313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976328803956403e-05,
+      "loss": 0.0017,
+      "step": 4314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.976088730974216e-05,
+      "loss": 0.002,
+      "step": 4315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97584865799203e-05,
+      "loss": 0.0022,
+      "step": 4316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975608585009844e-05,
+      "loss": 0.0023,
+      "step": 4317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.975368512027657e-05,
+      "loss": 0.7241,
+      "step": 4318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97512843904547e-05,
+      "loss": 0.0027,
+      "step": 4319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974888366063283e-05,
+      "loss": 0.0021,
+      "step": 4320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.974648293081097e-05,
+      "loss": 0.7858,
+      "step": 4321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.97440822009891e-05,
+      "loss": 0.0024,
+      "step": 4322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9741681471167234e-05,
+      "loss": 0.0022,
+      "step": 4323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9739280741345366e-05,
+      "loss": 0.7299,
+      "step": 4324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9736880011523505e-05,
+      "loss": 0.3264,
+      "step": 4325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9734479281701637e-05,
+      "loss": 0.0017,
+      "step": 4326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9732078551879775e-05,
+      "loss": 0.0019,
+      "step": 4327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972967782205791e-05,
+      "loss": 0.0031,
+      "step": 4328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9727277092236045e-05,
+      "loss": 0.0032,
+      "step": 4329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972487636241418e-05,
+      "loss": 0.0073,
+      "step": 4330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972247563259231e-05,
+      "loss": 0.0023,
+      "step": 4331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.972007490277044e-05,
+      "loss": 0.003,
+      "step": 4332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971767417294858e-05,
+      "loss": 0.004,
+      "step": 4333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971527344312671e-05,
+      "loss": 0.0028,
+      "step": 4334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971287271330484e-05,
+      "loss": 0.0024,
+      "step": 4335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.971047198348298e-05,
+      "loss": 0.0027,
+      "step": 4336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9708071253661113e-05,
+      "loss": 0.0022,
+      "step": 4337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.970567052383925e-05,
+      "loss": 0.0029,
+      "step": 4338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9703269794017384e-05,
+      "loss": 0.0032,
+      "step": 4339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9700869064195516e-05,
+      "loss": 0.0027,
+      "step": 4340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969846833437365e-05,
+      "loss": 0.6684,
+      "step": 4341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9696067604551786e-05,
+      "loss": 0.0029,
+      "step": 4342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969366687472992e-05,
+      "loss": 0.0025,
+      "step": 4343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969126614490805e-05,
+      "loss": 0.003,
+      "step": 4344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968886541508619e-05,
+      "loss": 0.0028,
+      "step": 4345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968646468526433e-05,
+      "loss": 0.0018,
+      "step": 4346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968406395544246e-05,
+      "loss": 0.0024,
+      "step": 4347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968166322562059e-05,
+      "loss": 0.6993,
+      "step": 4348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967926249579872e-05,
+      "loss": 0.0022,
+      "step": 4349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.967686176597686e-05,
+      "loss": 0.0036,
+      "step": 4350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.967446103615499e-05,
+      "loss": 0.0032,
+      "step": 4351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9672060306333124e-05,
+      "loss": 0.0023,
+      "step": 4352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9669659576511256e-05,
+      "loss": 0.0027,
+      "step": 4353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9667258846689395e-05,
+      "loss": 0.0027,
+      "step": 4354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.966485811686753e-05,
+      "loss": 0.0025,
+      "step": 4355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9662457387045665e-05,
+      "loss": 0.7192,
+      "step": 4356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96600566572238e-05,
+      "loss": 0.7155,
+      "step": 4357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9657655927401935e-05,
+      "loss": 0.0025,
+      "step": 4358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.965525519758007e-05,
+      "loss": 0.0022,
+      "step": 4359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.96528544677582e-05,
+      "loss": 0.5593,
+      "step": 4360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.965045373793633e-05,
+      "loss": 0.0037,
+      "step": 4361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964805300811447e-05,
+      "loss": 0.0022,
+      "step": 4362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964565227829261e-05,
+      "loss": 0.0028,
+      "step": 4363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964325154847074e-05,
+      "loss": 0.0029,
+      "step": 4364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.964085081864887e-05,
+      "loss": 0.0028,
+      "step": 4365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9638450088827003e-05,
+      "loss": 0.0025,
+      "step": 4366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.963604935900514e-05,
+      "loss": 0.0013,
+      "step": 4367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9633648629183274e-05,
+      "loss": 0.0034,
+      "step": 4368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9631247899361406e-05,
+      "loss": 0.0016,
+      "step": 4369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.962884716953954e-05,
+      "loss": 0.0024,
+      "step": 4370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9626446439717676e-05,
+      "loss": 0.7036,
+      "step": 4371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9624045709895815e-05,
+      "loss": 0.0022,
+      "step": 4372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9621644980073946e-05,
+      "loss": 0.0027,
+      "step": 4373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.961924425025208e-05,
+      "loss": 0.0026,
+      "step": 4374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.961684352043022e-05,
+      "loss": 0.0286,
+      "step": 4375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.961444279060835e-05,
+      "loss": 0.003,
+      "step": 4376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.961204206078648e-05,
+      "loss": 0.5795,
+      "step": 4377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960964133096461e-05,
+      "loss": 0.0033,
+      "step": 4378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960724060114275e-05,
+      "loss": 0.0195,
+      "step": 4379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960483987132088e-05,
+      "loss": 0.0023,
+      "step": 4380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960243914149902e-05,
+      "loss": 0.0036,
+      "step": 4381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.960003841167715e-05,
+      "loss": 0.0048,
+      "step": 4382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9597637681855285e-05,
+      "loss": 0.0024,
+      "step": 4383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959523695203342e-05,
+      "loss": 0.0019,
+      "step": 4384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9592836222211555e-05,
+      "loss": 0.0033,
+      "step": 4385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.959043549238969e-05,
+      "loss": 0.4394,
+      "step": 4386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958803476256782e-05,
+      "loss": 0.5066,
+      "step": 4387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958563403274596e-05,
+      "loss": 0.0034,
+      "step": 4388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958323330292409e-05,
+      "loss": 0.0024,
+      "step": 4389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.958083257310222e-05,
+      "loss": 0.2705,
+      "step": 4390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957843184328036e-05,
+      "loss": 0.0019,
+      "step": 4391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95760311134585e-05,
+      "loss": 0.014,
+      "step": 4392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957363038363663e-05,
+      "loss": 0.0025,
+      "step": 4393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957122965381476e-05,
+      "loss": 0.0035,
+      "step": 4394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956882892399289e-05,
+      "loss": 0.0013,
+      "step": 4395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.956642819417103e-05,
+      "loss": 0.0018,
+      "step": 4396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9564027464349164e-05,
+      "loss": 0.0031,
+      "step": 4397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9561626734527296e-05,
+      "loss": 0.0155,
+      "step": 4398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.955922600470543e-05,
+      "loss": 0.0022,
+      "step": 4399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9556825274883566e-05,
+      "loss": 0.0015,
+      "step": 4400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9554424545061705e-05,
+      "loss": 0.0032,
+      "step": 4401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9552023815239836e-05,
+      "loss": 0.0019,
+      "step": 4402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954962308541797e-05,
+      "loss": 0.0018,
+      "step": 4403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95472223555961e-05,
+      "loss": 0.0022,
+      "step": 4404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954482162577424e-05,
+      "loss": 0.0078,
+      "step": 4405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954242089595237e-05,
+      "loss": 0.0023,
+      "step": 4406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.95400201661305e-05,
+      "loss": 0.0016,
+      "step": 4407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9537619436308634e-05,
+      "loss": 0.0015,
+      "step": 4408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953521870648678e-05,
+      "loss": 0.0017,
+      "step": 4409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953281797666491e-05,
+      "loss": 0.0015,
+      "step": 4410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.953041724684304e-05,
+      "loss": 0.0021,
+      "step": 4411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9528016517021175e-05,
+      "loss": 0.0025,
+      "step": 4412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952561578719931e-05,
+      "loss": 0.8418,
+      "step": 4413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9523215057377445e-05,
+      "loss": 0.0032,
+      "step": 4414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.952081432755558e-05,
+      "loss": 0.0019,
+      "step": 4415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951841359773371e-05,
+      "loss": 0.0015,
+      "step": 4416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.951601286791185e-05,
+      "loss": 0.0018,
+      "step": 4417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9513612138089986e-05,
+      "loss": 0.002,
+      "step": 4418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.951121140826812e-05,
+      "loss": 0.2195,
+      "step": 4419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.950881067844625e-05,
+      "loss": 0.0016,
+      "step": 4420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.950640994862438e-05,
+      "loss": 0.0011,
+      "step": 4421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.950400921880252e-05,
+      "loss": 0.0016,
+      "step": 4422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.950160848898065e-05,
+      "loss": 0.0015,
+      "step": 4423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.949920775915878e-05,
+      "loss": 0.3853,
+      "step": 4424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.949680702933692e-05,
+      "loss": 0.0014,
+      "step": 4425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9494406299515054e-05,
+      "loss": 0.0022,
+      "step": 4426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.949200556969319e-05,
+      "loss": 0.0014,
+      "step": 4427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9489604839871324e-05,
+      "loss": 0.0018,
+      "step": 4428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9487204110049456e-05,
+      "loss": 0.7315,
+      "step": 4429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9484803380227594e-05,
+      "loss": 0.0018,
+      "step": 4430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9482402650405726e-05,
+      "loss": 0.0022,
+      "step": 4431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.948000192058386e-05,
+      "loss": 0.0023,
+      "step": 4432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947760119076199e-05,
+      "loss": 0.0015,
+      "step": 4433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947520046094013e-05,
+      "loss": 0.4311,
+      "step": 4434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.947279973111826e-05,
+      "loss": 0.1998,
+      "step": 4435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94703990012964e-05,
+      "loss": 0.0017,
+      "step": 4436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.946799827147453e-05,
+      "loss": 0.0021,
+      "step": 4437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.946559754165266e-05,
+      "loss": 0.2128,
+      "step": 4438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94631968118308e-05,
+      "loss": 0.0015,
+      "step": 4439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.946079608200893e-05,
+      "loss": 0.0161,
+      "step": 4440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9458395352187065e-05,
+      "loss": 0.0016,
+      "step": 4441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.94559946223652e-05,
+      "loss": 0.2808,
+      "step": 4442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9453593892543335e-05,
+      "loss": 0.0011,
+      "step": 4443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.945119316272147e-05,
+      "loss": 0.0107,
+      "step": 4444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9448792432899605e-05,
+      "loss": 0.0017,
+      "step": 4445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.944639170307774e-05,
+      "loss": 0.0021,
+      "step": 4446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9443990973255876e-05,
+      "loss": 0.0023,
+      "step": 4447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.944159024343401e-05,
+      "loss": 0.0016,
+      "step": 4448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943918951361214e-05,
+      "loss": 0.0012,
+      "step": 4449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943678878379027e-05,
+      "loss": 0.002,
+      "step": 4450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943438805396841e-05,
+      "loss": 0.0018,
+      "step": 4451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.943198732414654e-05,
+      "loss": 0.0117,
+      "step": 4452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942958659432467e-05,
+      "loss": 0.001,
+      "step": 4453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9427185864502805e-05,
+      "loss": 0.0019,
+      "step": 4454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942478513468095e-05,
+      "loss": 0.9802,
+      "step": 4455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.942238440485908e-05,
+      "loss": 0.002,
+      "step": 4456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9419983675037214e-05,
+      "loss": 0.1307,
+      "step": 4457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9417582945215346e-05,
+      "loss": 0.0016,
+      "step": 4458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9415182215393484e-05,
+      "loss": 0.0019,
+      "step": 4459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9412781485571616e-05,
+      "loss": 0.0016,
+      "step": 4460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.941038075574975e-05,
+      "loss": 0.0019,
+      "step": 4461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940798002592788e-05,
+      "loss": 0.0018,
+      "step": 4462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940557929610602e-05,
+      "loss": 0.0019,
+      "step": 4463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940317856628416e-05,
+      "loss": 0.0025,
+      "step": 4464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.940077783646229e-05,
+      "loss": 0.0022,
+      "step": 4465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939837710664042e-05,
+      "loss": 0.0014,
+      "step": 4466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939597637681855e-05,
+      "loss": 0.0182,
+      "step": 4467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939357564699669e-05,
+      "loss": 0.0018,
+      "step": 4468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.939117491717482e-05,
+      "loss": 0.0014,
+      "step": 4469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9388774187352955e-05,
+      "loss": 0.0014,
+      "step": 4470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9386373457531086e-05,
+      "loss": 0.0017,
+      "step": 4471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9383972727709225e-05,
+      "loss": 0.0026,
+      "step": 4472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9381571997887364e-05,
+      "loss": 0.3026,
+      "step": 4473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9379171268065495e-05,
+      "loss": 0.0014,
+      "step": 4474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937677053824363e-05,
+      "loss": 1.4534,
+      "step": 4475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9374369808421766e-05,
+      "loss": 0.0019,
+      "step": 4476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93719690785999e-05,
+      "loss": 0.0021,
+      "step": 4477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936956834877803e-05,
+      "loss": 0.7796,
+      "step": 4478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936716761895616e-05,
+      "loss": 0.0016,
+      "step": 4479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93647668891343e-05,
+      "loss": 0.0027,
+      "step": 4480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.936236615931243e-05,
+      "loss": 0.0019,
+      "step": 4481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935996542949057e-05,
+      "loss": 0.002,
+      "step": 4482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93575646996687e-05,
+      "loss": 0.0021,
+      "step": 4483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9355163969846834e-05,
+      "loss": 0.0023,
+      "step": 4484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935276324002497e-05,
+      "loss": 0.0028,
+      "step": 4485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9350362510203104e-05,
+      "loss": 0.9118,
+      "step": 4486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9347961780381236e-05,
+      "loss": 0.0014,
+      "step": 4487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.934556105055937e-05,
+      "loss": 0.0024,
+      "step": 4488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9343160320737506e-05,
+      "loss": 0.0015,
+      "step": 4489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.934075959091564e-05,
+      "loss": 0.0019,
+      "step": 4490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9338358861093777e-05,
+      "loss": 0.0021,
+      "step": 4491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.933595813127191e-05,
+      "loss": 0.8896,
+      "step": 4492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.933355740145005e-05,
+      "loss": 0.002,
+      "step": 4493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.933115667162818e-05,
+      "loss": 0.0023,
+      "step": 4494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932875594180631e-05,
+      "loss": 0.0028,
+      "step": 4495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932635521198444e-05,
+      "loss": 0.0022,
+      "step": 4496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932395448216258e-05,
+      "loss": 0.0028,
+      "step": 4497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.932155375234071e-05,
+      "loss": 0.0025,
+      "step": 4498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9319153022518845e-05,
+      "loss": 0.0029,
+      "step": 4499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.931675229269698e-05,
+      "loss": 0.6992,
+      "step": 4500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9314351562875115e-05,
+      "loss": 0.0033,
+      "step": 4501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9311950833053254e-05,
+      "loss": 0.003,
+      "step": 4502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9309550103231385e-05,
+      "loss": 0.7653,
+      "step": 4503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930714937340952e-05,
+      "loss": 0.0042,
+      "step": 4504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930474864358765e-05,
+      "loss": 0.0041,
+      "step": 4505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.930234791376579e-05,
+      "loss": 0.004,
+      "step": 4506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929994718394392e-05,
+      "loss": 0.0054,
+      "step": 4507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929754645412205e-05,
+      "loss": 0.0038,
+      "step": 4508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929514572430019e-05,
+      "loss": 0.0047,
+      "step": 4509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929274499447833e-05,
+      "loss": 0.0035,
+      "step": 4510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.929034426465646e-05,
+      "loss": 0.0036,
+      "step": 4511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.928794353483459e-05,
+      "loss": 0.0039,
+      "step": 4512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9285542805012724e-05,
+      "loss": 0.0041,
+      "step": 4513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.928314207519086e-05,
+      "loss": 0.0034,
+      "step": 4514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9280741345368994e-05,
+      "loss": 0.0033,
+      "step": 4515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9278340615547126e-05,
+      "loss": 0.003,
+      "step": 4516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.927593988572526e-05,
+      "loss": 0.0031,
+      "step": 4517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9273539155903396e-05,
+      "loss": 0.0029,
+      "step": 4518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9271138426081535e-05,
+      "loss": 0.0032,
+      "step": 4519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9268737696259667e-05,
+      "loss": 0.0025,
+      "step": 4520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92663369664378e-05,
+      "loss": 0.0022,
+      "step": 4521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926393623661594e-05,
+      "loss": 0.0027,
+      "step": 4522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.926153550679407e-05,
+      "loss": 0.0021,
+      "step": 4523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92591347769722e-05,
+      "loss": 0.0022,
+      "step": 4524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925673404715033e-05,
+      "loss": 0.0021,
+      "step": 4525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.925433331732847e-05,
+      "loss": 0.002,
+      "step": 4526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.92519325875066e-05,
+      "loss": 0.0018,
+      "step": 4527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924953185768474e-05,
+      "loss": 0.0026,
+      "step": 4528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.924713112786287e-05,
+      "loss": 0.0026,
+      "step": 4529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9244730398041005e-05,
+      "loss": 0.0025,
+      "step": 4530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9242329668219143e-05,
+      "loss": 0.8015,
+      "step": 4531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9239928938397275e-05,
+      "loss": 0.7775,
+      "step": 4532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923752820857541e-05,
+      "loss": 0.0026,
+      "step": 4533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923512747875354e-05,
+      "loss": 0.0026,
+      "step": 4534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923272674893168e-05,
+      "loss": 0.0024,
+      "step": 4535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923032601910981e-05,
+      "loss": 0.0022,
+      "step": 4536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922792528928795e-05,
+      "loss": 0.0024,
+      "step": 4537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922552455946608e-05,
+      "loss": 0.0025,
+      "step": 4538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922312382964422e-05,
+      "loss": 0.0026,
+      "step": 4539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922072309982235e-05,
+      "loss": 0.0025,
+      "step": 4540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921832237000048e-05,
+      "loss": 0.0025,
+      "step": 4541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9215921640178614e-05,
+      "loss": 0.0026,
+      "step": 4542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921352091035675e-05,
+      "loss": 0.7745,
+      "step": 4543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9211120180534884e-05,
+      "loss": 0.7853,
+      "step": 4544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9208719450713016e-05,
+      "loss": 0.0026,
+      "step": 4545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9206318720891154e-05,
+      "loss": 0.003,
+      "step": 4546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9203917991069286e-05,
+      "loss": 0.0028,
+      "step": 4547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9201517261247425e-05,
+      "loss": 0.0026,
+      "step": 4548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9199116531425557e-05,
+      "loss": 0.4825,
+      "step": 4549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919671580160369e-05,
+      "loss": 0.7597,
+      "step": 4550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919431507178182e-05,
+      "loss": 0.7574,
+      "step": 4551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919191434195996e-05,
+      "loss": 0.0032,
+      "step": 4552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918951361213809e-05,
+      "loss": 0.0036,
+      "step": 4553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918711288231622e-05,
+      "loss": 0.0031,
+      "step": 4554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.918471215249436e-05,
+      "loss": 0.0036,
+      "step": 4555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91823114226725e-05,
+      "loss": 0.0034,
+      "step": 4556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917991069285063e-05,
+      "loss": 0.0038,
+      "step": 4557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917750996302876e-05,
+      "loss": 0.0037,
+      "step": 4558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9175109233206895e-05,
+      "loss": 0.0038,
+      "step": 4559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9172708503385033e-05,
+      "loss": 0.0036,
+      "step": 4560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9170307773563165e-05,
+      "loss": 0.0045,
+      "step": 4561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.91679070437413e-05,
+      "loss": 0.0035,
+      "step": 4562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.916550631391943e-05,
+      "loss": 0.0046,
+      "step": 4563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.916310558409757e-05,
+      "loss": 0.0031,
+      "step": 4564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9160704854275706e-05,
+      "loss": 0.0033,
+      "step": 4565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.915830412445384e-05,
+      "loss": 0.0166,
+      "step": 4566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.915590339463197e-05,
+      "loss": 0.0037,
+      "step": 4567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.91535026648101e-05,
+      "loss": 0.0034,
+      "step": 4568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.915110193498824e-05,
+      "loss": 0.0036,
+      "step": 4569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.914870120516637e-05,
+      "loss": 0.0034,
+      "step": 4570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9146300475344504e-05,
+      "loss": 0.175,
+      "step": 4571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9143899745522635e-05,
+      "loss": 0.003,
+      "step": 4572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9141499015700774e-05,
+      "loss": 0.0036,
+      "step": 4573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.913909828587891e-05,
+      "loss": 0.0029,
+      "step": 4574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9136697556057044e-05,
+      "loss": 0.5151,
+      "step": 4575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9134296826235176e-05,
+      "loss": 0.003,
+      "step": 4576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9131896096413315e-05,
+      "loss": 0.0024,
+      "step": 4577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9129495366591447e-05,
+      "loss": 0.0031,
+      "step": 4578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.912709463676958e-05,
+      "loss": 0.003,
+      "step": 4579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.912469390694771e-05,
+      "loss": 0.003,
+      "step": 4580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.912229317712585e-05,
+      "loss": 0.0154,
+      "step": 4581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911989244730398e-05,
+      "loss": 0.5708,
+      "step": 4582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911749171748212e-05,
+      "loss": 0.0025,
+      "step": 4583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911509098766025e-05,
+      "loss": 0.0029,
+      "step": 4584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911269025783838e-05,
+      "loss": 0.0024,
+      "step": 4585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.911028952801652e-05,
+      "loss": 1.3885,
+      "step": 4586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910788879819465e-05,
+      "loss": 0.0622,
+      "step": 4587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9105488068372785e-05,
+      "loss": 0.7112,
+      "step": 4588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.910308733855092e-05,
+      "loss": 1.2259,
+      "step": 4589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9100686608729055e-05,
+      "loss": 0.6349,
+      "step": 4590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909828587890719e-05,
+      "loss": 0.0028,
+      "step": 4591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9095885149085326e-05,
+      "loss": 0.0024,
+      "step": 4592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909348441926346e-05,
+      "loss": 0.0024,
+      "step": 4593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9091083689441596e-05,
+      "loss": 0.0027,
+      "step": 4594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908868295961973e-05,
+      "loss": 0.003,
+      "step": 4595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908628222979786e-05,
+      "loss": 0.0051,
+      "step": 4596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908388149997599e-05,
+      "loss": 0.0062,
+      "step": 4597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908148077015413e-05,
+      "loss": 0.0067,
+      "step": 4598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907908004033226e-05,
+      "loss": 0.0081,
+      "step": 4599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9076679310510394e-05,
+      "loss": 0.0053,
+      "step": 4600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.907427858068853e-05,
+      "loss": 0.0032,
+      "step": 4601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9071877850866664e-05,
+      "loss": 0.0056,
+      "step": 4602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90694771210448e-05,
+      "loss": 0.0022,
+      "step": 4603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9067076391222934e-05,
+      "loss": 0.003,
+      "step": 4604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9064675661401066e-05,
+      "loss": 0.0021,
+      "step": 4605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9062274931579205e-05,
+      "loss": 0.0024,
+      "step": 4606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9059874201757336e-05,
+      "loss": 0.003,
+      "step": 4607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905747347193547e-05,
+      "loss": 0.7999,
+      "step": 4608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90550727421136e-05,
+      "loss": 0.3128,
+      "step": 4609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905267201229174e-05,
+      "loss": 0.0024,
+      "step": 4610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.905027128246988e-05,
+      "loss": 0.0024,
+      "step": 4611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904787055264801e-05,
+      "loss": 0.0033,
+      "step": 4612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904546982282614e-05,
+      "loss": 0.3061,
+      "step": 4613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904306909300427e-05,
+      "loss": 0.0025,
+      "step": 4614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904066836318241e-05,
+      "loss": 0.0024,
+      "step": 4615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903826763336054e-05,
+      "loss": 0.0024,
+      "step": 4616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9035866903538675e-05,
+      "loss": 0.0326,
+      "step": 4617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.903346617371681e-05,
+      "loss": 0.0024,
+      "step": 4618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9031065443894945e-05,
+      "loss": 0.0027,
+      "step": 4619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9028664714073084e-05,
+      "loss": 0.0021,
+      "step": 4620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9026263984251216e-05,
+      "loss": 0.0027,
+      "step": 4621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902386325442935e-05,
+      "loss": 0.002,
+      "step": 4622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9021462524607486e-05,
+      "loss": 0.0022,
+      "step": 4623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901906179478562e-05,
+      "loss": 0.0028,
+      "step": 4624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901666106496375e-05,
+      "loss": 0.0261,
+      "step": 4625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901426033514188e-05,
+      "loss": 0.0023,
+      "step": 4626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.901185960532002e-05,
+      "loss": 0.0021,
+      "step": 4627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.900945887549816e-05,
+      "loss": 0.0025,
+      "step": 4628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.900705814567629e-05,
+      "loss": 0.0018,
+      "step": 4629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.900465741585442e-05,
+      "loss": 0.7629,
+      "step": 4630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9002256686032554e-05,
+      "loss": 0.276,
+      "step": 4631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.899985595621069e-05,
+      "loss": 0.0023,
+      "step": 4632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8997455226388824e-05,
+      "loss": 0.0022,
+      "step": 4633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8995054496566956e-05,
+      "loss": 0.0351,
+      "step": 4634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.899265376674509e-05,
+      "loss": 0.0021,
+      "step": 4635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8990253036923226e-05,
+      "loss": 0.002,
+      "step": 4636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.898785230710136e-05,
+      "loss": 0.0024,
+      "step": 4637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89854515772795e-05,
+      "loss": 0.0024,
+      "step": 4638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.898305084745763e-05,
+      "loss": 0.0213,
+      "step": 4639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.898065011763577e-05,
+      "loss": 0.0021,
+      "step": 4640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89782493878139e-05,
+      "loss": 0.0101,
+      "step": 4641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.897584865799203e-05,
+      "loss": 0.0139,
+      "step": 4642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.897344792817016e-05,
+      "loss": 0.766,
+      "step": 4643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89710471983483e-05,
+      "loss": 0.0019,
+      "step": 4644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.896864646852643e-05,
+      "loss": 0.0018,
+      "step": 4645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8966245738704565e-05,
+      "loss": 0.002,
+      "step": 4646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.89638450088827e-05,
+      "loss": 0.0023,
+      "step": 4647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8961444279060835e-05,
+      "loss": 0.0021,
+      "step": 4648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8959043549238974e-05,
+      "loss": 0.0016,
+      "step": 4649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8956642819417106e-05,
+      "loss": 0.0021,
+      "step": 4650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.895424208959524e-05,
+      "loss": 0.0019,
+      "step": 4651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.895184135977337e-05,
+      "loss": 0.002,
+      "step": 4652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894944062995151e-05,
+      "loss": 0.003,
+      "step": 4653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894703990012964e-05,
+      "loss": 0.8127,
+      "step": 4654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894463917030777e-05,
+      "loss": 0.002,
+      "step": 4655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.894223844048591e-05,
+      "loss": 0.0018,
+      "step": 4656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893983771066405e-05,
+      "loss": 0.0018,
+      "step": 4657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893743698084218e-05,
+      "loss": 0.0039,
+      "step": 4658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893503625102031e-05,
+      "loss": 0.0017,
+      "step": 4659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8932635521198444e-05,
+      "loss": 0.0022,
+      "step": 4660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.893023479137658e-05,
+      "loss": 0.0016,
+      "step": 4661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8927834061554714e-05,
+      "loss": 0.0021,
+      "step": 4662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8925433331732846e-05,
+      "loss": 0.0018,
+      "step": 4663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892303260191098e-05,
+      "loss": 0.0042,
+      "step": 4664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8920631872089116e-05,
+      "loss": 0.0018,
+      "step": 4665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8918231142267255e-05,
+      "loss": 0.7442,
+      "step": 4666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891583041244539e-05,
+      "loss": 0.0016,
+      "step": 4667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891342968262352e-05,
+      "loss": 0.0031,
+      "step": 4668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.891102895280165e-05,
+      "loss": 0.0021,
+      "step": 4669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890862822297979e-05,
+      "loss": 0.1622,
+      "step": 4670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890622749315792e-05,
+      "loss": 0.5024,
+      "step": 4671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890382676333605e-05,
+      "loss": 0.0017,
+      "step": 4672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890142603351419e-05,
+      "loss": 0.0021,
+      "step": 4673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889902530369233e-05,
+      "loss": 0.002,
+      "step": 4674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889662457387046e-05,
+      "loss": 0.002,
+      "step": 4675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889422384404859e-05,
+      "loss": 0.002,
+      "step": 4676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8891823114226725e-05,
+      "loss": 0.0022,
+      "step": 4677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8889422384404864e-05,
+      "loss": 0.0018,
+      "step": 4678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8887021654582996e-05,
+      "loss": 0.0021,
+      "step": 4679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888462092476113e-05,
+      "loss": 0.0018,
+      "step": 4680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888222019493926e-05,
+      "loss": 0.5156,
+      "step": 4681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88798194651174e-05,
+      "loss": 0.0103,
+      "step": 4682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8877418735295536e-05,
+      "loss": 0.0019,
+      "step": 4683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887501800547367e-05,
+      "loss": 0.0018,
+      "step": 4684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88726172756518e-05,
+      "loss": 0.0022,
+      "step": 4685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887021654582993e-05,
+      "loss": 0.0019,
+      "step": 4686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886781581600807e-05,
+      "loss": 0.0016,
+      "step": 4687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88654150861862e-05,
+      "loss": 0.0021,
+      "step": 4688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8863014356364334e-05,
+      "loss": 0.7714,
+      "step": 4689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886061362654247e-05,
+      "loss": 0.0018,
+      "step": 4690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8858212896720604e-05,
+      "loss": 0.0018,
+      "step": 4691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885581216689874e-05,
+      "loss": 0.0018,
+      "step": 4692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8853411437076875e-05,
+      "loss": 0.0021,
+      "step": 4693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8851010707255006e-05,
+      "loss": 0.0038,
+      "step": 4694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8848609977433145e-05,
+      "loss": 0.0019,
+      "step": 4695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884620924761128e-05,
+      "loss": 0.002,
+      "step": 4696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884380851778941e-05,
+      "loss": 0.0018,
+      "step": 4697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.884140778796754e-05,
+      "loss": 0.0028,
+      "step": 4698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.883900705814568e-05,
+      "loss": 0.4835,
+      "step": 4699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.883660632832381e-05,
+      "loss": 0.0017,
+      "step": 4700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.883420559850194e-05,
+      "loss": 0.0016,
+      "step": 4701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.883180486868008e-05,
+      "loss": 0.0017,
+      "step": 4702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.882940413885822e-05,
+      "loss": 0.0021,
+      "step": 4703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.882700340903635e-05,
+      "loss": 0.0018,
+      "step": 4704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.882460267921448e-05,
+      "loss": 0.0017,
+      "step": 4705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8822201949392615e-05,
+      "loss": 0.0016,
+      "step": 4706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8819801219570754e-05,
+      "loss": 0.0034,
+      "step": 4707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8817400489748885e-05,
+      "loss": 0.0016,
+      "step": 4708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.881499975992702e-05,
+      "loss": 0.7445,
+      "step": 4709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.881259903010515e-05,
+      "loss": 0.0015,
+      "step": 4710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.881019830028329e-05,
+      "loss": 0.0017,
+      "step": 4711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8807797570461426e-05,
+      "loss": 0.0013,
+      "step": 4712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.880539684063956e-05,
+      "loss": 0.0014,
+      "step": 4713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.880299611081769e-05,
+      "loss": 0.0019,
+      "step": 4714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.880059538099582e-05,
+      "loss": 0.0017,
+      "step": 4715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.879819465117396e-05,
+      "loss": 0.0018,
+      "step": 4716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.879579392135209e-05,
+      "loss": 0.0018,
+      "step": 4717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8793393191530224e-05,
+      "loss": 0.0016,
+      "step": 4718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8790992461708356e-05,
+      "loss": 0.0015,
+      "step": 4719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87885917318865e-05,
+      "loss": 0.0028,
+      "step": 4720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.878619100206463e-05,
+      "loss": 0.0016,
+      "step": 4721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8783790272242765e-05,
+      "loss": 0.0016,
+      "step": 4722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8781389542420896e-05,
+      "loss": 0.0012,
+      "step": 4723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8778988812599035e-05,
+      "loss": 0.0023,
+      "step": 4724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877658808277717e-05,
+      "loss": 0.0017,
+      "step": 4725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87741873529553e-05,
+      "loss": 0.0015,
+      "step": 4726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.877178662313343e-05,
+      "loss": 0.0013,
+      "step": 4727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876938589331157e-05,
+      "loss": 0.0016,
+      "step": 4728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876698516348971e-05,
+      "loss": 0.0014,
+      "step": 4729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876458443366784e-05,
+      "loss": 0.0016,
+      "step": 4730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876218370384597e-05,
+      "loss": 0.0011,
+      "step": 4731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87597829740241e-05,
+      "loss": 1.288,
+      "step": 4732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875738224420224e-05,
+      "loss": 0.0011,
+      "step": 4733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875498151438037e-05,
+      "loss": 0.0012,
+      "step": 4734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8752580784558505e-05,
+      "loss": 0.0015,
+      "step": 4735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875018005473664e-05,
+      "loss": 0.0013,
+      "step": 4736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8747779324914775e-05,
+      "loss": 0.0014,
+      "step": 4737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8745378595092914e-05,
+      "loss": 0.0014,
+      "step": 4738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8742977865271046e-05,
+      "loss": 0.0028,
+      "step": 4739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874057713544918e-05,
+      "loss": 0.0014,
+      "step": 4740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8738176405627316e-05,
+      "loss": 0.0013,
+      "step": 4741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873577567580545e-05,
+      "loss": 0.0011,
+      "step": 4742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873337494598358e-05,
+      "loss": 0.0015,
+      "step": 4743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873097421616171e-05,
+      "loss": 0.0021,
+      "step": 4744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872857348633985e-05,
+      "loss": 0.004,
+      "step": 4745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872617275651798e-05,
+      "loss": 0.0014,
+      "step": 4746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872377202669612e-05,
+      "loss": 0.0014,
+      "step": 4747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872137129687425e-05,
+      "loss": 0.0013,
+      "step": 4748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8718970567052384e-05,
+      "loss": 0.0022,
+      "step": 4749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871656983723052e-05,
+      "loss": 0.0012,
+      "step": 4750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8714169107408655e-05,
+      "loss": 0.5944,
+      "step": 4751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8711768377586786e-05,
+      "loss": 0.0013,
+      "step": 4752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870936764776492e-05,
+      "loss": 0.8598,
+      "step": 4753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870696691794306e-05,
+      "loss": 0.0011,
+      "step": 4754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870456618812119e-05,
+      "loss": 0.0013,
+      "step": 4755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870216545829933e-05,
+      "loss": 0.0012,
+      "step": 4756
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869976472847746e-05,
+      "loss": 0.0014,
+      "step": 4757
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86973639986556e-05,
+      "loss": 0.0014,
+      "step": 4758
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869496326883373e-05,
+      "loss": 0.0014,
+      "step": 4759
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869256253901186e-05,
+      "loss": 0.0012,
+      "step": 4760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869016180918999e-05,
+      "loss": 0.0013,
+      "step": 4761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868776107936813e-05,
+      "loss": 0.7992,
+      "step": 4762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868536034954626e-05,
+      "loss": 0.0014,
+      "step": 4763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8682959619724395e-05,
+      "loss": 0.0013,
+      "step": 4764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868055888990253e-05,
+      "loss": 0.0016,
+      "step": 4765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8678158160080665e-05,
+      "loss": 0.0014,
+      "step": 4766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8675757430258804e-05,
+      "loss": 0.0014,
+      "step": 4767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8673356700436936e-05,
+      "loss": 0.0012,
+      "step": 4768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.867095597061507e-05,
+      "loss": 0.0015,
+      "step": 4769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86685552407932e-05,
+      "loss": 0.0012,
+      "step": 4770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.866615451097134e-05,
+      "loss": 0.0013,
+      "step": 4771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.866375378114947e-05,
+      "loss": 0.0014,
+      "step": 4772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.86613530513276e-05,
+      "loss": 0.0028,
+      "step": 4773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865895232150574e-05,
+      "loss": 0.4544,
+      "step": 4774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865655159168388e-05,
+      "loss": 0.0856,
+      "step": 4775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865415086186201e-05,
+      "loss": 0.0013,
+      "step": 4776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.865175013204014e-05,
+      "loss": 0.0032,
+      "step": 4777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8649349402218274e-05,
+      "loss": 0.0012,
+      "step": 4778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.864694867239641e-05,
+      "loss": 0.0014,
+      "step": 4779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8644547942574545e-05,
+      "loss": 0.0011,
+      "step": 4780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8642147212752676e-05,
+      "loss": 0.0014,
+      "step": 4781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863974648293081e-05,
+      "loss": 0.0012,
+      "step": 4782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863734575310895e-05,
+      "loss": 0.0015,
+      "step": 4783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8634945023287085e-05,
+      "loss": 0.0013,
+      "step": 4784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863254429346522e-05,
+      "loss": 0.0011,
+      "step": 4785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863014356364335e-05,
+      "loss": 0.0012,
+      "step": 4786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.862774283382149e-05,
+      "loss": 0.0013,
+      "step": 4787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.862534210399962e-05,
+      "loss": 0.0013,
+      "step": 4788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.862294137417775e-05,
+      "loss": 0.0013,
+      "step": 4789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.862054064435588e-05,
+      "loss": 0.4461,
+      "step": 4790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861813991453402e-05,
+      "loss": 0.0014,
+      "step": 4791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861573918471215e-05,
+      "loss": 0.0013,
+      "step": 4792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861333845489029e-05,
+      "loss": 0.0044,
+      "step": 4793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8610937725068424e-05,
+      "loss": 0.0012,
+      "step": 4794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8608536995246555e-05,
+      "loss": 0.0012,
+      "step": 4795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8606136265424694e-05,
+      "loss": 0.0012,
+      "step": 4796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8603735535602826e-05,
+      "loss": 0.0012,
+      "step": 4797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.860133480578096e-05,
+      "loss": 0.0012,
+      "step": 4798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859893407595909e-05,
+      "loss": 0.7413,
+      "step": 4799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859653334613723e-05,
+      "loss": 0.0012,
+      "step": 4800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859413261631536e-05,
+      "loss": 0.0011,
+      "step": 4801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85917318864935e-05,
+      "loss": 0.0039,
+      "step": 4802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858933115667163e-05,
+      "loss": 0.0012,
+      "step": 4803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858693042684977e-05,
+      "loss": 0.0013,
+      "step": 4804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85845296970279e-05,
+      "loss": 0.0012,
+      "step": 4805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858212896720603e-05,
+      "loss": 0.0013,
+      "step": 4806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8579728237384164e-05,
+      "loss": 0.0012,
+      "step": 4807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.85773275075623e-05,
+      "loss": 0.0023,
+      "step": 4808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8574926777740434e-05,
+      "loss": 0.0013,
+      "step": 4809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8572526047918566e-05,
+      "loss": 0.0013,
+      "step": 4810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8570125318096705e-05,
+      "loss": 0.0013,
+      "step": 4811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856772458827484e-05,
+      "loss": 0.0016,
+      "step": 4812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8565323858452975e-05,
+      "loss": 0.0014,
+      "step": 4813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856292312863111e-05,
+      "loss": 0.0011,
+      "step": 4814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856052239880924e-05,
+      "loss": 0.0026,
+      "step": 4815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855812166898737e-05,
+      "loss": 0.0047,
+      "step": 4816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855572093916551e-05,
+      "loss": 0.001,
+      "step": 4817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855332020934364e-05,
+      "loss": 0.0013,
+      "step": 4818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855091947952177e-05,
+      "loss": 0.0012,
+      "step": 4819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854851874969991e-05,
+      "loss": 0.0013,
+      "step": 4820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854611801987805e-05,
+      "loss": 0.0014,
+      "step": 4821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854371729005618e-05,
+      "loss": 0.0009,
+      "step": 4822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8541316560234314e-05,
+      "loss": 0.0012,
+      "step": 4823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8538915830412445e-05,
+      "loss": 0.0013,
+      "step": 4824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8536515100590584e-05,
+      "loss": 0.0016,
+      "step": 4825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8534114370768716e-05,
+      "loss": 0.002,
+      "step": 4826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853171364094685e-05,
+      "loss": 0.001,
+      "step": 4827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852931291112498e-05,
+      "loss": 2.611,
+      "step": 4828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852691218130312e-05,
+      "loss": 0.0009,
+      "step": 4829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8524511451481257e-05,
+      "loss": 0.001,
+      "step": 4830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.852211072165939e-05,
+      "loss": 0.001,
+      "step": 4831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851970999183752e-05,
+      "loss": 0.0009,
+      "step": 4832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851730926201565e-05,
+      "loss": 0.0018,
+      "step": 4833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851490853219379e-05,
+      "loss": 0.0011,
+      "step": 4834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851250780237192e-05,
+      "loss": 0.0008,
+      "step": 4835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8510107072550054e-05,
+      "loss": 0.001,
+      "step": 4836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8507706342728186e-05,
+      "loss": 0.0009,
+      "step": 4837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8505305612906324e-05,
+      "loss": 0.001,
+      "step": 4838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.850290488308446e-05,
+      "loss": 0.001,
+      "step": 4839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8500504153262595e-05,
+      "loss": 0.5604,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.849810342344073e-05,
+      "loss": 0.001,
+      "step": 4841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8495702693618865e-05,
+      "loss": 0.0008,
+      "step": 4842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8493301963797e-05,
+      "loss": 0.001,
+      "step": 4843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.849090123397513e-05,
+      "loss": 0.0011,
+      "step": 4844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848850050415326e-05,
+      "loss": 0.0009,
+      "step": 4845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84860997743314e-05,
+      "loss": 0.0011,
+      "step": 4846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848369904450953e-05,
+      "loss": 0.0008,
+      "step": 4847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.848129831468767e-05,
+      "loss": 0.0009,
+      "step": 4848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.84788975848658e-05,
+      "loss": 0.001,
+      "step": 4849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.847649685504393e-05,
+      "loss": 0.8281,
+      "step": 4850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.847409612522207e-05,
+      "loss": 0.0011,
+      "step": 4851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8471695395400204e-05,
+      "loss": 0.0023,
+      "step": 4852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8469294665578335e-05,
+      "loss": 0.6769,
+      "step": 4853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8466893935756474e-05,
+      "loss": 0.0011,
+      "step": 4854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8464493205934606e-05,
+      "loss": 0.601,
+      "step": 4855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.846209247611274e-05,
+      "loss": 0.0011,
+      "step": 4856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8459691746290876e-05,
+      "loss": 0.0009,
+      "step": 4857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.845729101646901e-05,
+      "loss": 0.0011,
+      "step": 4858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8454890286647146e-05,
+      "loss": 0.001,
+      "step": 4859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.845248955682528e-05,
+      "loss": 0.001,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.845008882700341e-05,
+      "loss": 0.001,
+      "step": 4861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844768809718154e-05,
+      "loss": 0.0011,
+      "step": 4862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844528736735968e-05,
+      "loss": 0.0011,
+      "step": 4863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.844288663753781e-05,
+      "loss": 0.8591,
+      "step": 4864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8440485907715944e-05,
+      "loss": 0.7959,
+      "step": 4865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843808517789408e-05,
+      "loss": 0.0012,
+      "step": 4866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8435684448072214e-05,
+      "loss": 0.0016,
+      "step": 4867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.843328371825035e-05,
+      "loss": 0.0009,
+      "step": 4868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8430882988428485e-05,
+      "loss": 0.0014,
+      "step": 4869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842848225860662e-05,
+      "loss": 0.0015,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8426081528784755e-05,
+      "loss": 0.7912,
+      "step": 4871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842368079896289e-05,
+      "loss": 0.6143,
+      "step": 4872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842128006914102e-05,
+      "loss": 0.0013,
+      "step": 4873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841887933931915e-05,
+      "loss": 0.0016,
+      "step": 4874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841647860949729e-05,
+      "loss": 0.0015,
+      "step": 4875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841407787967543e-05,
+      "loss": 0.0016,
+      "step": 4876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.841167714985356e-05,
+      "loss": 0.0021,
+      "step": 4877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840927642003169e-05,
+      "loss": 0.0018,
+      "step": 4878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840687569020982e-05,
+      "loss": 0.0016,
+      "step": 4879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840447496038796e-05,
+      "loss": 0.0018,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8402074230566094e-05,
+      "loss": 0.0019,
+      "step": 4881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8399673500744225e-05,
+      "loss": 0.0017,
+      "step": 4882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839727277092236e-05,
+      "loss": 0.0019,
+      "step": 4883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8394872041100496e-05,
+      "loss": 0.0018,
+      "step": 4884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8392471311278634e-05,
+      "loss": 0.002,
+      "step": 4885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8390070581456766e-05,
+      "loss": 0.0011,
+      "step": 4886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83876698516349e-05,
+      "loss": 0.0017,
+      "step": 4887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8385269121813036e-05,
+      "loss": 0.0032,
+      "step": 4888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838286839199117e-05,
+      "loss": 0.0015,
+      "step": 4889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83804676621693e-05,
+      "loss": 0.0021,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837806693234743e-05,
+      "loss": 0.0037,
+      "step": 4891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837566620252557e-05,
+      "loss": 0.0015,
+      "step": 4892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83732654727037e-05,
+      "loss": 0.0014,
+      "step": 4893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.837086474288184e-05,
+      "loss": 0.0016,
+      "step": 4894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836846401305997e-05,
+      "loss": 0.0017,
+      "step": 4895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8366063283238104e-05,
+      "loss": 0.0017,
+      "step": 4896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836366255341624e-05,
+      "loss": 0.0014,
+      "step": 4897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8361261823594375e-05,
+      "loss": 0.0012,
+      "step": 4898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8358861093772507e-05,
+      "loss": 0.0012,
+      "step": 4899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835646036395064e-05,
+      "loss": 0.0016,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835405963412878e-05,
+      "loss": 0.8816,
+      "step": 4901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835165890430691e-05,
+      "loss": 0.001,
+      "step": 4902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834925817448505e-05,
+      "loss": 0.0011,
+      "step": 4903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834685744466318e-05,
+      "loss": 0.0009,
+      "step": 4904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834445671484132e-05,
+      "loss": 0.0014,
+      "step": 4905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834205598501945e-05,
+      "loss": 0.0013,
+      "step": 4906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833965525519758e-05,
+      "loss": 0.0012,
+      "step": 4907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833725452537571e-05,
+      "loss": 0.0013,
+      "step": 4908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833485379555385e-05,
+      "loss": 0.6305,
+      "step": 4909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8332453065731983e-05,
+      "loss": 0.0012,
+      "step": 4910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8330052335910115e-05,
+      "loss": 0.0014,
+      "step": 4911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8327651606088254e-05,
+      "loss": 0.0012,
+      "step": 4912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8325250876266386e-05,
+      "loss": 0.0012,
+      "step": 4913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8322850146444524e-05,
+      "loss": 0.0009,
+      "step": 4914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8320449416622656e-05,
+      "loss": 0.0016,
+      "step": 4915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831804868680079e-05,
+      "loss": 0.0014,
+      "step": 4916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831564795697892e-05,
+      "loss": 0.0011,
+      "step": 4917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831324722715706e-05,
+      "loss": 0.0027,
+      "step": 4918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831084649733519e-05,
+      "loss": 0.0013,
+      "step": 4919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830844576751332e-05,
+      "loss": 0.5328,
+      "step": 4920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830604503769146e-05,
+      "loss": 0.001,
+      "step": 4921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83036443078696e-05,
+      "loss": 0.0014,
+      "step": 4922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830124357804773e-05,
+      "loss": 0.0012,
+      "step": 4923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829884284822586e-05,
+      "loss": 0.0063,
+      "step": 4924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8296442118403994e-05,
+      "loss": 0.0011,
+      "step": 4925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829404138858213e-05,
+      "loss": 0.0027,
+      "step": 4926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8291640658760265e-05,
+      "loss": 0.001,
+      "step": 4927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8289239928938397e-05,
+      "loss": 0.199,
+      "step": 4928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828683919911653e-05,
+      "loss": 0.0013,
+      "step": 4929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828443846929467e-05,
+      "loss": 0.0049,
+      "step": 4930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8282037739472806e-05,
+      "loss": 0.0016,
+      "step": 4931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827963700965094e-05,
+      "loss": 0.0011,
+      "step": 4932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827723627982907e-05,
+      "loss": 0.0011,
+      "step": 4933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82748355500072e-05,
+      "loss": 0.0013,
+      "step": 4934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827243482018534e-05,
+      "loss": 0.001,
+      "step": 4935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.827003409036347e-05,
+      "loss": 0.0011,
+      "step": 4936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82676333605416e-05,
+      "loss": 0.0009,
+      "step": 4937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826523263071974e-05,
+      "loss": 0.8668,
+      "step": 4938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826283190089788e-05,
+      "loss": 0.0013,
+      "step": 4939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826043117107601e-05,
+      "loss": 0.0012,
+      "step": 4940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8258030441254144e-05,
+      "loss": 0.0013,
+      "step": 4941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8255629711432276e-05,
+      "loss": 0.0011,
+      "step": 4942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8253228981610414e-05,
+      "loss": 0.0009,
+      "step": 4943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8250828251788546e-05,
+      "loss": 0.0014,
+      "step": 4944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824842752196668e-05,
+      "loss": 0.0023,
+      "step": 4945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824602679214481e-05,
+      "loss": 0.0009,
+      "step": 4946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824362606232295e-05,
+      "loss": 0.0009,
+      "step": 4947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824122533250108e-05,
+      "loss": 0.0012,
+      "step": 4948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823882460267922e-05,
+      "loss": 0.0009,
+      "step": 4949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823642387285735e-05,
+      "loss": 0.0011,
+      "step": 4950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823402314303549e-05,
+      "loss": 0.0013,
+      "step": 4951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823162241321362e-05,
+      "loss": 0.0011,
+      "step": 4952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822922168339175e-05,
+      "loss": 0.2554,
+      "step": 4953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8226820953569884e-05,
+      "loss": 0.0012,
+      "step": 4954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822442022374802e-05,
+      "loss": 0.0011,
+      "step": 4955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8222019493926155e-05,
+      "loss": 0.0013,
+      "step": 4956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8219618764104287e-05,
+      "loss": 0.0011,
+      "step": 4957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8217218034282425e-05,
+      "loss": 0.4567,
+      "step": 4958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821481730446056e-05,
+      "loss": 0.0011,
+      "step": 4959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8212416574638695e-05,
+      "loss": 0.001,
+      "step": 4960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821001584481683e-05,
+      "loss": 0.0011,
+      "step": 4961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820761511499496e-05,
+      "loss": 0.0009,
+      "step": 4962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820521438517309e-05,
+      "loss": 0.0014,
+      "step": 4963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820281365535123e-05,
+      "loss": 0.0011,
+      "step": 4964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820041292552936e-05,
+      "loss": 0.0012,
+      "step": 4965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819801219570749e-05,
+      "loss": 0.0012,
+      "step": 4966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819561146588563e-05,
+      "loss": 0.0012,
+      "step": 4967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819321073606377e-05,
+      "loss": 0.0013,
+      "step": 4968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81908100062419e-05,
+      "loss": 0.001,
+      "step": 4969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8188409276420034e-05,
+      "loss": 0.0012,
+      "step": 4970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8186008546598166e-05,
+      "loss": 0.0011,
+      "step": 4971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8183607816776304e-05,
+      "loss": 0.0013,
+      "step": 4972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8181207086954436e-05,
+      "loss": 0.0014,
+      "step": 4973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817880635713257e-05,
+      "loss": 0.001,
+      "step": 4974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81764056273107e-05,
+      "loss": 0.001,
+      "step": 4975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817400489748884e-05,
+      "loss": 0.8133,
+      "step": 4976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817160416766698e-05,
+      "loss": 0.0008,
+      "step": 4977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816920343784511e-05,
+      "loss": 0.001,
+      "step": 4978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816680270802324e-05,
+      "loss": 0.001,
+      "step": 4979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816440197820137e-05,
+      "loss": 0.0011,
+      "step": 4980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816200124837951e-05,
+      "loss": 0.9351,
+      "step": 4981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815960051855764e-05,
+      "loss": 0.0024,
+      "step": 4982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8157199788735774e-05,
+      "loss": 0.0012,
+      "step": 4983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8154799058913906e-05,
+      "loss": 0.0012,
+      "step": 4984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815239832909205e-05,
+      "loss": 0.0011,
+      "step": 4985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814999759927018e-05,
+      "loss": 0.0018,
+      "step": 4986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8147596869448315e-05,
+      "loss": 0.0011,
+      "step": 4987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814519613962645e-05,
+      "loss": 0.0011,
+      "step": 4988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8142795409804585e-05,
+      "loss": 0.0011,
+      "step": 4989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814039467998272e-05,
+      "loss": 0.001,
+      "step": 4990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813799395016085e-05,
+      "loss": 0.9348,
+      "step": 4991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813559322033898e-05,
+      "loss": 0.001,
+      "step": 4992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813319249051712e-05,
+      "loss": 0.0009,
+      "step": 4993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813079176069526e-05,
+      "loss": 0.001,
+      "step": 4994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812839103087339e-05,
+      "loss": 0.0018,
+      "step": 4995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812599030105152e-05,
+      "loss": 0.0016,
+      "step": 4996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8123589571229653e-05,
+      "loss": 0.0013,
+      "step": 4997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.812118884140779e-05,
+      "loss": 0.0072,
+      "step": 4998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8118788111585924e-05,
+      "loss": 0.0013,
+      "step": 4999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8116387381764056e-05,
+      "loss": 0.0011,
+      "step": 5000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811398665194219e-05,
+      "loss": 0.0014,
+      "step": 5001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8111585922120326e-05,
+      "loss": 0.0011,
+      "step": 5002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8109185192298465e-05,
+      "loss": 0.0011,
+      "step": 5003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8106784462476596e-05,
+      "loss": 0.0012,
+      "step": 5004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810438373265473e-05,
+      "loss": 0.0013,
+      "step": 5005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810198300283287e-05,
+      "loss": 0.0017,
+      "step": 5006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8099582273011e-05,
+      "loss": 0.001,
+      "step": 5007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809718154318913e-05,
+      "loss": 0.0011,
+      "step": 5008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809478081336726e-05,
+      "loss": 0.0012,
+      "step": 5009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80923800835454e-05,
+      "loss": 0.0014,
+      "step": 5010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808997935372353e-05,
+      "loss": 0.0009,
+      "step": 5011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8087578623901664e-05,
+      "loss": 0.0012,
+      "step": 5012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80851778940798e-05,
+      "loss": 0.0035,
+      "step": 5013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8082777164257935e-05,
+      "loss": 0.0012,
+      "step": 5014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808037643443607e-05,
+      "loss": 0.0015,
+      "step": 5015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8077975704614205e-05,
+      "loss": 0.001,
+      "step": 5016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807557497479234e-05,
+      "loss": 0.0012,
+      "step": 5017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807317424497047e-05,
+      "loss": 0.5306,
+      "step": 5018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807077351514861e-05,
+      "loss": 0.0012,
+      "step": 5019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806837278532674e-05,
+      "loss": 0.0032,
+      "step": 5020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806597205550487e-05,
+      "loss": 0.0013,
+      "step": 5021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806357132568301e-05,
+      "loss": 0.001,
+      "step": 5022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806117059586115e-05,
+      "loss": 0.0011,
+      "step": 5023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805876986603928e-05,
+      "loss": 0.5177,
+      "step": 5024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805636913621741e-05,
+      "loss": 0.0009,
+      "step": 5025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805396840639554e-05,
+      "loss": 0.0011,
+      "step": 5026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805156767657368e-05,
+      "loss": 0.0014,
+      "step": 5027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8049166946751814e-05,
+      "loss": 0.001,
+      "step": 5028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8046766216929946e-05,
+      "loss": 0.001,
+      "step": 5029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804436548710808e-05,
+      "loss": 0.0012,
+      "step": 5030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8041964757286216e-05,
+      "loss": 0.0011,
+      "step": 5031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8039564027464355e-05,
+      "loss": 0.0009,
+      "step": 5032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8037163297642486e-05,
+      "loss": 0.2399,
+      "step": 5033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803476256782062e-05,
+      "loss": 0.0009,
+      "step": 5034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803236183799876e-05,
+      "loss": 0.0009,
+      "step": 5035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802996110817689e-05,
+      "loss": 0.0011,
+      "step": 5036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802756037835502e-05,
+      "loss": 0.0023,
+      "step": 5037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802515964853315e-05,
+      "loss": 0.8886,
+      "step": 5038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802275891871129e-05,
+      "loss": 0.001,
+      "step": 5039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802035818888943e-05,
+      "loss": 0.001,
+      "step": 5040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801795745906756e-05,
+      "loss": 0.001,
+      "step": 5041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801555672924569e-05,
+      "loss": 0.0084,
+      "step": 5042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8013155999423825e-05,
+      "loss": 0.0011,
+      "step": 5043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801075526960196e-05,
+      "loss": 0.0011,
+      "step": 5044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8008354539780095e-05,
+      "loss": 0.0013,
+      "step": 5045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800595380995823e-05,
+      "loss": 0.001,
+      "step": 5046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.800355308013636e-05,
+      "loss": 0.001,
+      "step": 5047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.80011523503145e-05,
+      "loss": 0.0011,
+      "step": 5048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7998751620492636e-05,
+      "loss": 0.0012,
+      "step": 5049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.799635089067077e-05,
+      "loss": 0.0013,
+      "step": 5050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79939501608489e-05,
+      "loss": 0.0012,
+      "step": 5051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.799154943102704e-05,
+      "loss": 0.0011,
+      "step": 5052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.798914870120517e-05,
+      "loss": 0.0011,
+      "step": 5053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79867479713833e-05,
+      "loss": 0.0009,
+      "step": 5054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.798434724156143e-05,
+      "loss": 0.0012,
+      "step": 5055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.798194651173957e-05,
+      "loss": 0.0011,
+      "step": 5056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7979545781917704e-05,
+      "loss": 0.917,
+      "step": 5057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.797714505209584e-05,
+      "loss": 0.8765,
+      "step": 5058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7974744322273974e-05,
+      "loss": 0.0011,
+      "step": 5059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7972343592452106e-05,
+      "loss": 0.8079,
+      "step": 5060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7969942862630244e-05,
+      "loss": 0.0012,
+      "step": 5061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7967542132808376e-05,
+      "loss": 0.0015,
+      "step": 5062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.796514140298651e-05,
+      "loss": 0.0013,
+      "step": 5063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.796274067316464e-05,
+      "loss": 0.0016,
+      "step": 5064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.796033994334278e-05,
+      "loss": 0.0014,
+      "step": 5065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795793921352091e-05,
+      "loss": 0.0017,
+      "step": 5066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795553848369905e-05,
+      "loss": 0.0015,
+      "step": 5067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795313775387718e-05,
+      "loss": 0.0016,
+      "step": 5068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.795073702405532e-05,
+      "loss": 0.0015,
+      "step": 5069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794833629423345e-05,
+      "loss": 0.0015,
+      "step": 5070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794593556441158e-05,
+      "loss": 0.0015,
+      "step": 5071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7943534834589715e-05,
+      "loss": 0.0014,
+      "step": 5072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.794113410476785e-05,
+      "loss": 0.0012,
+      "step": 5073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7938733374945985e-05,
+      "loss": 0.0022,
+      "step": 5074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793633264512412e-05,
+      "loss": 0.0015,
+      "step": 5075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793393191530225e-05,
+      "loss": 0.0052,
+      "step": 5076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.793153118548039e-05,
+      "loss": 0.0015,
+      "step": 5077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7929130455658526e-05,
+      "loss": 0.0013,
+      "step": 5078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792672972583666e-05,
+      "loss": 0.0015,
+      "step": 5079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792432899601479e-05,
+      "loss": 0.0014,
+      "step": 5080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.792192826619292e-05,
+      "loss": 0.0011,
+      "step": 5081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791952753637106e-05,
+      "loss": 0.0018,
+      "step": 5082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791712680654919e-05,
+      "loss": 0.0017,
+      "step": 5083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.791472607672732e-05,
+      "loss": 0.0013,
+      "step": 5084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7912325346905455e-05,
+      "loss": 0.0013,
+      "step": 5085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79099246170836e-05,
+      "loss": 0.0014,
+      "step": 5086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790752388726173e-05,
+      "loss": 0.0015,
+      "step": 5087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7905123157439864e-05,
+      "loss": 0.0015,
+      "step": 5088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7902722427617996e-05,
+      "loss": 0.0086,
+      "step": 5089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7900321697796134e-05,
+      "loss": 0.0017,
+      "step": 5090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7897920967974266e-05,
+      "loss": 0.2425,
+      "step": 5091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78955202381524e-05,
+      "loss": 0.0016,
+      "step": 5092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789311950833053e-05,
+      "loss": 0.0013,
+      "step": 5093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789071877850867e-05,
+      "loss": 0.0012,
+      "step": 5094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788831804868681e-05,
+      "loss": 0.0017,
+      "step": 5095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788591731886494e-05,
+      "loss": 0.0025,
+      "step": 5096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788351658904307e-05,
+      "loss": 0.0011,
+      "step": 5097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78811158592212e-05,
+      "loss": 0.0014,
+      "step": 5098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787871512939934e-05,
+      "loss": 0.0016,
+      "step": 5099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787631439957747e-05,
+      "loss": 0.8626,
+      "step": 5100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7873913669755605e-05,
+      "loss": 0.0011,
+      "step": 5101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787151293993374e-05,
+      "loss": 0.0015,
+      "step": 5102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7869112210111875e-05,
+      "loss": 0.0016,
+      "step": 5103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7866711480290014e-05,
+      "loss": 0.0024,
+      "step": 5104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7864310750468145e-05,
+      "loss": 0.0013,
+      "step": 5105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786191002064628e-05,
+      "loss": 0.0011,
+      "step": 5106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7859509290824416e-05,
+      "loss": 0.0012,
+      "step": 5107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785710856100255e-05,
+      "loss": 0.9246,
+      "step": 5108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785470783118068e-05,
+      "loss": 0.0011,
+      "step": 5109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785230710135881e-05,
+      "loss": 0.0013,
+      "step": 5110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784990637153695e-05,
+      "loss": 0.0012,
+      "step": 5111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784750564171508e-05,
+      "loss": 0.0014,
+      "step": 5112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784510491189322e-05,
+      "loss": 0.0014,
+      "step": 5113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784270418207135e-05,
+      "loss": 0.0012,
+      "step": 5114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7840303452249484e-05,
+      "loss": 0.0015,
+      "step": 5115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.783790272242762e-05,
+      "loss": 0.0011,
+      "step": 5116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7835501992605754e-05,
+      "loss": 0.8951,
+      "step": 5117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7833101262783886e-05,
+      "loss": 0.0015,
+      "step": 5118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7830700532962024e-05,
+      "loss": 0.0015,
+      "step": 5119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7828299803140156e-05,
+      "loss": 0.0014,
+      "step": 5120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.782589907331829e-05,
+      "loss": 0.0016,
+      "step": 5121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7823498343496427e-05,
+      "loss": 0.0016,
+      "step": 5122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.782109761367456e-05,
+      "loss": 0.0014,
+      "step": 5123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.78186968838527e-05,
+      "loss": 0.0016,
+      "step": 5124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781629615403083e-05,
+      "loss": 0.0018,
+      "step": 5125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781389542420896e-05,
+      "loss": 0.0018,
+      "step": 5126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.781149469438709e-05,
+      "loss": 0.7984,
+      "step": 5127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.780909396456523e-05,
+      "loss": 0.0018,
+      "step": 5128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.780669323474336e-05,
+      "loss": 0.0026,
+      "step": 5129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7804292504921495e-05,
+      "loss": 0.3914,
+      "step": 5130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.780189177509963e-05,
+      "loss": 0.0018,
+      "step": 5131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779949104527777e-05,
+      "loss": 0.0014,
+      "step": 5132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7797090315455904e-05,
+      "loss": 0.0017,
+      "step": 5133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7794689585634035e-05,
+      "loss": 0.0019,
+      "step": 5134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.779228885581217e-05,
+      "loss": 0.0014,
+      "step": 5135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7789888125990306e-05,
+      "loss": 0.0017,
+      "step": 5136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.778748739616844e-05,
+      "loss": 0.0017,
+      "step": 5137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.778508666634657e-05,
+      "loss": 0.0033,
+      "step": 5138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77826859365247e-05,
+      "loss": 0.0017,
+      "step": 5139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.778028520670284e-05,
+      "loss": 0.0017,
+      "step": 5140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777788447688098e-05,
+      "loss": 0.0016,
+      "step": 5141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777548374705911e-05,
+      "loss": 0.0016,
+      "step": 5142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.777308301723724e-05,
+      "loss": 0.0016,
+      "step": 5143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7770682287415374e-05,
+      "loss": 0.1161,
+      "step": 5144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776828155759351e-05,
+      "loss": 0.0018,
+      "step": 5145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7765880827771644e-05,
+      "loss": 0.0012,
+      "step": 5146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7763480097949776e-05,
+      "loss": 0.0015,
+      "step": 5147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776107936812791e-05,
+      "loss": 0.0016,
+      "step": 5148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7758678638306046e-05,
+      "loss": 0.0014,
+      "step": 5149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7756277908484185e-05,
+      "loss": 0.0014,
+      "step": 5150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7753877178662317e-05,
+      "loss": 0.8349,
+      "step": 5151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775147644884045e-05,
+      "loss": 0.0017,
+      "step": 5152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774907571901859e-05,
+      "loss": 0.0015,
+      "step": 5153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774667498919672e-05,
+      "loss": 0.0015,
+      "step": 5154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774427425937485e-05,
+      "loss": 0.7628,
+      "step": 5155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774187352955298e-05,
+      "loss": 0.0017,
+      "step": 5156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773947279973112e-05,
+      "loss": 0.0017,
+      "step": 5157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773707206990925e-05,
+      "loss": 0.002,
+      "step": 5158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773467134008739e-05,
+      "loss": 0.0341,
+      "step": 5159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773227061026552e-05,
+      "loss": 0.002,
+      "step": 5160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7729869880443655e-05,
+      "loss": 0.0021,
+      "step": 5161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7727469150621793e-05,
+      "loss": 0.1142,
+      "step": 5162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7725068420799925e-05,
+      "loss": 0.8353,
+      "step": 5163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772266769097806e-05,
+      "loss": 0.0022,
+      "step": 5164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.772026696115619e-05,
+      "loss": 0.0044,
+      "step": 5165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771786623133433e-05,
+      "loss": 0.002,
+      "step": 5166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771546550151246e-05,
+      "loss": 0.0023,
+      "step": 5167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77130647716906e-05,
+      "loss": 0.0022,
+      "step": 5168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.771066404186873e-05,
+      "loss": 0.561,
+      "step": 5169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770826331204687e-05,
+      "loss": 0.0024,
+      "step": 5170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7705862582225e-05,
+      "loss": 0.002,
+      "step": 5171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770346185240313e-05,
+      "loss": 0.0019,
+      "step": 5172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7701061122581264e-05,
+      "loss": 0.002,
+      "step": 5173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.76986603927594e-05,
+      "loss": 0.81,
+      "step": 5174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7696259662937534e-05,
+      "loss": 0.0053,
+      "step": 5175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7693858933115666e-05,
+      "loss": 0.4023,
+      "step": 5176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7691458203293804e-05,
+      "loss": 0.0025,
+      "step": 5177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7689057473471936e-05,
+      "loss": 0.0385,
+      "step": 5178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7686656743650075e-05,
+      "loss": 0.0194,
+      "step": 5179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7684256013828207e-05,
+      "loss": 0.0021,
+      "step": 5180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768185528400634e-05,
+      "loss": 0.0021,
+      "step": 5181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767945455418447e-05,
+      "loss": 0.0024,
+      "step": 5182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767705382436261e-05,
+      "loss": 0.0026,
+      "step": 5183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.767465309454074e-05,
+      "loss": 0.0024,
+      "step": 5184
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.767225236471887e-05,
+      "loss": 0.0019,
+      "step": 5185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766985163489701e-05,
+      "loss": 0.0022,
+      "step": 5186
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766745090507515e-05,
+      "loss": 0.0022,
+      "step": 5187
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766505017525328e-05,
+      "loss": 0.0017,
+      "step": 5188
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.766264944543141e-05,
+      "loss": 0.0022,
+      "step": 5189
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7660248715609545e-05,
+      "loss": 0.0024,
+      "step": 5190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7657847985787683e-05,
+      "loss": 0.0017,
+      "step": 5191
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7655447255965815e-05,
+      "loss": 0.0021,
+      "step": 5192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.765304652614395e-05,
+      "loss": 0.0019,
+      "step": 5193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.765064579632208e-05,
+      "loss": 0.002,
+      "step": 5194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.764824506650022e-05,
+      "loss": 1.0348,
+      "step": 5195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7645844336678356e-05,
+      "loss": 0.0022,
+      "step": 5196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.764344360685649e-05,
+      "loss": 0.002,
+      "step": 5197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.764104287703462e-05,
+      "loss": 0.0017,
+      "step": 5198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.763864214721276e-05,
+      "loss": 0.002,
+      "step": 5199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.763624141739089e-05,
+      "loss": 0.0017,
+      "step": 5200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.763384068756902e-05,
+      "loss": 0.4727,
+      "step": 5201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7631439957747154e-05,
+      "loss": 0.0015,
+      "step": 5202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.762903922792529e-05,
+      "loss": 0.002,
+      "step": 5203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7626638498103424e-05,
+      "loss": 0.0019,
+      "step": 5204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.762423776828156e-05,
+      "loss": 0.0017,
+      "step": 5205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7621837038459694e-05,
+      "loss": 0.0018,
+      "step": 5206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7619436308637826e-05,
+      "loss": 0.0016,
+      "step": 5207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7617035578815965e-05,
+      "loss": 0.0018,
+      "step": 5208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7614634848994097e-05,
+      "loss": 0.0016,
+      "step": 5209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.761223411917223e-05,
+      "loss": 0.0066,
+      "step": 5210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760983338935036e-05,
+      "loss": 0.0016,
+      "step": 5211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.76074326595285e-05,
+      "loss": 0.7778,
+      "step": 5212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760503192970663e-05,
+      "loss": 0.7589,
+      "step": 5213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.760263119988477e-05,
+      "loss": 0.0018,
+      "step": 5214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.76002304700629e-05,
+      "loss": 0.0019,
+      "step": 5215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759782974024104e-05,
+      "loss": 0.2595,
+      "step": 5216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.759542901041917e-05,
+      "loss": 0.6151,
+      "step": 5217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75930282805973e-05,
+      "loss": 0.0024,
+      "step": 5218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7590627550775435e-05,
+      "loss": 0.0019,
+      "step": 5219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7588226820953573e-05,
+      "loss": 0.0018,
+      "step": 5220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7585826091131705e-05,
+      "loss": 0.0022,
+      "step": 5221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.758342536130984e-05,
+      "loss": 0.8394,
+      "step": 5222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7581024631487976e-05,
+      "loss": 0.0024,
+      "step": 5223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757862390166611e-05,
+      "loss": 0.002,
+      "step": 5224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7576223171844246e-05,
+      "loss": 0.0022,
+      "step": 5225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757382244202238e-05,
+      "loss": 0.0027,
+      "step": 5226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.757142171220051e-05,
+      "loss": 0.0021,
+      "step": 5227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756902098237864e-05,
+      "loss": 0.0021,
+      "step": 5228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756662025255678e-05,
+      "loss": 0.0016,
+      "step": 5229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.756421952273491e-05,
+      "loss": 0.0022,
+      "step": 5230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7561818792913044e-05,
+      "loss": 0.002,
+      "step": 5231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755941806309118e-05,
+      "loss": 0.0024,
+      "step": 5232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755701733326932e-05,
+      "loss": 0.0017,
+      "step": 5233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755461660344745e-05,
+      "loss": 0.0022,
+      "step": 5234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7552215873625584e-05,
+      "loss": 0.0019,
+      "step": 5235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7549815143803716e-05,
+      "loss": 0.8226,
+      "step": 5236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7547414413981855e-05,
+      "loss": 0.8841,
+      "step": 5237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7545013684159986e-05,
+      "loss": 0.0019,
+      "step": 5238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754261295433812e-05,
+      "loss": 0.0022,
+      "step": 5239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754021222451625e-05,
+      "loss": 0.0022,
+      "step": 5240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753781149469439e-05,
+      "loss": 0.002,
+      "step": 5241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753541076487253e-05,
+      "loss": 0.0022,
+      "step": 5242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753301003505066e-05,
+      "loss": 0.002,
+      "step": 5243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753060930522879e-05,
+      "loss": 0.7756,
+      "step": 5244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752820857540692e-05,
+      "loss": 0.0026,
+      "step": 5245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752580784558506e-05,
+      "loss": 0.0062,
+      "step": 5246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752340711576319e-05,
+      "loss": 0.0025,
+      "step": 5247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7521006385941325e-05,
+      "loss": 0.0021,
+      "step": 5248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751860565611946e-05,
+      "loss": 0.0023,
+      "step": 5249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75162049262976e-05,
+      "loss": 0.0037,
+      "step": 5250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7513804196475734e-05,
+      "loss": 0.7587,
+      "step": 5251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7511403466653866e-05,
+      "loss": 0.002,
+      "step": 5252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7509002736832e-05,
+      "loss": 0.0025,
+      "step": 5253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7506602007010136e-05,
+      "loss": 0.0023,
+      "step": 5254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.750420127718827e-05,
+      "loss": 0.0029,
+      "step": 5255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.75018005473664e-05,
+      "loss": 0.0023,
+      "step": 5256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.749939981754453e-05,
+      "loss": 0.0022,
+      "step": 5257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.749699908772267e-05,
+      "loss": 0.0026,
+      "step": 5258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74945983579008e-05,
+      "loss": 0.0024,
+      "step": 5259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.749219762807894e-05,
+      "loss": 0.0022,
+      "step": 5260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.748979689825707e-05,
+      "loss": 0.0023,
+      "step": 5261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7487396168435204e-05,
+      "loss": 0.0021,
+      "step": 5262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.748499543861334e-05,
+      "loss": 0.002,
+      "step": 5263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7482594708791474e-05,
+      "loss": 0.0598,
+      "step": 5264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7480193978969606e-05,
+      "loss": 0.0023,
+      "step": 5265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.747779324914774e-05,
+      "loss": 0.0022,
+      "step": 5266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7475392519325876e-05,
+      "loss": 0.0022,
+      "step": 5267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.747299178950401e-05,
+      "loss": 0.0021,
+      "step": 5268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.747059105968215e-05,
+      "loss": 0.0022,
+      "step": 5269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.746819032986028e-05,
+      "loss": 0.0017,
+      "step": 5270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.746578960003842e-05,
+      "loss": 0.0023,
+      "step": 5271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.746338887021655e-05,
+      "loss": 0.0025,
+      "step": 5272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.746098814039468e-05,
+      "loss": 0.0022,
+      "step": 5273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.745858741057281e-05,
+      "loss": 0.1645,
+      "step": 5274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.745618668075095e-05,
+      "loss": 0.0025,
+      "step": 5275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.745378595092908e-05,
+      "loss": 0.0019,
+      "step": 5276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7451385221107215e-05,
+      "loss": 0.0017,
+      "step": 5277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.744898449128535e-05,
+      "loss": 0.0018,
+      "step": 5278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7446583761463485e-05,
+      "loss": 0.0018,
+      "step": 5279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7444183031641624e-05,
+      "loss": 0.0015,
+      "step": 5280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7441782301819756e-05,
+      "loss": 0.0018,
+      "step": 5281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743938157199789e-05,
+      "loss": 0.0059,
+      "step": 5282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7436980842176026e-05,
+      "loss": 0.0016,
+      "step": 5283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743458011235416e-05,
+      "loss": 0.002,
+      "step": 5284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743217938253229e-05,
+      "loss": 0.0016,
+      "step": 5285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742977865271042e-05,
+      "loss": 0.0039,
+      "step": 5286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742737792288856e-05,
+      "loss": 0.0014,
+      "step": 5287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.74249771930667e-05,
+      "loss": 0.0017,
+      "step": 5288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742257646324483e-05,
+      "loss": 0.0016,
+      "step": 5289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742017573342296e-05,
+      "loss": 0.0015,
+      "step": 5290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7417775003601094e-05,
+      "loss": 0.8212,
+      "step": 5291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741537427377923e-05,
+      "loss": 0.0017,
+      "step": 5292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7412973543957364e-05,
+      "loss": 0.0016,
+      "step": 5293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7410572814135496e-05,
+      "loss": 0.7926,
+      "step": 5294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740817208431363e-05,
+      "loss": 0.0014,
+      "step": 5295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740577135449177e-05,
+      "loss": 0.0018,
+      "step": 5296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7403370624669905e-05,
+      "loss": 0.0016,
+      "step": 5297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740096989484804e-05,
+      "loss": 0.8091,
+      "step": 5298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739856916502617e-05,
+      "loss": 0.0018,
+      "step": 5299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739616843520431e-05,
+      "loss": 0.7487,
+      "step": 5300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739376770538244e-05,
+      "loss": 0.0019,
+      "step": 5301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739136697556057e-05,
+      "loss": 0.0017,
+      "step": 5302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73889662457387e-05,
+      "loss": 0.0017,
+      "step": 5303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738656551591684e-05,
+      "loss": 0.0019,
+      "step": 5304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738416478609498e-05,
+      "loss": 0.7403,
+      "step": 5305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.738176405627311e-05,
+      "loss": 0.0024,
+      "step": 5306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737936332645124e-05,
+      "loss": 0.0028,
+      "step": 5307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7376962596629375e-05,
+      "loss": 0.2463,
+      "step": 5308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7374561866807514e-05,
+      "loss": 0.0031,
+      "step": 5309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7372161136985646e-05,
+      "loss": 0.0021,
+      "step": 5310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736976040716378e-05,
+      "loss": 0.0029,
+      "step": 5311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736735967734191e-05,
+      "loss": 0.0041,
+      "step": 5312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736495894752005e-05,
+      "loss": 0.0024,
+      "step": 5313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7362558217698186e-05,
+      "loss": 0.004,
+      "step": 5314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736015748787632e-05,
+      "loss": 0.0035,
+      "step": 5315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735775675805445e-05,
+      "loss": 0.0031,
+      "step": 5316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735535602823259e-05,
+      "loss": 0.0023,
+      "step": 5317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735295529841072e-05,
+      "loss": 0.0032,
+      "step": 5318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735055456858885e-05,
+      "loss": 0.0031,
+      "step": 5319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7348153838766984e-05,
+      "loss": 0.0022,
+      "step": 5320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734575310894512e-05,
+      "loss": 0.0022,
+      "step": 5321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7343352379123254e-05,
+      "loss": 0.0028,
+      "step": 5322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7340951649301386e-05,
+      "loss": 0.0027,
+      "step": 5323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7338550919479525e-05,
+      "loss": 0.0026,
+      "step": 5324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7336150189657656e-05,
+      "loss": 0.0024,
+      "step": 5325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7333749459835795e-05,
+      "loss": 0.0024,
+      "step": 5326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.733134873001393e-05,
+      "loss": 0.0027,
+      "step": 5327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732894800019206e-05,
+      "loss": 0.0035,
+      "step": 5328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732654727037019e-05,
+      "loss": 0.002,
+      "step": 5329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732414654054833e-05,
+      "loss": 0.0021,
+      "step": 5330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.732174581072646e-05,
+      "loss": 0.0021,
+      "step": 5331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731934508090459e-05,
+      "loss": 0.002,
+      "step": 5332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731694435108273e-05,
+      "loss": 0.0023,
+      "step": 5333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.731454362126087e-05,
+      "loss": 0.0014,
+      "step": 5334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7312142891439e-05,
+      "loss": 0.0023,
+      "step": 5335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.730974216161713e-05,
+      "loss": 1.5853,
+      "step": 5336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7307341431795265e-05,
+      "loss": 0.0023,
+      "step": 5337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7304940701973404e-05,
+      "loss": 0.0026,
+      "step": 5338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7302539972151535e-05,
+      "loss": 0.0016,
+      "step": 5339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.730013924232967e-05,
+      "loss": 0.0021,
+      "step": 5340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72977385125078e-05,
+      "loss": 0.0023,
+      "step": 5341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.729533778268594e-05,
+      "loss": 0.0024,
+      "step": 5342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7292937052864076e-05,
+      "loss": 0.0028,
+      "step": 5343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.729053632304221e-05,
+      "loss": 0.7432,
+      "step": 5344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728813559322034e-05,
+      "loss": 0.0102,
+      "step": 5345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728573486339847e-05,
+      "loss": 0.0022,
+      "step": 5346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728333413357661e-05,
+      "loss": 0.5578,
+      "step": 5347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.728093340375474e-05,
+      "loss": 0.0025,
+      "step": 5348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7278532673932874e-05,
+      "loss": 0.0022,
+      "step": 5349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727613194411101e-05,
+      "loss": 0.0031,
+      "step": 5350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727373121428915e-05,
+      "loss": 0.0028,
+      "step": 5351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.727133048446728e-05,
+      "loss": 0.002,
+      "step": 5352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7268929754645415e-05,
+      "loss": 0.0024,
+      "step": 5353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7266529024823546e-05,
+      "loss": 0.0023,
+      "step": 5354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7264128295001685e-05,
+      "loss": 0.0023,
+      "step": 5355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726172756517982e-05,
+      "loss": 0.0024,
+      "step": 5356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725932683535795e-05,
+      "loss": 0.0021,
+      "step": 5357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725692610553608e-05,
+      "loss": 0.003,
+      "step": 5358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725452537571422e-05,
+      "loss": 0.0024,
+      "step": 5359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725212464589236e-05,
+      "loss": 0.0027,
+      "step": 5360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724972391607049e-05,
+      "loss": 0.0016,
+      "step": 5361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724732318624862e-05,
+      "loss": 0.0022,
+      "step": 5362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724492245642675e-05,
+      "loss": 0.7165,
+      "step": 5363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724252172660489e-05,
+      "loss": 0.0037,
+      "step": 5364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724012099678302e-05,
+      "loss": 0.0023,
+      "step": 5365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7237720266961155e-05,
+      "loss": 0.0021,
+      "step": 5366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7235319537139294e-05,
+      "loss": 0.7682,
+      "step": 5367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7232918807317425e-05,
+      "loss": 0.0029,
+      "step": 5368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7230518077495564e-05,
+      "loss": 0.0025,
+      "step": 5369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7228117347673696e-05,
+      "loss": 0.7284,
+      "step": 5370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722571661785183e-05,
+      "loss": 0.0023,
+      "step": 5371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7223315888029966e-05,
+      "loss": 0.0018,
+      "step": 5372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72209151582081e-05,
+      "loss": 0.003,
+      "step": 5373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721851442838623e-05,
+      "loss": 0.0019,
+      "step": 5374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721611369856436e-05,
+      "loss": 0.0021,
+      "step": 5375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72137129687425e-05,
+      "loss": 0.0024,
+      "step": 5376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721131223892063e-05,
+      "loss": 0.4128,
+      "step": 5377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720891150909877e-05,
+      "loss": 0.0019,
+      "step": 5378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72065107792769e-05,
+      "loss": 0.0027,
+      "step": 5379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720411004945504e-05,
+      "loss": 0.0021,
+      "step": 5380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.720170931963317e-05,
+      "loss": 0.0021,
+      "step": 5381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7199308589811305e-05,
+      "loss": 0.0019,
+      "step": 5382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7196907859989436e-05,
+      "loss": 0.0023,
+      "step": 5383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7194507130167575e-05,
+      "loss": 0.0023,
+      "step": 5384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719210640034571e-05,
+      "loss": 0.0022,
+      "step": 5385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718970567052384e-05,
+      "loss": 0.0025,
+      "step": 5386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718730494070197e-05,
+      "loss": 0.0015,
+      "step": 5387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718490421088011e-05,
+      "loss": 0.0015,
+      "step": 5388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718250348105825e-05,
+      "loss": 0.1867,
+      "step": 5389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718010275123638e-05,
+      "loss": 0.0019,
+      "step": 5390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717770202141451e-05,
+      "loss": 0.002,
+      "step": 5391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717530129159264e-05,
+      "loss": 0.0024,
+      "step": 5392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717290056177078e-05,
+      "loss": 0.0021,
+      "step": 5393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.717049983194891e-05,
+      "loss": 0.0018,
+      "step": 5394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7168099102127045e-05,
+      "loss": 0.002,
+      "step": 5395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.716569837230518e-05,
+      "loss": 0.0022,
+      "step": 5396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.716329764248332e-05,
+      "loss": 0.7256,
+      "step": 5397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7160896912661454e-05,
+      "loss": 0.0015,
+      "step": 5398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7158496182839586e-05,
+      "loss": 0.0016,
+      "step": 5399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.715609545301772e-05,
+      "loss": 0.7349,
+      "step": 5400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7153694723195856e-05,
+      "loss": 0.002,
+      "step": 5401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.715129399337399e-05,
+      "loss": 0.002,
+      "step": 5402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714889326355212e-05,
+      "loss": 0.0017,
+      "step": 5403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714649253373025e-05,
+      "loss": 0.002,
+      "step": 5404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714409180390839e-05,
+      "loss": 0.0021,
+      "step": 5405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.714169107408653e-05,
+      "loss": 0.0016,
+      "step": 5406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.713929034426466e-05,
+      "loss": 0.0016,
+      "step": 5407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.713688961444279e-05,
+      "loss": 0.0019,
+      "step": 5408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7134488884620924e-05,
+      "loss": 0.0015,
+      "step": 5409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.713208815479906e-05,
+      "loss": 0.3441,
+      "step": 5410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7129687424977195e-05,
+      "loss": 0.0014,
+      "step": 5411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7127286695155326e-05,
+      "loss": 0.0021,
+      "step": 5412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.712488596533346e-05,
+      "loss": 0.003,
+      "step": 5413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71224852355116e-05,
+      "loss": 0.4357,
+      "step": 5414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7120084505689735e-05,
+      "loss": 0.7114,
+      "step": 5415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.711768377586787e-05,
+      "loss": 0.0021,
+      "step": 5416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7115283046046e-05,
+      "loss": 0.0024,
+      "step": 5417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.711288231622414e-05,
+      "loss": 0.0022,
+      "step": 5418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.711048158640227e-05,
+      "loss": 0.002,
+      "step": 5419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71080808565804e-05,
+      "loss": 0.0022,
+      "step": 5420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710568012675853e-05,
+      "loss": 0.0451,
+      "step": 5421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.710327939693667e-05,
+      "loss": 0.0021,
+      "step": 5422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.71008786671148e-05,
+      "loss": 0.1715,
+      "step": 5423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.709847793729294e-05,
+      "loss": 0.0021,
+      "step": 5424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7096077207471074e-05,
+      "loss": 0.0024,
+      "step": 5425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7093676477649205e-05,
+      "loss": 0.0187,
+      "step": 5426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7091275747827344e-05,
+      "loss": 0.0028,
+      "step": 5427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7088875018005476e-05,
+      "loss": 0.0011,
+      "step": 5428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708647428818361e-05,
+      "loss": 0.0022,
+      "step": 5429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708407355836174e-05,
+      "loss": 0.0019,
+      "step": 5430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708167282853988e-05,
+      "loss": 0.0243,
+      "step": 5431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707927209871801e-05,
+      "loss": 0.7017,
+      "step": 5432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707687136889615e-05,
+      "loss": 0.0023,
+      "step": 5433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707447063907428e-05,
+      "loss": 0.0022,
+      "step": 5434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.707206990925242e-05,
+      "loss": 0.0019,
+      "step": 5435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706966917943055e-05,
+      "loss": 0.003,
+      "step": 5436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706726844960868e-05,
+      "loss": 0.0023,
+      "step": 5437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7064867719786814e-05,
+      "loss": 0.0028,
+      "step": 5438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.706246698996495e-05,
+      "loss": 0.0018,
+      "step": 5439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7060066260143084e-05,
+      "loss": 0.0027,
+      "step": 5440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7057665530321216e-05,
+      "loss": 0.0022,
+      "step": 5441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7055264800499355e-05,
+      "loss": 0.0022,
+      "step": 5442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.705286407067749e-05,
+      "loss": 0.003,
+      "step": 5443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7050463340855625e-05,
+      "loss": 0.2294,
+      "step": 5444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704806261103376e-05,
+      "loss": 0.002,
+      "step": 5445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704566188121189e-05,
+      "loss": 0.0096,
+      "step": 5446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704326115139003e-05,
+      "loss": 0.0025,
+      "step": 5447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704086042156816e-05,
+      "loss": 0.0018,
+      "step": 5448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703845969174629e-05,
+      "loss": 0.0022,
+      "step": 5449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703605896192442e-05,
+      "loss": 0.0012,
+      "step": 5450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.703365823210256e-05,
+      "loss": 0.0023,
+      "step": 5451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70312575022807e-05,
+      "loss": 0.0047,
+      "step": 5452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702885677245883e-05,
+      "loss": 0.0078,
+      "step": 5453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7026456042636964e-05,
+      "loss": 0.0016,
+      "step": 5454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7024055312815095e-05,
+      "loss": 0.002,
+      "step": 5455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7021654582993234e-05,
+      "loss": 0.0013,
+      "step": 5456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7019253853171366e-05,
+      "loss": 0.0024,
+      "step": 5457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70168531233495e-05,
+      "loss": 0.0017,
+      "step": 5458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701445239352763e-05,
+      "loss": 0.4776,
+      "step": 5459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.701205166370577e-05,
+      "loss": 0.7102,
+      "step": 5460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7009650933883907e-05,
+      "loss": 0.0012,
+      "step": 5461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700725020406204e-05,
+      "loss": 0.0018,
+      "step": 5462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.700484947424017e-05,
+      "loss": 0.0022,
+      "step": 5463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.700244874441831e-05,
+      "loss": 0.0019,
+      "step": 5464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.700004801459644e-05,
+      "loss": 0.0032,
+      "step": 5465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.699764728477457e-05,
+      "loss": 0.0018,
+      "step": 5466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6995246554952704e-05,
+      "loss": 0.0022,
+      "step": 5467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.699284582513084e-05,
+      "loss": 0.813,
+      "step": 5468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6990445095308974e-05,
+      "loss": 0.0015,
+      "step": 5469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.698804436548711e-05,
+      "loss": 0.0018,
+      "step": 5470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6985643635665245e-05,
+      "loss": 0.8523,
+      "step": 5471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.698324290584338e-05,
+      "loss": 0.3986,
+      "step": 5472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6980842176021515e-05,
+      "loss": 0.0019,
+      "step": 5473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697844144619965e-05,
+      "loss": 0.0017,
+      "step": 5474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697604071637778e-05,
+      "loss": 0.0012,
+      "step": 5475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697363998655591e-05,
+      "loss": 0.002,
+      "step": 5476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.697123925673405e-05,
+      "loss": 0.0019,
+      "step": 5477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.696883852691218e-05,
+      "loss": 0.0021,
+      "step": 5478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.696643779709032e-05,
+      "loss": 0.0021,
+      "step": 5479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.696403706726845e-05,
+      "loss": 0.2194,
+      "step": 5480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.696163633744659e-05,
+      "loss": 0.0023,
+      "step": 5481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.695923560762472e-05,
+      "loss": 1.5738,
+      "step": 5482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6956834877802854e-05,
+      "loss": 1.6697,
+      "step": 5483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6954434147980985e-05,
+      "loss": 0.8144,
+      "step": 5484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6952033418159124e-05,
+      "loss": 0.0021,
+      "step": 5485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6949632688337256e-05,
+      "loss": 0.0026,
+      "step": 5486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.694723195851539e-05,
+      "loss": 0.3633,
+      "step": 5487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6944831228693526e-05,
+      "loss": 0.0027,
+      "step": 5488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.694243049887166e-05,
+      "loss": 0.6917,
+      "step": 5489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6940029769049796e-05,
+      "loss": 0.0038,
+      "step": 5490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693762903922793e-05,
+      "loss": 0.5994,
+      "step": 5491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693522830940606e-05,
+      "loss": 0.0091,
+      "step": 5492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693282757958419e-05,
+      "loss": 0.0397,
+      "step": 5493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.693042684976233e-05,
+      "loss": 0.011,
+      "step": 5494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692802611994046e-05,
+      "loss": 0.0261,
+      "step": 5495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6925625390118594e-05,
+      "loss": 0.0145,
+      "step": 5496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692322466029673e-05,
+      "loss": 0.0162,
+      "step": 5497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692082393047487e-05,
+      "loss": 0.0152,
+      "step": 5498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6918423200653e-05,
+      "loss": 0.0046,
+      "step": 5499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6916022470831135e-05,
+      "loss": 0.005,
+      "step": 5500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.691362174100927e-05,
+      "loss": 0.0053,
+      "step": 5501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6911221011187405e-05,
+      "loss": 0.0038,
+      "step": 5502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690882028136554e-05,
+      "loss": 0.0046,
+      "step": 5503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690641955154367e-05,
+      "loss": 0.0045,
+      "step": 5504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.69040188217218e-05,
+      "loss": 0.0032,
+      "step": 5505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690161809189994e-05,
+      "loss": 0.0028,
+      "step": 5506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689921736207808e-05,
+      "loss": 0.003,
+      "step": 5507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689681663225621e-05,
+      "loss": 0.0029,
+      "step": 5508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689441590243434e-05,
+      "loss": 0.0025,
+      "step": 5509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689201517261247e-05,
+      "loss": 0.7463,
+      "step": 5510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688961444279061e-05,
+      "loss": 0.0031,
+      "step": 5511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6887213712968744e-05,
+      "loss": 0.0028,
+      "step": 5512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6884812983146875e-05,
+      "loss": 0.684,
+      "step": 5513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688241225332501e-05,
+      "loss": 0.0023,
+      "step": 5514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6880011523503146e-05,
+      "loss": 0.0027,
+      "step": 5515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6877610793681284e-05,
+      "loss": 0.0026,
+      "step": 5516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6875210063859416e-05,
+      "loss": 0.0023,
+      "step": 5517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687280933403755e-05,
+      "loss": 0.0027,
+      "step": 5518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6870408604215686e-05,
+      "loss": 0.0024,
+      "step": 5519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686800787439382e-05,
+      "loss": 0.003,
+      "step": 5520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686560714457195e-05,
+      "loss": 0.0026,
+      "step": 5521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686320641475008e-05,
+      "loss": 0.0026,
+      "step": 5522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686080568492822e-05,
+      "loss": 0.0024,
+      "step": 5523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685840495510635e-05,
+      "loss": 0.0023,
+      "step": 5524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685600422528449e-05,
+      "loss": 0.0023,
+      "step": 5525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.685360349546262e-05,
+      "loss": 0.0022,
+      "step": 5526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6851202765640754e-05,
+      "loss": 0.0024,
+      "step": 5527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684880203581889e-05,
+      "loss": 0.0024,
+      "step": 5528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6846401305997025e-05,
+      "loss": 0.0022,
+      "step": 5529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6844000576175157e-05,
+      "loss": 0.002,
+      "step": 5530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6841599846353295e-05,
+      "loss": 0.002,
+      "step": 5531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.683919911653143e-05,
+      "loss": 0.0022,
+      "step": 5532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.683679838670956e-05,
+      "loss": 0.0021,
+      "step": 5533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68343976568877e-05,
+      "loss": 0.7693,
+      "step": 5534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.683199692706583e-05,
+      "loss": 0.0019,
+      "step": 5535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.682959619724397e-05,
+      "loss": 0.0022,
+      "step": 5536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68271954674221e-05,
+      "loss": 0.0019,
+      "step": 5537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.682479473760023e-05,
+      "loss": 0.0026,
+      "step": 5538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.682239400777836e-05,
+      "loss": 0.0084,
+      "step": 5539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.68199932779565e-05,
+      "loss": 0.002,
+      "step": 5540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6817592548134633e-05,
+      "loss": 0.002,
+      "step": 5541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6815191818312765e-05,
+      "loss": 0.0022,
+      "step": 5542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6812791088490904e-05,
+      "loss": 0.0019,
+      "step": 5543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.681039035866904e-05,
+      "loss": 0.0021,
+      "step": 5544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6807989628847174e-05,
+      "loss": 0.002,
+      "step": 5545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6805588899025306e-05,
+      "loss": 0.002,
+      "step": 5546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.680318816920344e-05,
+      "loss": 0.0018,
+      "step": 5547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6800787439381576e-05,
+      "loss": 0.0017,
+      "step": 5548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.679838670955971e-05,
+      "loss": 0.0017,
+      "step": 5549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.679598597973784e-05,
+      "loss": 0.0018,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.679358524991597e-05,
+      "loss": 0.002,
+      "step": 5551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.679118452009411e-05,
+      "loss": 0.0018,
+      "step": 5552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678878379027225e-05,
+      "loss": 0.0162,
+      "step": 5553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678638306045038e-05,
+      "loss": 0.0021,
+      "step": 5554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.678398233062851e-05,
+      "loss": 0.0019,
+      "step": 5555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6781581600806644e-05,
+      "loss": 0.002,
+      "step": 5556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677918087098478e-05,
+      "loss": 0.0018,
+      "step": 5557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6776780141162915e-05,
+      "loss": 0.0018,
+      "step": 5558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6774379411341047e-05,
+      "loss": 0.0018,
+      "step": 5559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.677197868151918e-05,
+      "loss": 0.0125,
+      "step": 5560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6769577951697324e-05,
+      "loss": 0.0087,
+      "step": 5561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6767177221875456e-05,
+      "loss": 0.0015,
+      "step": 5562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676477649205359e-05,
+      "loss": 0.0015,
+      "step": 5563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676237576223172e-05,
+      "loss": 0.0015,
+      "step": 5564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675997503240986e-05,
+      "loss": 0.8467,
+      "step": 5565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675757430258799e-05,
+      "loss": 0.0015,
+      "step": 5566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675517357276612e-05,
+      "loss": 0.0017,
+      "step": 5567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675277284294425e-05,
+      "loss": 0.8061,
+      "step": 5568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675037211312239e-05,
+      "loss": 0.0012,
+      "step": 5569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747971383300523e-05,
+      "loss": 0.0018,
+      "step": 5570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674557065347866e-05,
+      "loss": 0.2985,
+      "step": 5571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6743169923656794e-05,
+      "loss": 0.002,
+      "step": 5572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6740769193834926e-05,
+      "loss": 0.0017,
+      "step": 5573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6738368464013064e-05,
+      "loss": 0.0017,
+      "step": 5574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6735967734191196e-05,
+      "loss": 0.0015,
+      "step": 5575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673356700436933e-05,
+      "loss": 0.0015,
+      "step": 5576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.673116627454746e-05,
+      "loss": 0.0018,
+      "step": 5577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67287655447256e-05,
+      "loss": 0.0019,
+      "step": 5578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672636481490373e-05,
+      "loss": 0.0017,
+      "step": 5579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672396408508187e-05,
+      "loss": 0.0018,
+      "step": 5580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672156335526e-05,
+      "loss": 0.0018,
+      "step": 5581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671916262543814e-05,
+      "loss": 0.0019,
+      "step": 5582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671676189561627e-05,
+      "loss": 0.0015,
+      "step": 5583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.67143611657944e-05,
+      "loss": 0.0016,
+      "step": 5584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6711960435972534e-05,
+      "loss": 0.0223,
+      "step": 5585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670955970615067e-05,
+      "loss": 0.0017,
+      "step": 5586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6707158976328805e-05,
+      "loss": 0.0016,
+      "step": 5587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6704758246506937e-05,
+      "loss": 0.0017,
+      "step": 5588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6702357516685075e-05,
+      "loss": 0.0019,
+      "step": 5589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669995678686321e-05,
+      "loss": 0.0018,
+      "step": 5590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6697556057041345e-05,
+      "loss": 0.0016,
+      "step": 5591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669515532721948e-05,
+      "loss": 0.0014,
+      "step": 5592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669275459739761e-05,
+      "loss": 0.8886,
+      "step": 5593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669035386757574e-05,
+      "loss": 0.0018,
+      "step": 5594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668795313775388e-05,
+      "loss": 0.0024,
+      "step": 5595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668555240793201e-05,
+      "loss": 0.0016,
+      "step": 5596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668315167811014e-05,
+      "loss": 0.0015,
+      "step": 5597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668075094828828e-05,
+      "loss": 0.0013,
+      "step": 5598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667835021846642e-05,
+      "loss": 0.0017,
+      "step": 5599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667594948864455e-05,
+      "loss": 0.2587,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6673548758822684e-05,
+      "loss": 0.0013,
+      "step": 5601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6671148029000816e-05,
+      "loss": 0.0014,
+      "step": 5602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6668747299178954e-05,
+      "loss": 0.0019,
+      "step": 5603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6666346569357086e-05,
+      "loss": 0.0015,
+      "step": 5604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.666394583953522e-05,
+      "loss": 0.0015,
+      "step": 5605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.666154510971335e-05,
+      "loss": 0.0017,
+      "step": 5606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.665914437989149e-05,
+      "loss": 0.0015,
+      "step": 5607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.665674365006963e-05,
+      "loss": 0.6371,
+      "step": 5608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.665434292024776e-05,
+      "loss": 0.0015,
+      "step": 5609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.665194219042589e-05,
+      "loss": 0.0017,
+      "step": 5610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664954146060402e-05,
+      "loss": 0.0406,
+      "step": 5611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664714073078216e-05,
+      "loss": 0.0017,
+      "step": 5612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.664474000096029e-05,
+      "loss": 0.0014,
+      "step": 5613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6642339271138424e-05,
+      "loss": 0.0014,
+      "step": 5614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663993854131656e-05,
+      "loss": 0.0017,
+      "step": 5615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66375378114947e-05,
+      "loss": 0.0016,
+      "step": 5616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663513708167283e-05,
+      "loss": 0.0578,
+      "step": 5617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6632736351850965e-05,
+      "loss": 0.0018,
+      "step": 5618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66303356220291e-05,
+      "loss": 0.0014,
+      "step": 5619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6627934892207235e-05,
+      "loss": 0.869,
+      "step": 5620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662553416238537e-05,
+      "loss": 0.0015,
+      "step": 5621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.66231334325635e-05,
+      "loss": 0.0012,
+      "step": 5622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.662073270274163e-05,
+      "loss": 0.0017,
+      "step": 5623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661833197291977e-05,
+      "loss": 0.0017,
+      "step": 5624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661593124309791e-05,
+      "loss": 0.3453,
+      "step": 5625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661353051327604e-05,
+      "loss": 0.0015,
+      "step": 5626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661112978345417e-05,
+      "loss": 0.0016,
+      "step": 5627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660872905363231e-05,
+      "loss": 0.8573,
+      "step": 5628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.660632832381044e-05,
+      "loss": 0.8574,
+      "step": 5629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6603927593988574e-05,
+      "loss": 0.8279,
+      "step": 5630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6601526864166706e-05,
+      "loss": 0.8106,
+      "step": 5631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6599126134344844e-05,
+      "loss": 0.0014,
+      "step": 5632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6596725404522976e-05,
+      "loss": 0.0085,
+      "step": 5633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659432467470111e-05,
+      "loss": 0.0022,
+      "step": 5634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6591923944879246e-05,
+      "loss": 0.4626,
+      "step": 5635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658952321505738e-05,
+      "loss": 0.13,
+      "step": 5636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658712248523552e-05,
+      "loss": 0.0026,
+      "step": 5637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658472175541365e-05,
+      "loss": 0.0023,
+      "step": 5638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658232102559178e-05,
+      "loss": 0.0039,
+      "step": 5639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657992029576991e-05,
+      "loss": 0.0026,
+      "step": 5640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657751956594805e-05,
+      "loss": 0.0039,
+      "step": 5641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657511883612618e-05,
+      "loss": 0.0039,
+      "step": 5642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6572718106304314e-05,
+      "loss": 0.3233,
+      "step": 5643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657031737648245e-05,
+      "loss": 0.0025,
+      "step": 5644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656791664666059e-05,
+      "loss": 0.0026,
+      "step": 5645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656551591683872e-05,
+      "loss": 0.003,
+      "step": 5646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6563115187016855e-05,
+      "loss": 0.0031,
+      "step": 5647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656071445719499e-05,
+      "loss": 0.0028,
+      "step": 5648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6558313727373125e-05,
+      "loss": 0.0032,
+      "step": 5649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655591299755126e-05,
+      "loss": 0.0025,
+      "step": 5650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655351226772939e-05,
+      "loss": 0.0027,
+      "step": 5651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655111153790752e-05,
+      "loss": 0.0036,
+      "step": 5652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654871080808566e-05,
+      "loss": 0.0024,
+      "step": 5653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65463100782638e-05,
+      "loss": 0.003,
+      "step": 5654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654390934844193e-05,
+      "loss": 0.0036,
+      "step": 5655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654150861862006e-05,
+      "loss": 0.0029,
+      "step": 5656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653910788879819e-05,
+      "loss": 0.8239,
+      "step": 5657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653670715897633e-05,
+      "loss": 0.0025,
+      "step": 5658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6534306429154464e-05,
+      "loss": 0.0036,
+      "step": 5659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6531905699332596e-05,
+      "loss": 0.0031,
+      "step": 5660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652950496951073e-05,
+      "loss": 0.0025,
+      "step": 5661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652710423968887e-05,
+      "loss": 0.0031,
+      "step": 5662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6524703509867005e-05,
+      "loss": 0.6892,
+      "step": 5663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6522302780045136e-05,
+      "loss": 0.003,
+      "step": 5664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651990205022327e-05,
+      "loss": 0.0028,
+      "step": 5665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651750132040141e-05,
+      "loss": 0.0034,
+      "step": 5666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651510059057954e-05,
+      "loss": 0.0037,
+      "step": 5667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.651269986075767e-05,
+      "loss": 1.5827,
+      "step": 5668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65102991309358e-05,
+      "loss": 0.9242,
+      "step": 5669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650789840111394e-05,
+      "loss": 0.0022,
+      "step": 5670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650549767129208e-05,
+      "loss": 0.0032,
+      "step": 5671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.650309694147021e-05,
+      "loss": 0.0023,
+      "step": 5672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.650069621164834e-05,
+      "loss": 0.3282,
+      "step": 5673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6498295481826475e-05,
+      "loss": 0.0026,
+      "step": 5674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649589475200461e-05,
+      "loss": 0.003,
+      "step": 5675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6493494022182745e-05,
+      "loss": 0.0032,
+      "step": 5676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649109329236088e-05,
+      "loss": 0.0141,
+      "step": 5677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648869256253901e-05,
+      "loss": 0.003,
+      "step": 5678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648629183271715e-05,
+      "loss": 0.0033,
+      "step": 5679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6483891102895286e-05,
+      "loss": 0.0031,
+      "step": 5680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648149037307342e-05,
+      "loss": 0.0033,
+      "step": 5681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647908964325155e-05,
+      "loss": 0.0031,
+      "step": 5682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647668891342969e-05,
+      "loss": 0.003,
+      "step": 5683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647428818360782e-05,
+      "loss": 0.0026,
+      "step": 5684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647188745378595e-05,
+      "loss": 0.0028,
+      "step": 5685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646948672396408e-05,
+      "loss": 0.0032,
+      "step": 5686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646708599414222e-05,
+      "loss": 0.0417,
+      "step": 5687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6464685264320354e-05,
+      "loss": 0.0025,
+      "step": 5688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646228453449849e-05,
+      "loss": 0.0189,
+      "step": 5689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6459883804676624e-05,
+      "loss": 0.0025,
+      "step": 5690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6457483074854756e-05,
+      "loss": 0.0115,
+      "step": 5691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6455082345032894e-05,
+      "loss": 0.002,
+      "step": 5692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6452681615211026e-05,
+      "loss": 0.3415,
+      "step": 5693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.645028088538916e-05,
+      "loss": 0.0095,
+      "step": 5694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64478801555673e-05,
+      "loss": 0.0126,
+      "step": 5695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644547942574543e-05,
+      "loss": 0.0094,
+      "step": 5696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644307869592356e-05,
+      "loss": 0.0016,
+      "step": 5697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644067796610169e-05,
+      "loss": 0.0046,
+      "step": 5698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643827723627983e-05,
+      "loss": 0.0018,
+      "step": 5699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643587650645797e-05,
+      "loss": 0.0103,
+      "step": 5700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64334757766361e-05,
+      "loss": 0.0019,
+      "step": 5701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643107504681423e-05,
+      "loss": 0.0019,
+      "step": 5702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6428674316992365e-05,
+      "loss": 0.0017,
+      "step": 5703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64262735871705e-05,
+      "loss": 0.0018,
+      "step": 5704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6423872857348635e-05,
+      "loss": 0.002,
+      "step": 5705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642147212752677e-05,
+      "loss": 0.0018,
+      "step": 5706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64190713977049e-05,
+      "loss": 0.0018,
+      "step": 5707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641667066788304e-05,
+      "loss": 0.0017,
+      "step": 5708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6414269938061176e-05,
+      "loss": 0.002,
+      "step": 5709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641186920823931e-05,
+      "loss": 0.0076,
+      "step": 5710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640946847841744e-05,
+      "loss": 0.0018,
+      "step": 5711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640706774859558e-05,
+      "loss": 0.0016,
+      "step": 5712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640466701877371e-05,
+      "loss": 0.0049,
+      "step": 5713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640226628895184e-05,
+      "loss": 0.0018,
+      "step": 5714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639986555912997e-05,
+      "loss": 0.0015,
+      "step": 5715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639746482930811e-05,
+      "loss": 0.0013,
+      "step": 5716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639506409948625e-05,
+      "loss": 0.0014,
+      "step": 5717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639266336966438e-05,
+      "loss": 0.0015,
+      "step": 5718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6390262639842514e-05,
+      "loss": 0.0013,
+      "step": 5719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6387861910020646e-05,
+      "loss": 0.0017,
+      "step": 5720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6385461180198784e-05,
+      "loss": 0.0018,
+      "step": 5721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6383060450376916e-05,
+      "loss": 0.0012,
+      "step": 5722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638065972055505e-05,
+      "loss": 0.7887,
+      "step": 5723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637825899073318e-05,
+      "loss": 0.0015,
+      "step": 5724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637585826091132e-05,
+      "loss": 0.0016,
+      "step": 5725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637345753108946e-05,
+      "loss": 0.0018,
+      "step": 5726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637105680126759e-05,
+      "loss": 0.0014,
+      "step": 5727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636865607144572e-05,
+      "loss": 0.0011,
+      "step": 5728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636625534162386e-05,
+      "loss": 0.0013,
+      "step": 5729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636385461180199e-05,
+      "loss": 0.8458,
+      "step": 5730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636145388198012e-05,
+      "loss": 0.0015,
+      "step": 5731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6359053152158255e-05,
+      "loss": 0.0014,
+      "step": 5732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635665242233639e-05,
+      "loss": 0.0015,
+      "step": 5733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6354251692514525e-05,
+      "loss": 0.0017,
+      "step": 5734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6351850962692664e-05,
+      "loss": 0.0015,
+      "step": 5735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6349450232870795e-05,
+      "loss": 0.0013,
+      "step": 5736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634704950304893e-05,
+      "loss": 0.0014,
+      "step": 5737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6344648773227066e-05,
+      "loss": 0.0015,
+      "step": 5738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63422480434052e-05,
+      "loss": 0.0014,
+      "step": 5739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633984731358333e-05,
+      "loss": 0.0015,
+      "step": 5740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633744658376146e-05,
+      "loss": 0.8751,
+      "step": 5741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63350458539396e-05,
+      "loss": 0.6139,
+      "step": 5742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633264512411773e-05,
+      "loss": 0.002,
+      "step": 5743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.633024439429587e-05,
+      "loss": 0.0017,
+      "step": 5744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6327843664474e-05,
+      "loss": 0.0015,
+      "step": 5745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632544293465214e-05,
+      "loss": 0.0015,
+      "step": 5746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632304220483027e-05,
+      "loss": 0.0014,
+      "step": 5747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6320641475008404e-05,
+      "loss": 0.0016,
+      "step": 5748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6318240745186536e-05,
+      "loss": 0.0016,
+      "step": 5749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6315840015364674e-05,
+      "loss": 0.0019,
+      "step": 5750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6313439285542806e-05,
+      "loss": 0.0016,
+      "step": 5751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.631103855572094e-05,
+      "loss": 0.0014,
+      "step": 5752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6308637825899077e-05,
+      "loss": 0.0062,
+      "step": 5753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630623709607721e-05,
+      "loss": 0.0017,
+      "step": 5754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630383636625535e-05,
+      "loss": 0.0016,
+      "step": 5755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630143563643348e-05,
+      "loss": 0.0018,
+      "step": 5756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629903490661161e-05,
+      "loss": 0.0019,
+      "step": 5757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629663417678974e-05,
+      "loss": 0.0019,
+      "step": 5758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629423344696788e-05,
+      "loss": 0.0016,
+      "step": 5759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.629183271714601e-05,
+      "loss": 0.0016,
+      "step": 5760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6289431987324145e-05,
+      "loss": 0.0018,
+      "step": 5761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6287031257502276e-05,
+      "loss": 0.0016,
+      "step": 5762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.628463052768042e-05,
+      "loss": 0.0017,
+      "step": 5763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6282229797858554e-05,
+      "loss": 0.0016,
+      "step": 5764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6279829068036685e-05,
+      "loss": 0.0017,
+      "step": 5765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627742833821482e-05,
+      "loss": 0.0015,
+      "step": 5766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6275027608392956e-05,
+      "loss": 0.0015,
+      "step": 5767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627262687857109e-05,
+      "loss": 0.0018,
+      "step": 5768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627022614874922e-05,
+      "loss": 0.8634,
+      "step": 5769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626782541892735e-05,
+      "loss": 0.0018,
+      "step": 5770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626542468910549e-05,
+      "loss": 0.0015,
+      "step": 5771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626302395928363e-05,
+      "loss": 0.8018,
+      "step": 5772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626062322946176e-05,
+      "loss": 0.0014,
+      "step": 5773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625822249963989e-05,
+      "loss": 0.0015,
+      "step": 5774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6255821769818024e-05,
+      "loss": 0.0016,
+      "step": 5775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625342103999616e-05,
+      "loss": 0.0016,
+      "step": 5776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6251020310174294e-05,
+      "loss": 0.0018,
+      "step": 5777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6248619580352426e-05,
+      "loss": 0.0018,
+      "step": 5778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6246218850530564e-05,
+      "loss": 0.0018,
+      "step": 5779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6243818120708696e-05,
+      "loss": 0.5777,
+      "step": 5780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6241417390886835e-05,
+      "loss": 0.0064,
+      "step": 5781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6239016661064967e-05,
+      "loss": 0.0016,
+      "step": 5782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62366159312431e-05,
+      "loss": 0.0016,
+      "step": 5783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623421520142124e-05,
+      "loss": 0.0018,
+      "step": 5784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623181447159937e-05,
+      "loss": 0.0018,
+      "step": 5785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62294137417775e-05,
+      "loss": 0.0019,
+      "step": 5786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622701301195563e-05,
+      "loss": 0.0017,
+      "step": 5787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622461228213377e-05,
+      "loss": 0.0014,
+      "step": 5788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62222115523119e-05,
+      "loss": 0.002,
+      "step": 5789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621981082249004e-05,
+      "loss": 0.0018,
+      "step": 5790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621741009266817e-05,
+      "loss": 0.0017,
+      "step": 5791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6215009362846305e-05,
+      "loss": 0.0017,
+      "step": 5792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6212608633024443e-05,
+      "loss": 0.0017,
+      "step": 5793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6210207903202575e-05,
+      "loss": 0.0015,
+      "step": 5794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620780717338071e-05,
+      "loss": 0.0016,
+      "step": 5795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6205406443558846e-05,
+      "loss": 0.0017,
+      "step": 5796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620300571373698e-05,
+      "loss": 0.0017,
+      "step": 5797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620060498391511e-05,
+      "loss": 0.0014,
+      "step": 5798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619820425409325e-05,
+      "loss": 0.0016,
+      "step": 5799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619580352427138e-05,
+      "loss": 0.0017,
+      "step": 5800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619340279444952e-05,
+      "loss": 0.0018,
+      "step": 5801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619100206462765e-05,
+      "loss": 0.0015,
+      "step": 5802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618860133480578e-05,
+      "loss": 0.0013,
+      "step": 5803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6186200604983914e-05,
+      "loss": 0.0017,
+      "step": 5804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618379987516205e-05,
+      "loss": 0.0016,
+      "step": 5805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6181399145340184e-05,
+      "loss": 0.0015,
+      "step": 5806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6178998415518316e-05,
+      "loss": 0.0064,
+      "step": 5807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6176597685696454e-05,
+      "loss": 0.0014,
+      "step": 5808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617419695587459e-05,
+      "loss": 0.0017,
+      "step": 5809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6171796226052725e-05,
+      "loss": 0.0016,
+      "step": 5810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6169395496230857e-05,
+      "loss": 0.0018,
+      "step": 5811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.616699476640899e-05,
+      "loss": 0.0013,
+      "step": 5812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.616459403658713e-05,
+      "loss": 0.0017,
+      "step": 5813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.616219330676526e-05,
+      "loss": 0.0014,
+      "step": 5814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615979257694339e-05,
+      "loss": 0.0015,
+      "step": 5815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615739184712152e-05,
+      "loss": 0.0014,
+      "step": 5816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615499111729966e-05,
+      "loss": 0.0012,
+      "step": 5817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61525903874778e-05,
+      "loss": 0.0017,
+      "step": 5818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.615018965765593e-05,
+      "loss": 0.0035,
+      "step": 5819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.614778892783406e-05,
+      "loss": 0.0053,
+      "step": 5820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6145388198012195e-05,
+      "loss": 0.0015,
+      "step": 5821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6142987468190333e-05,
+      "loss": 0.0016,
+      "step": 5822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6140586738368465e-05,
+      "loss": 0.0014,
+      "step": 5823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.61381860085466e-05,
+      "loss": 0.0013,
+      "step": 5824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613578527872473e-05,
+      "loss": 0.0012,
+      "step": 5825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.613338454890287e-05,
+      "loss": 0.3408,
+      "step": 5826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6130983819081006e-05,
+      "loss": 0.003,
+      "step": 5827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612858308925914e-05,
+      "loss": 0.0012,
+      "step": 5828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612618235943727e-05,
+      "loss": 0.8491,
+      "step": 5829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612378162961541e-05,
+      "loss": 0.0013,
+      "step": 5830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612138089979354e-05,
+      "loss": 0.0012,
+      "step": 5831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611898016997167e-05,
+      "loss": 0.0014,
+      "step": 5832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6116579440149804e-05,
+      "loss": 0.0013,
+      "step": 5833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.611417871032794e-05,
+      "loss": 0.0015,
+      "step": 5834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6111777980506074e-05,
+      "loss": 0.0039,
+      "step": 5835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610937725068421e-05,
+      "loss": 0.0014,
+      "step": 5836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6106976520862344e-05,
+      "loss": 0.0014,
+      "step": 5837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6104575791040476e-05,
+      "loss": 0.8239,
+      "step": 5838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6102175061218615e-05,
+      "loss": 0.8297,
+      "step": 5839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6099774331396747e-05,
+      "loss": 0.0014,
+      "step": 5840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609737360157488e-05,
+      "loss": 0.0016,
+      "step": 5841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609497287175301e-05,
+      "loss": 0.0014,
+      "step": 5842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609257214193115e-05,
+      "loss": 0.0012,
+      "step": 5843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609017141210928e-05,
+      "loss": 0.0013,
+      "step": 5844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608777068228742e-05,
+      "loss": 0.004,
+      "step": 5845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608536995246555e-05,
+      "loss": 0.0052,
+      "step": 5846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608296922264369e-05,
+      "loss": 0.0013,
+      "step": 5847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608056849282182e-05,
+      "loss": 0.0015,
+      "step": 5848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607816776299995e-05,
+      "loss": 0.0019,
+      "step": 5849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6075767033178085e-05,
+      "loss": 0.0015,
+      "step": 5850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6073366303356223e-05,
+      "loss": 0.0017,
+      "step": 5851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6070965573534355e-05,
+      "loss": 0.0017,
+      "step": 5852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606856484371249e-05,
+      "loss": 0.0018,
+      "step": 5853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6066164113890626e-05,
+      "loss": 0.0045,
+      "step": 5854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606376338406876e-05,
+      "loss": 0.0017,
+      "step": 5855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6061362654246896e-05,
+      "loss": 0.0015,
+      "step": 5856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605896192442503e-05,
+      "loss": 0.0016,
+      "step": 5857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605656119460316e-05,
+      "loss": 0.0014,
+      "step": 5858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605416046478129e-05,
+      "loss": 0.0012,
+      "step": 5859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605175973495943e-05,
+      "loss": 0.0012,
+      "step": 5860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604935900513756e-05,
+      "loss": 0.0013,
+      "step": 5861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6046958275315694e-05,
+      "loss": 0.0015,
+      "step": 5862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604455754549383e-05,
+      "loss": 0.0018,
+      "step": 5863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604215681567197e-05,
+      "loss": 0.8327,
+      "step": 5864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60397560858501e-05,
+      "loss": 0.0017,
+      "step": 5865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6037355356028234e-05,
+      "loss": 0.0015,
+      "step": 5866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6034954626206366e-05,
+      "loss": 0.0017,
+      "step": 5867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6032553896384505e-05,
+      "loss": 0.0017,
+      "step": 5868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6030153166562636e-05,
+      "loss": 0.0021,
+      "step": 5869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602775243674077e-05,
+      "loss": 0.0012,
+      "step": 5870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60253517069189e-05,
+      "loss": 0.0014,
+      "step": 5871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602295097709704e-05,
+      "loss": 0.0053,
+      "step": 5872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602055024727518e-05,
+      "loss": 0.0013,
+      "step": 5873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601814951745331e-05,
+      "loss": 0.0018,
+      "step": 5874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601574878763144e-05,
+      "loss": 0.0064,
+      "step": 5875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601334805780958e-05,
+      "loss": 0.0038,
+      "step": 5876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.601094732798771e-05,
+      "loss": 0.0013,
+      "step": 5877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.600854659816584e-05,
+      "loss": 0.0016,
+      "step": 5878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6006145868343975e-05,
+      "loss": 0.0017,
+      "step": 5879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6003745138522113e-05,
+      "loss": 0.0014,
+      "step": 5880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6001344408700245e-05,
+      "loss": 0.0015,
+      "step": 5881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5998943678878384e-05,
+      "loss": 0.0016,
+      "step": 5882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5996542949056516e-05,
+      "loss": 0.0039,
+      "step": 5883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.599414221923465e-05,
+      "loss": 0.0016,
+      "step": 5884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5991741489412786e-05,
+      "loss": 0.0011,
+      "step": 5885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598934075959092e-05,
+      "loss": 0.001,
+      "step": 5886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598694002976905e-05,
+      "loss": 0.001,
+      "step": 5887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598453929994718e-05,
+      "loss": 0.001,
+      "step": 5888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598213857012532e-05,
+      "loss": 0.0015,
+      "step": 5889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597973784030345e-05,
+      "loss": 0.0033,
+      "step": 5890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597733711048159e-05,
+      "loss": 0.0013,
+      "step": 5891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597493638065972e-05,
+      "loss": 0.0015,
+      "step": 5892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597253565083786e-05,
+      "loss": 0.4608,
+      "step": 5893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.597013492101599e-05,
+      "loss": 0.0019,
+      "step": 5894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5967734191194124e-05,
+      "loss": 0.0014,
+      "step": 5895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5965333461372256e-05,
+      "loss": 0.0012,
+      "step": 5896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5962932731550395e-05,
+      "loss": 0.0017,
+      "step": 5897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5960532001728526e-05,
+      "loss": 0.0013,
+      "step": 5898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.595813127190666e-05,
+      "loss": 0.0012,
+      "step": 5899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59557305420848e-05,
+      "loss": 0.5355,
+      "step": 5900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.595332981226293e-05,
+      "loss": 0.0034,
+      "step": 5901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.595092908244107e-05,
+      "loss": 0.0016,
+      "step": 5902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59485283526192e-05,
+      "loss": 0.0012,
+      "step": 5903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.594612762279733e-05,
+      "loss": 0.0036,
+      "step": 5904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.594372689297546e-05,
+      "loss": 0.0019,
+      "step": 5905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59413261631536e-05,
+      "loss": 0.0016,
+      "step": 5906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593892543333173e-05,
+      "loss": 0.4923,
+      "step": 5907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5936524703509865e-05,
+      "loss": 0.7563,
+      "step": 5908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5934123973688e-05,
+      "loss": 0.0016,
+      "step": 5909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.593172324386614e-05,
+      "loss": 0.0015,
+      "step": 5910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5929322514044274e-05,
+      "loss": 0.0011,
+      "step": 5911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5926921784222406e-05,
+      "loss": 0.0012,
+      "step": 5912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.592452105440054e-05,
+      "loss": 0.0072,
+      "step": 5913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5922120324578676e-05,
+      "loss": 0.0017,
+      "step": 5914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591971959475681e-05,
+      "loss": 0.0014,
+      "step": 5915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591731886493494e-05,
+      "loss": 0.0014,
+      "step": 5916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591491813511307e-05,
+      "loss": 0.0015,
+      "step": 5917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591251740529121e-05,
+      "loss": 0.0013,
+      "step": 5918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.591011667546935e-05,
+      "loss": 0.0012,
+      "step": 5919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590771594564748e-05,
+      "loss": 0.0016,
+      "step": 5920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590531521582561e-05,
+      "loss": 0.0017,
+      "step": 5921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5902914486003744e-05,
+      "loss": 0.0018,
+      "step": 5922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.590051375618188e-05,
+      "loss": 0.0016,
+      "step": 5923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5898113026360014e-05,
+      "loss": 0.0015,
+      "step": 5924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5895712296538146e-05,
+      "loss": 0.0013,
+      "step": 5925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589331156671628e-05,
+      "loss": 0.0016,
+      "step": 5926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.589091083689442e-05,
+      "loss": 0.0082,
+      "step": 5927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5888510107072555e-05,
+      "loss": 0.0015,
+      "step": 5928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588610937725069e-05,
+      "loss": 0.0014,
+      "step": 5929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588370864742882e-05,
+      "loss": 0.2535,
+      "step": 5930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588130791760696e-05,
+      "loss": 0.0017,
+      "step": 5931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587890718778509e-05,
+      "loss": 0.0015,
+      "step": 5932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587650645796322e-05,
+      "loss": 0.001,
+      "step": 5933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587410572814135e-05,
+      "loss": 0.0014,
+      "step": 5934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587170499831949e-05,
+      "loss": 0.0208,
+      "step": 5935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586930426849763e-05,
+      "loss": 0.0019,
+      "step": 5936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586690353867576e-05,
+      "loss": 0.0017,
+      "step": 5937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586450280885389e-05,
+      "loss": 0.0016,
+      "step": 5938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5862102079032025e-05,
+      "loss": 0.0015,
+      "step": 5939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5859701349210164e-05,
+      "loss": 0.0013,
+      "step": 5940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5857300619388296e-05,
+      "loss": 0.0042,
+      "step": 5941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.585489988956643e-05,
+      "loss": 0.0014,
+      "step": 5942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5852499159744566e-05,
+      "loss": 0.0012,
+      "step": 5943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58500984299227e-05,
+      "loss": 0.0014,
+      "step": 5944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584769770010083e-05,
+      "loss": 0.0014,
+      "step": 5945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584529697027897e-05,
+      "loss": 0.0015,
+      "step": 5946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.58428962404571e-05,
+      "loss": 0.0015,
+      "step": 5947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.584049551063524e-05,
+      "loss": 0.0014,
+      "step": 5948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583809478081337e-05,
+      "loss": 0.0017,
+      "step": 5949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.58356940509915e-05,
+      "loss": 0.0014,
+      "step": 5950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5833293321169634e-05,
+      "loss": 0.0013,
+      "step": 5951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.583089259134777e-05,
+      "loss": 0.304,
+      "step": 5952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5828491861525904e-05,
+      "loss": 0.0013,
+      "step": 5953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5826091131704036e-05,
+      "loss": 0.7861,
+      "step": 5954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5823690401882175e-05,
+      "loss": 0.001,
+      "step": 5955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5821289672060306e-05,
+      "loss": 0.0329,
+      "step": 5956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5818888942238445e-05,
+      "loss": 0.0014,
+      "step": 5957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.581648821241658e-05,
+      "loss": 0.7952,
+      "step": 5958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.581408748259471e-05,
+      "loss": 0.0011,
+      "step": 5959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.581168675277285e-05,
+      "loss": 0.0015,
+      "step": 5960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.580928602295098e-05,
+      "loss": 0.0011,
+      "step": 5961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.580688529312911e-05,
+      "loss": 0.0014,
+      "step": 5962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.580448456330724e-05,
+      "loss": 0.0014,
+      "step": 5963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.580208383348538e-05,
+      "loss": 0.0009,
+      "step": 5964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.579968310366352e-05,
+      "loss": 0.0012,
+      "step": 5965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.579728237384165e-05,
+      "loss": 0.0011,
+      "step": 5966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.579488164401978e-05,
+      "loss": 0.0017,
+      "step": 5967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5792480914197915e-05,
+      "loss": 0.0023,
+      "step": 5968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5790080184376054e-05,
+      "loss": 0.728,
+      "step": 5969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5787679454554185e-05,
+      "loss": 0.0015,
+      "step": 5970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.578527872473232e-05,
+      "loss": 0.0018,
+      "step": 5971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.578287799491045e-05,
+      "loss": 0.0024,
+      "step": 5972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5780477265088594e-05,
+      "loss": 0.0014,
+      "step": 5973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5778076535266726e-05,
+      "loss": 0.0019,
+      "step": 5974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577567580544486e-05,
+      "loss": 0.0013,
+      "step": 5975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577327507562299e-05,
+      "loss": 0.0022,
+      "step": 5976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.577087434580113e-05,
+      "loss": 0.0021,
+      "step": 5977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576847361597926e-05,
+      "loss": 0.0025,
+      "step": 5978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576607288615739e-05,
+      "loss": 0.0029,
+      "step": 5979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5763672156335524e-05,
+      "loss": 0.0023,
+      "step": 5980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576127142651366e-05,
+      "loss": 0.0015,
+      "step": 5981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57588706966918e-05,
+      "loss": 0.0025,
+      "step": 5982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575646996686993e-05,
+      "loss": 0.0012,
+      "step": 5983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5754069237048065e-05,
+      "loss": 0.0013,
+      "step": 5984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5751668507226196e-05,
+      "loss": 0.0015,
+      "step": 5985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5749267777404335e-05,
+      "loss": 0.0016,
+      "step": 5986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574686704758247e-05,
+      "loss": 0.0026,
+      "step": 5987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.57444663177606e-05,
+      "loss": 0.0011,
+      "step": 5988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574206558793873e-05,
+      "loss": 0.0013,
+      "step": 5989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573966485811687e-05,
+      "loss": 0.0014,
+      "step": 5990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573726412829501e-05,
+      "loss": 0.0016,
+      "step": 5991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573486339847314e-05,
+      "loss": 0.0019,
+      "step": 5992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573246266865127e-05,
+      "loss": 0.0023,
+      "step": 5993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573006193882941e-05,
+      "loss": 0.0014,
+      "step": 5994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572766120900754e-05,
+      "loss": 0.0013,
+      "step": 5995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572526047918567e-05,
+      "loss": 0.0019,
+      "step": 5996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5722859749363805e-05,
+      "loss": 0.0017,
+      "step": 5997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5720459019541944e-05,
+      "loss": 0.0022,
+      "step": 5998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5718058289720075e-05,
+      "loss": 0.0021,
+      "step": 5999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5715657559898214e-05,
+      "loss": 1.3207,
+      "step": 6000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5713256830076346e-05,
+      "loss": 0.0017,
+      "step": 6001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571085610025448e-05,
+      "loss": 0.0012,
+      "step": 6002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5708455370432616e-05,
+      "loss": 0.0018,
+      "step": 6003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570605464061075e-05,
+      "loss": 0.7584,
+      "step": 6004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570365391078888e-05,
+      "loss": 0.0015,
+      "step": 6005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570125318096701e-05,
+      "loss": 0.0024,
+      "step": 6006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569885245114515e-05,
+      "loss": 0.0015,
+      "step": 6007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569645172132328e-05,
+      "loss": 0.0015,
+      "step": 6008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5694050991501414e-05,
+      "loss": 0.0021,
+      "step": 6009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569165026167955e-05,
+      "loss": 0.0018,
+      "step": 6010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568924953185769e-05,
+      "loss": 0.002,
+      "step": 6011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.568684880203582e-05,
+      "loss": 0.0017,
+      "step": 6012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5684448072213955e-05,
+      "loss": 0.0017,
+      "step": 6013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5682047342392086e-05,
+      "loss": 0.0025,
+      "step": 6014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5679646612570225e-05,
+      "loss": 0.0017,
+      "step": 6015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567724588274836e-05,
+      "loss": 0.0018,
+      "step": 6016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567484515292649e-05,
+      "loss": 0.0018,
+      "step": 6017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567244442310462e-05,
+      "loss": 0.0024,
+      "step": 6018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567004369328276e-05,
+      "loss": 0.0011,
+      "step": 6019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.56676429634609e-05,
+      "loss": 0.0021,
+      "step": 6020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.566524223363903e-05,
+      "loss": 0.0017,
+      "step": 6021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.566284150381716e-05,
+      "loss": 0.0068,
+      "step": 6022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.566044077399529e-05,
+      "loss": 0.0015,
+      "step": 6023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.565804004417343e-05,
+      "loss": 0.0013,
+      "step": 6024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.565563931435156e-05,
+      "loss": 0.0018,
+      "step": 6025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5653238584529695e-05,
+      "loss": 0.0026,
+      "step": 6026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5650837854707834e-05,
+      "loss": 0.0016,
+      "step": 6027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.564843712488597e-05,
+      "loss": 0.0015,
+      "step": 6028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5646036395064104e-05,
+      "loss": 0.9478,
+      "step": 6029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5643635665242236e-05,
+      "loss": 0.0013,
+      "step": 6030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.564123493542037e-05,
+      "loss": 0.0055,
+      "step": 6031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5638834205598506e-05,
+      "loss": 0.0019,
+      "step": 6032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.563643347577664e-05,
+      "loss": 0.0067,
+      "step": 6033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.563403274595477e-05,
+      "loss": 0.0015,
+      "step": 6034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.56316320161329e-05,
+      "loss": 0.0012,
+      "step": 6035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562923128631104e-05,
+      "loss": 0.0012,
+      "step": 6036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562683055648918e-05,
+      "loss": 0.0013,
+      "step": 6037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562442982666731e-05,
+      "loss": 0.001,
+      "step": 6038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.562202909684544e-05,
+      "loss": 0.0015,
+      "step": 6039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5619628367023574e-05,
+      "loss": 0.0014,
+      "step": 6040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.561722763720171e-05,
+      "loss": 0.0015,
+      "step": 6041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5614826907379845e-05,
+      "loss": 0.0009,
+      "step": 6042
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5612426177557976e-05,
+      "loss": 0.0017,
+      "step": 6043
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5610025447736115e-05,
+      "loss": 0.0016,
+      "step": 6044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560762471791425e-05,
+      "loss": 0.0016,
+      "step": 6045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5605223988092385e-05,
+      "loss": 0.0011,
+      "step": 6046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560282325827052e-05,
+      "loss": 0.0015,
+      "step": 6047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560042252844865e-05,
+      "loss": 0.0012,
+      "step": 6048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559802179862679e-05,
+      "loss": 0.0012,
+      "step": 6049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559562106880492e-05,
+      "loss": 0.0011,
+      "step": 6050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559322033898305e-05,
+      "loss": 0.0011,
+      "step": 6051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559081960916118e-05,
+      "loss": 0.0014,
+      "step": 6052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558841887933932e-05,
+      "loss": 0.0014,
+      "step": 6053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558601814951745e-05,
+      "loss": 0.0096,
+      "step": 6054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558361741969559e-05,
+      "loss": 0.0009,
+      "step": 6055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5581216689873724e-05,
+      "loss": 0.5105,
+      "step": 6056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557881596005186e-05,
+      "loss": 0.0013,
+      "step": 6057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5576415230229994e-05,
+      "loss": 0.0012,
+      "step": 6058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5574014500408126e-05,
+      "loss": 0.001,
+      "step": 6059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557161377058626e-05,
+      "loss": 0.8918,
+      "step": 6060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5569213040764396e-05,
+      "loss": 0.0011,
+      "step": 6061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556681231094253e-05,
+      "loss": 0.0007,
+      "step": 6062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556441158112066e-05,
+      "loss": 0.0012,
+      "step": 6063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55620108512988e-05,
+      "loss": 0.0011,
+      "step": 6064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555961012147693e-05,
+      "loss": 0.8778,
+      "step": 6065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555720939165507e-05,
+      "loss": 0.0009,
+      "step": 6066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55548086618332e-05,
+      "loss": 0.0012,
+      "step": 6067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555240793201133e-05,
+      "loss": 0.0011,
+      "step": 6068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5550007202189464e-05,
+      "loss": 0.0012,
+      "step": 6069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55476064723676e-05,
+      "loss": 0.0009,
+      "step": 6070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5545205742545735e-05,
+      "loss": 0.0017,
+      "step": 6071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5542805012723866e-05,
+      "loss": 0.0013,
+      "step": 6072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5540404282902e-05,
+      "loss": 0.0013,
+      "step": 6073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5538003553080143e-05,
+      "loss": 0.0012,
+      "step": 6074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5535602823258275e-05,
+      "loss": 0.0009,
+      "step": 6075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553320209343641e-05,
+      "loss": 0.0012,
+      "step": 6076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.553080136361454e-05,
+      "loss": 0.0011,
+      "step": 6077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552840063379268e-05,
+      "loss": 0.0011,
+      "step": 6078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552599990397081e-05,
+      "loss": 0.0009,
+      "step": 6079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552359917414894e-05,
+      "loss": 0.0014,
+      "step": 6080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552119844432707e-05,
+      "loss": 0.001,
+      "step": 6081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551879771450521e-05,
+      "loss": 0.0013,
+      "step": 6082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551639698468335e-05,
+      "loss": 0.0011,
+      "step": 6083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.551399625486148e-05,
+      "loss": 0.0011,
+      "step": 6084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5511595525039614e-05,
+      "loss": 0.0011,
+      "step": 6085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5509194795217745e-05,
+      "loss": 0.0012,
+      "step": 6086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5506794065395884e-05,
+      "loss": 0.0013,
+      "step": 6087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5504393335574016e-05,
+      "loss": 0.0011,
+      "step": 6088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550199260575215e-05,
+      "loss": 0.0012,
+      "step": 6089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.549959187593028e-05,
+      "loss": 0.0012,
+      "step": 6090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.549719114610842e-05,
+      "loss": 0.0014,
+      "step": 6091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5494790416286557e-05,
+      "loss": 0.0012,
+      "step": 6092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.549238968646469e-05,
+      "loss": 0.001,
+      "step": 6093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548998895664282e-05,
+      "loss": 0.001,
+      "step": 6094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548758822682096e-05,
+      "loss": 0.001,
+      "step": 6095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548518749699909e-05,
+      "loss": 0.0012,
+      "step": 6096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548278676717722e-05,
+      "loss": 0.0013,
+      "step": 6097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5480386037355354e-05,
+      "loss": 0.0009,
+      "step": 6098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547798530753349e-05,
+      "loss": 0.0009,
+      "step": 6099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5475584577711624e-05,
+      "loss": 0.0033,
+      "step": 6100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547318384788976e-05,
+      "loss": 0.0009,
+      "step": 6101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5470783118067895e-05,
+      "loss": 0.0012,
+      "step": 6102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546838238824603e-05,
+      "loss": 0.001,
+      "step": 6103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5465981658424165e-05,
+      "loss": 0.0009,
+      "step": 6104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54635809286023e-05,
+      "loss": 0.001,
+      "step": 6105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546118019878043e-05,
+      "loss": 0.001,
+      "step": 6106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.545877946895856e-05,
+      "loss": 0.001,
+      "step": 6107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54563787391367e-05,
+      "loss": 0.0009,
+      "step": 6108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.545397800931483e-05,
+      "loss": 0.0009,
+      "step": 6109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.545157727949297e-05,
+      "loss": 0.0011,
+      "step": 6110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.54491765496711e-05,
+      "loss": 0.0008,
+      "step": 6111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.544677581984924e-05,
+      "loss": 0.0011,
+      "step": 6112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.544437509002737e-05,
+      "loss": 0.0009,
+      "step": 6113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5441974360205504e-05,
+      "loss": 0.0008,
+      "step": 6114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5439573630383635e-05,
+      "loss": 0.0011,
+      "step": 6115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5437172900561774e-05,
+      "loss": 0.0009,
+      "step": 6116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5434772170739906e-05,
+      "loss": 0.0009,
+      "step": 6117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543237144091804e-05,
+      "loss": 0.0012,
+      "step": 6118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5429970711096176e-05,
+      "loss": 0.0009,
+      "step": 6119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542756998127431e-05,
+      "loss": 0.0009,
+      "step": 6120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5425169251452446e-05,
+      "loss": 0.0009,
+      "step": 6121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542276852163058e-05,
+      "loss": 0.0009,
+      "step": 6122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.542036779180871e-05,
+      "loss": 0.001,
+      "step": 6123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541796706198685e-05,
+      "loss": 0.0009,
+      "step": 6124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541556633216498e-05,
+      "loss": 0.6225,
+      "step": 6125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541316560234311e-05,
+      "loss": 0.0007,
+      "step": 6126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5410764872521244e-05,
+      "loss": 0.8967,
+      "step": 6127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540836414269938e-05,
+      "loss": 0.0008,
+      "step": 6128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540596341287752e-05,
+      "loss": 0.0009,
+      "step": 6129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.540356268305565e-05,
+      "loss": 0.0011,
+      "step": 6130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5401161953233785e-05,
+      "loss": 0.0009,
+      "step": 6131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539876122341192e-05,
+      "loss": 0.0011,
+      "step": 6132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5396360493590055e-05,
+      "loss": 0.8556,
+      "step": 6133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539395976376819e-05,
+      "loss": 0.001,
+      "step": 6134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539155903394632e-05,
+      "loss": 0.8859,
+      "step": 6135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538915830412445e-05,
+      "loss": 0.001,
+      "step": 6136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538675757430259e-05,
+      "loss": 0.0012,
+      "step": 6137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538435684448073e-05,
+      "loss": 0.001,
+      "step": 6138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538195611465886e-05,
+      "loss": 0.0011,
+      "step": 6139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537955538483699e-05,
+      "loss": 0.001,
+      "step": 6140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537715465501513e-05,
+      "loss": 0.0009,
+      "step": 6141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.537475392519326e-05,
+      "loss": 0.0011,
+      "step": 6142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5372353195371394e-05,
+      "loss": 0.0011,
+      "step": 6143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5369952465549525e-05,
+      "loss": 0.0011,
+      "step": 6144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5367551735727664e-05,
+      "loss": 0.0013,
+      "step": 6145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5365151005905796e-05,
+      "loss": 0.0011,
+      "step": 6146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5362750276083934e-05,
+      "loss": 0.001,
+      "step": 6147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5360349546262066e-05,
+      "loss": 0.0014,
+      "step": 6148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53579488164402e-05,
+      "loss": 0.001,
+      "step": 6149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5355548086618336e-05,
+      "loss": 0.0011,
+      "step": 6150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.535314735679647e-05,
+      "loss": 0.0012,
+      "step": 6151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53507466269746e-05,
+      "loss": 0.0012,
+      "step": 6152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534834589715273e-05,
+      "loss": 0.0014,
+      "step": 6153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534594516733087e-05,
+      "loss": 0.872,
+      "step": 6154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5343544437509e-05,
+      "loss": 0.0011,
+      "step": 6155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534114370768714e-05,
+      "loss": 0.001,
+      "step": 6156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533874297786527e-05,
+      "loss": 0.0012,
+      "step": 6157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.533634224804341e-05,
+      "loss": 0.0012,
+      "step": 6158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.533394151822154e-05,
+      "loss": 0.0011,
+      "step": 6159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5331540788399675e-05,
+      "loss": 0.8527,
+      "step": 6160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5329140058577807e-05,
+      "loss": 0.0011,
+      "step": 6161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5326739328755945e-05,
+      "loss": 0.0015,
+      "step": 6162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.532433859893408e-05,
+      "loss": 0.0013,
+      "step": 6163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.532193786911221e-05,
+      "loss": 0.0015,
+      "step": 6164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531953713929035e-05,
+      "loss": 0.0012,
+      "step": 6165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531713640946848e-05,
+      "loss": 0.0015,
+      "step": 6166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531473567964662e-05,
+      "loss": 0.0012,
+      "step": 6167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.531233494982475e-05,
+      "loss": 0.0015,
+      "step": 6168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530993422000288e-05,
+      "loss": 0.0015,
+      "step": 6169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530753349018101e-05,
+      "loss": 0.0013,
+      "step": 6170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530513276035915e-05,
+      "loss": 0.0014,
+      "step": 6171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5302732030537284e-05,
+      "loss": 0.0014,
+      "step": 6172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5300331300715415e-05,
+      "loss": 0.0015,
+      "step": 6173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5297930570893554e-05,
+      "loss": 0.0014,
+      "step": 6174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529552984107169e-05,
+      "loss": 0.0013,
+      "step": 6175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5293129111249824e-05,
+      "loss": 0.0014,
+      "step": 6176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5290728381427956e-05,
+      "loss": 0.0015,
+      "step": 6177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528832765160609e-05,
+      "loss": 0.0011,
+      "step": 6178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5285926921784226e-05,
+      "loss": 0.0013,
+      "step": 6179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528352619196236e-05,
+      "loss": 0.0011,
+      "step": 6180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.528112546214049e-05,
+      "loss": 0.001,
+      "step": 6181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527872473231862e-05,
+      "loss": 0.001,
+      "step": 6182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527632400249676e-05,
+      "loss": 0.0015,
+      "step": 6183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52739232726749e-05,
+      "loss": 0.0036,
+      "step": 6184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.527152254285303e-05,
+      "loss": 0.0012,
+      "step": 6185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526912181303116e-05,
+      "loss": 0.0015,
+      "step": 6186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5266721083209294e-05,
+      "loss": 0.0012,
+      "step": 6187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526432035338743e-05,
+      "loss": 0.0014,
+      "step": 6188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5261919623565565e-05,
+      "loss": 0.0014,
+      "step": 6189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5259518893743697e-05,
+      "loss": 0.0012,
+      "step": 6190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5257118163921835e-05,
+      "loss": 0.0013,
+      "step": 6191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5254717434099974e-05,
+      "loss": 0.0013,
+      "step": 6192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5252316704278106e-05,
+      "loss": 0.0014,
+      "step": 6193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524991597445624e-05,
+      "loss": 0.001,
+      "step": 6194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524751524463437e-05,
+      "loss": 0.0014,
+      "step": 6195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524511451481251e-05,
+      "loss": 0.0011,
+      "step": 6196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524271378499064e-05,
+      "loss": 0.0015,
+      "step": 6197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524031305516877e-05,
+      "loss": 0.0013,
+      "step": 6198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52379123253469e-05,
+      "loss": 0.0014,
+      "step": 6199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523551159552504e-05,
+      "loss": 0.0012,
+      "step": 6200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5233110865703173e-05,
+      "loss": 0.0014,
+      "step": 6201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523071013588131e-05,
+      "loss": 0.8454,
+      "step": 6202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5228309406059444e-05,
+      "loss": 0.0011,
+      "step": 6203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5225908676237576e-05,
+      "loss": 0.0013,
+      "step": 6204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5223507946415714e-05,
+      "loss": 0.0015,
+      "step": 6205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5221107216593846e-05,
+      "loss": 0.0011,
+      "step": 6206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521870648677198e-05,
+      "loss": 0.0013,
+      "step": 6207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5216305756950116e-05,
+      "loss": 0.0011,
+      "step": 6208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521390502712825e-05,
+      "loss": 0.0014,
+      "step": 6209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521150429730638e-05,
+      "loss": 0.001,
+      "step": 6210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520910356748452e-05,
+      "loss": 0.001,
+      "step": 6211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520670283766265e-05,
+      "loss": 0.0009,
+      "step": 6212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520430210784079e-05,
+      "loss": 0.8233,
+      "step": 6213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520190137801892e-05,
+      "loss": 0.0013,
+      "step": 6214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519950064819705e-05,
+      "loss": 0.0012,
+      "step": 6215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5197099918375184e-05,
+      "loss": 0.0014,
+      "step": 6216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519469918855332e-05,
+      "loss": 0.0009,
+      "step": 6217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5192298458731455e-05,
+      "loss": 0.0012,
+      "step": 6218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5189897728909587e-05,
+      "loss": 0.0017,
+      "step": 6219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5187496999087725e-05,
+      "loss": 0.81,
+      "step": 6220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5185096269265864e-05,
+      "loss": 0.0016,
+      "step": 6221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5182695539443995e-05,
+      "loss": 0.0014,
+      "step": 6222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518029480962213e-05,
+      "loss": 0.0013,
+      "step": 6223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517789407980026e-05,
+      "loss": 0.0016,
+      "step": 6224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51754933499784e-05,
+      "loss": 0.0011,
+      "step": 6225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517309262015653e-05,
+      "loss": 0.0013,
+      "step": 6226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517069189033466e-05,
+      "loss": 0.0017,
+      "step": 6227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516829116051279e-05,
+      "loss": 0.0018,
+      "step": 6228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516589043069093e-05,
+      "loss": 0.0015,
+      "step": 6229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516348970086907e-05,
+      "loss": 0.0019,
+      "step": 6230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.51610889710472e-05,
+      "loss": 0.0017,
+      "step": 6231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5158688241225334e-05,
+      "loss": 0.0021,
+      "step": 6232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5156287511403466e-05,
+      "loss": 0.0013,
+      "step": 6233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5153886781581604e-05,
+      "loss": 0.0017,
+      "step": 6234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5151486051759736e-05,
+      "loss": 0.0015,
+      "step": 6235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.514908532193787e-05,
+      "loss": 0.0015,
+      "step": 6236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5146684592116e-05,
+      "loss": 0.0015,
+      "step": 6237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5144283862294145e-05,
+      "loss": 0.0018,
+      "step": 6238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.514188313247228e-05,
+      "loss": 0.0016,
+      "step": 6239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.513948240265041e-05,
+      "loss": 0.0016,
+      "step": 6240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.513708167282854e-05,
+      "loss": 0.0011,
+      "step": 6241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.513468094300668e-05,
+      "loss": 0.0011,
+      "step": 6242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.513228021318481e-05,
+      "loss": 0.7804,
+      "step": 6243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512987948336294e-05,
+      "loss": 0.0011,
+      "step": 6244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5127478753541074e-05,
+      "loss": 0.0015,
+      "step": 6245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512507802371921e-05,
+      "loss": 0.0017,
+      "step": 6246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512267729389735e-05,
+      "loss": 0.0014,
+      "step": 6247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.512027656407548e-05,
+      "loss": 0.0019,
+      "step": 6248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5117875834253615e-05,
+      "loss": 0.0014,
+      "step": 6249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.511547510443175e-05,
+      "loss": 0.002,
+      "step": 6250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5113074374609885e-05,
+      "loss": 0.0025,
+      "step": 6251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.511067364478802e-05,
+      "loss": 0.0016,
+      "step": 6252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510827291496615e-05,
+      "loss": 0.0021,
+      "step": 6253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510587218514428e-05,
+      "loss": 0.002,
+      "step": 6254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510347145532242e-05,
+      "loss": 1.5059,
+      "step": 6255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510107072550055e-05,
+      "loss": 0.002,
+      "step": 6256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509866999567869e-05,
+      "loss": 0.0018,
+      "step": 6257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509626926585682e-05,
+      "loss": 0.0017,
+      "step": 6258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509386853603496e-05,
+      "loss": 0.002,
+      "step": 6259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509146780621309e-05,
+      "loss": 0.0018,
+      "step": 6260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5089067076391224e-05,
+      "loss": 0.0052,
+      "step": 6261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5086666346569356e-05,
+      "loss": 0.0023,
+      "step": 6262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5084265616747494e-05,
+      "loss": 0.002,
+      "step": 6263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5081864886925626e-05,
+      "loss": 0.0022,
+      "step": 6264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507946415710376e-05,
+      "loss": 0.0024,
+      "step": 6265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5077063427281896e-05,
+      "loss": 0.0017,
+      "step": 6266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507466269746003e-05,
+      "loss": 0.0027,
+      "step": 6267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507226196763817e-05,
+      "loss": 0.0014,
+      "step": 6268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50698612378163e-05,
+      "loss": 0.0015,
+      "step": 6269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506746050799443e-05,
+      "loss": 0.0017,
+      "step": 6270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506505977817256e-05,
+      "loss": 0.0013,
+      "step": 6271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50626590483507e-05,
+      "loss": 0.7427,
+      "step": 6272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506025831852883e-05,
+      "loss": 0.7254,
+      "step": 6273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5057857588706964e-05,
+      "loss": 0.0015,
+      "step": 6274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50554568588851e-05,
+      "loss": 0.0023,
+      "step": 6275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505305612906324e-05,
+      "loss": 0.0025,
+      "step": 6276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505065539924137e-05,
+      "loss": 0.0016,
+      "step": 6277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5048254669419505e-05,
+      "loss": 0.0019,
+      "step": 6278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504585393959764e-05,
+      "loss": 0.002,
+      "step": 6279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5043453209775775e-05,
+      "loss": 0.0027,
+      "step": 6280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504105247995391e-05,
+      "loss": 0.0025,
+      "step": 6281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503865175013204e-05,
+      "loss": 0.003,
+      "step": 6282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503625102031017e-05,
+      "loss": 0.0027,
+      "step": 6283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503385029048831e-05,
+      "loss": 0.0085,
+      "step": 6284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503144956066645e-05,
+      "loss": 0.0022,
+      "step": 6285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502904883084458e-05,
+      "loss": 0.0022,
+      "step": 6286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502664810102271e-05,
+      "loss": 0.0011,
+      "step": 6287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5024247371200843e-05,
+      "loss": 0.0022,
+      "step": 6288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502184664137898e-05,
+      "loss": 0.0012,
+      "step": 6289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5019445911557114e-05,
+      "loss": 0.0017,
+      "step": 6290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5017045181735246e-05,
+      "loss": 0.0016,
+      "step": 6291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5014644451913384e-05,
+      "loss": 0.0015,
+      "step": 6292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501224372209152e-05,
+      "loss": 0.0021,
+      "step": 6293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5009842992269655e-05,
+      "loss": 0.4547,
+      "step": 6294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5007442262447786e-05,
+      "loss": 0.0015,
+      "step": 6295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500504153262592e-05,
+      "loss": 0.0011,
+      "step": 6296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.500264080280406e-05,
+      "loss": 0.0019,
+      "step": 6297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.500024007298219e-05,
+      "loss": 0.0016,
+      "step": 6298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499783934316032e-05,
+      "loss": 0.0009,
+      "step": 6299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499543861333845e-05,
+      "loss": 0.0014,
+      "step": 6300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499303788351659e-05,
+      "loss": 0.9436,
+      "step": 6301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.499063715369473e-05,
+      "loss": 0.002,
+      "step": 6302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.498823642387286e-05,
+      "loss": 0.0013,
+      "step": 6303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.498583569405099e-05,
+      "loss": 0.0019,
+      "step": 6304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.498343496422913e-05,
+      "loss": 0.9385,
+      "step": 6305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.498103423440726e-05,
+      "loss": 0.0014,
+      "step": 6306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4978633504585395e-05,
+      "loss": 0.002,
+      "step": 6307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.497623277476353e-05,
+      "loss": 0.0018,
+      "step": 6308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4973832044941665e-05,
+      "loss": 0.0015,
+      "step": 6309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49714313151198e-05,
+      "loss": 0.0016,
+      "step": 6310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4969030585297936e-05,
+      "loss": 0.0009,
+      "step": 6311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.496662985547607e-05,
+      "loss": 0.0018,
+      "step": 6312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49642291256542e-05,
+      "loss": 0.8466,
+      "step": 6313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.496182839583234e-05,
+      "loss": 0.0015,
+      "step": 6314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.495942766601047e-05,
+      "loss": 0.0014,
+      "step": 6315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.49570269361886e-05,
+      "loss": 0.0018,
+      "step": 6316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.495462620636673e-05,
+      "loss": 0.0014,
+      "step": 6317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.495222547654487e-05,
+      "loss": 0.0013,
+      "step": 6318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4949824746723004e-05,
+      "loss": 0.0017,
+      "step": 6319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4947424016901136e-05,
+      "loss": 0.8593,
+      "step": 6320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4945023287079274e-05,
+      "loss": 0.8615,
+      "step": 6321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.494262255725741e-05,
+      "loss": 0.0019,
+      "step": 6322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4940221827435544e-05,
+      "loss": 0.002,
+      "step": 6323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4937821097613676e-05,
+      "loss": 0.789,
+      "step": 6324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493542036779181e-05,
+      "loss": 0.8826,
+      "step": 6325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493301963796995e-05,
+      "loss": 0.002,
+      "step": 6326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493061890814808e-05,
+      "loss": 0.8254,
+      "step": 6327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492821817832621e-05,
+      "loss": 0.0026,
+      "step": 6328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492581744850434e-05,
+      "loss": 1.5724,
+      "step": 6329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492341671868248e-05,
+      "loss": 1.5308,
+      "step": 6330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492101598886062e-05,
+      "loss": 2.3184,
+      "step": 6331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491861525903875e-05,
+      "loss": 0.8285,
+      "step": 6332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491621452921688e-05,
+      "loss": 0.4381,
+      "step": 6333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4913813799395015e-05,
+      "loss": 0.6808,
+      "step": 6334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491141306957315e-05,
+      "loss": 0.6998,
+      "step": 6335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4909012339751285e-05,
+      "loss": 0.8025,
+      "step": 6336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490661160992942e-05,
+      "loss": 0.6884,
+      "step": 6337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490421088010755e-05,
+      "loss": 0.0046,
+      "step": 6338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4901810150285694e-05,
+      "loss": 0.0067,
+      "step": 6339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4899409420463826e-05,
+      "loss": 0.5537,
+      "step": 6340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489700869064196e-05,
+      "loss": 0.0056,
+      "step": 6341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489460796082009e-05,
+      "loss": 0.0068,
+      "step": 6342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.489220723099823e-05,
+      "loss": 0.0103,
+      "step": 6343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488980650117636e-05,
+      "loss": 1.7049,
+      "step": 6344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488740577135449e-05,
+      "loss": 0.009,
+      "step": 6345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488500504153262e-05,
+      "loss": 1.0762,
+      "step": 6346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488260431171076e-05,
+      "loss": 0.0074,
+      "step": 6347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48802035818889e-05,
+      "loss": 0.4467,
+      "step": 6348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.487780285206703e-05,
+      "loss": 1.0303,
+      "step": 6349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4875402122245164e-05,
+      "loss": 1.2556,
+      "step": 6350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4873001392423296e-05,
+      "loss": 1.2343,
+      "step": 6351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4870600662601434e-05,
+      "loss": 1.2273,
+      "step": 6352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4868199932779566e-05,
+      "loss": 0.4256,
+      "step": 6353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48657992029577e-05,
+      "loss": 0.6994,
+      "step": 6354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486339847313583e-05,
+      "loss": 1.0068,
+      "step": 6355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.486099774331397e-05,
+      "loss": 0.5926,
+      "step": 6356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485859701349211e-05,
+      "loss": 0.3532,
+      "step": 6357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485619628367024e-05,
+      "loss": 0.9585,
+      "step": 6358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485379555384837e-05,
+      "loss": 0.5249,
+      "step": 6359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.485139482402651e-05,
+      "loss": 0.9306,
+      "step": 6360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484899409420464e-05,
+      "loss": 0.4785,
+      "step": 6361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484659336438277e-05,
+      "loss": 0.8189,
+      "step": 6362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4844192634560905e-05,
+      "loss": 0.3354,
+      "step": 6363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484179190473904e-05,
+      "loss": 0.4493,
+      "step": 6364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4839391174917175e-05,
+      "loss": 0.6538,
+      "step": 6365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4836990445095314e-05,
+      "loss": 0.5286,
+      "step": 6366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4834589715273445e-05,
+      "loss": 0.8228,
+      "step": 6367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483218898545158e-05,
+      "loss": 0.6078,
+      "step": 6368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4829788255629716e-05,
+      "loss": 0.5977,
+      "step": 6369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482738752580785e-05,
+      "loss": 0.5649,
+      "step": 6370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482498679598598e-05,
+      "loss": 0.6042,
+      "step": 6371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482258606616412e-05,
+      "loss": 0.5579,
+      "step": 6372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482018533634225e-05,
+      "loss": 0.5827,
+      "step": 6373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481778460652038e-05,
+      "loss": 0.5106,
+      "step": 6374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481538387669852e-05,
+      "loss": 0.6786,
+      "step": 6375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481298314687665e-05,
+      "loss": 0.5493,
+      "step": 6376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481058241705479e-05,
+      "loss": 0.4287,
+      "step": 6377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.480818168723292e-05,
+      "loss": 0.7603,
+      "step": 6378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4805780957411054e-05,
+      "loss": 0.6732,
+      "step": 6379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4803380227589186e-05,
+      "loss": 0.7296,
+      "step": 6380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4800979497767324e-05,
+      "loss": 0.7189,
+      "step": 6381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4798578767945456e-05,
+      "loss": 0.6965,
+      "step": 6382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479617803812359e-05,
+      "loss": 0.6801,
+      "step": 6383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479377730830172e-05,
+      "loss": 0.6648,
+      "step": 6384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479137657847986e-05,
+      "loss": 0.7108,
+      "step": 6385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4788975848658e-05,
+      "loss": 0.5566,
+      "step": 6386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.478657511883613e-05,
+      "loss": 0.7017,
+      "step": 6387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.478417438901426e-05,
+      "loss": 0.6143,
+      "step": 6388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47817736591924e-05,
+      "loss": 0.6453,
+      "step": 6389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477937292937053e-05,
+      "loss": 0.6412,
+      "step": 6390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477697219954866e-05,
+      "loss": 0.6498,
+      "step": 6391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4774571469726795e-05,
+      "loss": 0.495,
+      "step": 6392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477217073990493e-05,
+      "loss": 0.5199,
+      "step": 6393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476977001008307e-05,
+      "loss": 0.4938,
+      "step": 6394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4767369280261204e-05,
+      "loss": 0.6782,
+      "step": 6395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4764968550439335e-05,
+      "loss": 0.5059,
+      "step": 6396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476256782061747e-05,
+      "loss": 0.5666,
+      "step": 6397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4760167090795606e-05,
+      "loss": 0.475,
+      "step": 6398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475776636097374e-05,
+      "loss": 0.4748,
+      "step": 6399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475536563115187e-05,
+      "loss": 0.8595,
+      "step": 6400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475296490133e-05,
+      "loss": 0.5383,
+      "step": 6401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475056417150814e-05,
+      "loss": 0.8807,
+      "step": 6402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474816344168628e-05,
+      "loss": 0.9433,
+      "step": 6403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474576271186441e-05,
+      "loss": 0.8134,
+      "step": 6404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474336198204254e-05,
+      "loss": 0.9936,
+      "step": 6405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474096125222068e-05,
+      "loss": 0.4211,
+      "step": 6406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473856052239881e-05,
+      "loss": 1.0329,
+      "step": 6407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4736159792576944e-05,
+      "loss": 0.7704,
+      "step": 6408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4733759062755076e-05,
+      "loss": 0.849,
+      "step": 6409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4731358332933214e-05,
+      "loss": 0.5962,
+      "step": 6410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4728957603111346e-05,
+      "loss": 0.4871,
+      "step": 6411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4726556873289485e-05,
+      "loss": 0.793,
+      "step": 6412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4724156143467617e-05,
+      "loss": 0.3831,
+      "step": 6413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472175541364575e-05,
+      "loss": 0.5887,
+      "step": 6414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471935468382389e-05,
+      "loss": 0.7589,
+      "step": 6415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471695395400202e-05,
+      "loss": 0.6389,
+      "step": 6416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471455322418015e-05,
+      "loss": 0.6013,
+      "step": 6417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471215249435828e-05,
+      "loss": 0.5844,
+      "step": 6418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470975176453642e-05,
+      "loss": 0.5536,
+      "step": 6419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470735103471455e-05,
+      "loss": 0.6606,
+      "step": 6420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470495030489269e-05,
+      "loss": 0.6012,
+      "step": 6421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470254957507082e-05,
+      "loss": 0.5935,
+      "step": 6422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470014884524896e-05,
+      "loss": 0.7781,
+      "step": 6423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4697748115427093e-05,
+      "loss": 0.6968,
+      "step": 6424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4695347385605225e-05,
+      "loss": 0.4039,
+      "step": 6425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469294665578336e-05,
+      "loss": 0.6528,
+      "step": 6426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4690545925961496e-05,
+      "loss": 0.7574,
+      "step": 6427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468814519613963e-05,
+      "loss": 0.5556,
+      "step": 6428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468574446631776e-05,
+      "loss": 0.6349,
+      "step": 6429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46833437364959e-05,
+      "loss": 0.5894,
+      "step": 6430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468094300667403e-05,
+      "loss": 0.7083,
+      "step": 6431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467854227685217e-05,
+      "loss": 0.6218,
+      "step": 6432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46761415470303e-05,
+      "loss": 0.5415,
+      "step": 6433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467374081720843e-05,
+      "loss": 0.6972,
+      "step": 6434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4671340087386564e-05,
+      "loss": 0.7763,
+      "step": 6435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46689393575647e-05,
+      "loss": 0.5105,
+      "step": 6436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4666538627742834e-05,
+      "loss": 0.7215,
+      "step": 6437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4664137897920966e-05,
+      "loss": 0.4969,
+      "step": 6438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4661737168099104e-05,
+      "loss": 0.5112,
+      "step": 6439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.465933643827724e-05,
+      "loss": 0.8489,
+      "step": 6440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4656935708455375e-05,
+      "loss": 0.7596,
+      "step": 6441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4654534978633507e-05,
+      "loss": 0.7833,
+      "step": 6442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.465213424881164e-05,
+      "loss": 0.5946,
+      "step": 6443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464973351898978e-05,
+      "loss": 0.4847,
+      "step": 6444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464733278916791e-05,
+      "loss": 0.7133,
+      "step": 6445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464493205934604e-05,
+      "loss": 0.6324,
+      "step": 6446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464253132952417e-05,
+      "loss": 0.5678,
+      "step": 6447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.464013059970231e-05,
+      "loss": 0.7687,
+      "step": 6448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463772986988045e-05,
+      "loss": 0.6579,
+      "step": 6449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463532914005858e-05,
+      "loss": 0.6707,
+      "step": 6450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.463292841023671e-05,
+      "loss": 0.6311,
+      "step": 6451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4630527680414845e-05,
+      "loss": 0.7504,
+      "step": 6452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4628126950592983e-05,
+      "loss": 0.4942,
+      "step": 6453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4625726220771115e-05,
+      "loss": 0.6613,
+      "step": 6454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.462332549094925e-05,
+      "loss": 0.4593,
+      "step": 6455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4620924761127386e-05,
+      "loss": 0.5618,
+      "step": 6456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.461852403130552e-05,
+      "loss": 0.6785,
+      "step": 6457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4616123301483656e-05,
+      "loss": 0.7039,
+      "step": 6458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.461372257166179e-05,
+      "loss": 0.5322,
+      "step": 6459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.461132184183992e-05,
+      "loss": 0.4858,
+      "step": 6460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460892111201806e-05,
+      "loss": 0.6818,
+      "step": 6461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460652038219619e-05,
+      "loss": 0.5454,
+      "step": 6462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.460411965237432e-05,
+      "loss": 0.7493,
+      "step": 6463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4601718922552454e-05,
+      "loss": 0.6994,
+      "step": 6464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459931819273059e-05,
+      "loss": 0.4893,
+      "step": 6465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4596917462908724e-05,
+      "loss": 0.7663,
+      "step": 6466
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459451673308686e-05,
+      "loss": 0.5326,
+      "step": 6467
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4592116003264994e-05,
+      "loss": 0.748,
+      "step": 6468
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458971527344313e-05,
+      "loss": 0.5711,
+      "step": 6469
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4587314543621265e-05,
+      "loss": 0.4653,
+      "step": 6470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4584913813799397e-05,
+      "loss": 0.5366,
+      "step": 6471
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458251308397753e-05,
+      "loss": 0.7484,
+      "step": 6472
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458011235415567e-05,
+      "loss": 0.8419,
+      "step": 6473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45777116243338e-05,
+      "loss": 0.6278,
+      "step": 6474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457531089451193e-05,
+      "loss": 0.6377,
+      "step": 6475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457291016469007e-05,
+      "loss": 0.5193,
+      "step": 6476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45705094348682e-05,
+      "loss": 0.5162,
+      "step": 6477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456810870504634e-05,
+      "loss": 0.6497,
+      "step": 6478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456570797522447e-05,
+      "loss": 0.6875,
+      "step": 6479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45633072454026e-05,
+      "loss": 0.5288,
+      "step": 6480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4560906515580735e-05,
+      "loss": 0.6621,
+      "step": 6481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4558505785758873e-05,
+      "loss": 0.5931,
+      "step": 6482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4556105055937005e-05,
+      "loss": 0.5328,
+      "step": 6483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455370432611514e-05,
+      "loss": 0.5349,
+      "step": 6484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4551303596293276e-05,
+      "loss": 0.691,
+      "step": 6485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4548902866471414e-05,
+      "loss": 0.7072,
+      "step": 6486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4546502136649546e-05,
+      "loss": 0.7581,
+      "step": 6487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454410140682768e-05,
+      "loss": 0.6465,
+      "step": 6488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454170067700581e-05,
+      "loss": 0.6154,
+      "step": 6489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453929994718395e-05,
+      "loss": 0.8694,
+      "step": 6490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453689921736208e-05,
+      "loss": 0.4988,
+      "step": 6491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.453449848754021e-05,
+      "loss": 0.5792,
+      "step": 6492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4532097757718344e-05,
+      "loss": 0.6657,
+      "step": 6493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452969702789648e-05,
+      "loss": 0.575,
+      "step": 6494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452729629807462e-05,
+      "loss": 0.6698,
+      "step": 6495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452489556825275e-05,
+      "loss": 0.6486,
+      "step": 6496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4522494838430884e-05,
+      "loss": 0.6347,
+      "step": 6497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4520094108609016e-05,
+      "loss": 0.6743,
+      "step": 6498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4517693378787155e-05,
+      "loss": 0.5976,
+      "step": 6499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4515292648965286e-05,
+      "loss": 0.6602,
+      "step": 6500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451289191914342e-05,
+      "loss": 0.6798,
+      "step": 6501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.451049118932155e-05,
+      "loss": 0.523,
+      "step": 6502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4508090459499695e-05,
+      "loss": 0.4006,
+      "step": 6503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450568972967783e-05,
+      "loss": 0.6384,
+      "step": 6504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450328899985596e-05,
+      "loss": 0.5289,
+      "step": 6505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450088827003409e-05,
+      "loss": 0.7171,
+      "step": 6506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449848754021223e-05,
+      "loss": 0.4856,
+      "step": 6507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449608681039036e-05,
+      "loss": 0.6721,
+      "step": 6508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.449368608056849e-05,
+      "loss": 0.4868,
+      "step": 6509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4491285350746625e-05,
+      "loss": 0.582,
+      "step": 6510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4488884620924763e-05,
+      "loss": 0.659,
+      "step": 6511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4486483891102895e-05,
+      "loss": 0.5607,
+      "step": 6512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4484083161281034e-05,
+      "loss": 0.5223,
+      "step": 6513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4481682431459166e-05,
+      "loss": 0.5997,
+      "step": 6514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44792817016373e-05,
+      "loss": 0.5031,
+      "step": 6515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4476880971815436e-05,
+      "loss": 0.6178,
+      "step": 6516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.447448024199357e-05,
+      "loss": 0.5792,
+      "step": 6517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44720795121717e-05,
+      "loss": 0.6105,
+      "step": 6518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.446967878234983e-05,
+      "loss": 0.5309,
+      "step": 6519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.446727805252797e-05,
+      "loss": 0.369,
+      "step": 6520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44648773227061e-05,
+      "loss": 0.6167,
+      "step": 6521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.446247659288424e-05,
+      "loss": 0.6119,
+      "step": 6522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.446007586306237e-05,
+      "loss": 0.4972,
+      "step": 6523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.445767513324051e-05,
+      "loss": 0.5665,
+      "step": 6524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.445527440341864e-05,
+      "loss": 0.6264,
+      "step": 6525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4452873673596774e-05,
+      "loss": 0.6019,
+      "step": 6526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4450472943774906e-05,
+      "loss": 0.493,
+      "step": 6527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4448072213953045e-05,
+      "loss": 0.3839,
+      "step": 6528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4445671484131176e-05,
+      "loss": 0.6535,
+      "step": 6529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.444327075430931e-05,
+      "loss": 0.4533,
+      "step": 6530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.444087002448745e-05,
+      "loss": 0.282,
+      "step": 6531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443846929466558e-05,
+      "loss": 0.5829,
+      "step": 6532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443606856484372e-05,
+      "loss": 0.5141,
+      "step": 6533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443366783502185e-05,
+      "loss": 0.6588,
+      "step": 6534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.443126710519998e-05,
+      "loss": 0.6224,
+      "step": 6535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442886637537811e-05,
+      "loss": 0.5482,
+      "step": 6536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442646564555625e-05,
+      "loss": 0.4218,
+      "step": 6537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442406491573438e-05,
+      "loss": 0.9323,
+      "step": 6538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4421664185912515e-05,
+      "loss": 0.7078,
+      "step": 6539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441926345609065e-05,
+      "loss": 0.6504,
+      "step": 6540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441686272626879e-05,
+      "loss": 0.3159,
+      "step": 6541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4414461996446924e-05,
+      "loss": 0.4775,
+      "step": 6542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4412061266625056e-05,
+      "loss": 0.5447,
+      "step": 6543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440966053680319e-05,
+      "loss": 0.6622,
+      "step": 6544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4407259806981326e-05,
+      "loss": 0.4208,
+      "step": 6545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440485907715946e-05,
+      "loss": 0.3909,
+      "step": 6546
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440245834733759e-05,
+      "loss": 0.6254,
+      "step": 6547
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440005761751572e-05,
+      "loss": 0.2602,
+      "step": 6548
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439765688769386e-05,
+      "loss": 0.4934,
+      "step": 6549
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4395256157872e-05,
+      "loss": 0.523,
+      "step": 6550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439285542805013e-05,
+      "loss": 0.3136,
+      "step": 6551
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.439045469822826e-05,
+      "loss": 0.3576,
+      "step": 6552
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43880539684064e-05,
+      "loss": 0.453,
+      "step": 6553
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438565323858453e-05,
+      "loss": 0.7156,
+      "step": 6554
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4383252508762664e-05,
+      "loss": 0.501,
+      "step": 6555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4380851778940796e-05,
+      "loss": 0.6464,
+      "step": 6556
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4378451049118935e-05,
+      "loss": 0.7164,
+      "step": 6557
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437605031929707e-05,
+      "loss": 0.4529,
+      "step": 6558
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4373649589475205e-05,
+      "loss": 0.5264,
+      "step": 6559
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437124885965334e-05,
+      "loss": 0.6459,
+      "step": 6560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436884812983147e-05,
+      "loss": 0.3419,
+      "step": 6561
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436644740000961e-05,
+      "loss": 0.5572,
+      "step": 6562
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436404667018774e-05,
+      "loss": 0.4231,
+      "step": 6563
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436164594036587e-05,
+      "loss": 0.4731,
+      "step": 6564
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4359245210544e-05,
+      "loss": 0.422,
+      "step": 6565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435684448072214e-05,
+      "loss": 0.4527,
+      "step": 6566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435444375090027e-05,
+      "loss": 0.3125,
+      "step": 6567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.435204302107841e-05,
+      "loss": 0.7767,
+      "step": 6568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434964229125654e-05,
+      "loss": 0.4152,
+      "step": 6569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434724156143468e-05,
+      "loss": 0.4495,
+      "step": 6570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4344840831612814e-05,
+      "loss": 0.5405,
+      "step": 6571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4342440101790946e-05,
+      "loss": 0.5407,
+      "step": 6572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434003937196908e-05,
+      "loss": 0.3527,
+      "step": 6573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4337638642147216e-05,
+      "loss": 0.3424,
+      "step": 6574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433523791232535e-05,
+      "loss": 0.5764,
+      "step": 6575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.433283718250348e-05,
+      "loss": 0.5572,
+      "step": 6576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.433043645268162e-05,
+      "loss": 0.4137,
+      "step": 6577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432803572285975e-05,
+      "loss": 0.403,
+      "step": 6578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432563499303789e-05,
+      "loss": 0.3036,
+      "step": 6579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432323426321602e-05,
+      "loss": 0.3726,
+      "step": 6580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.432083353339415e-05,
+      "loss": 0.3045,
+      "step": 6581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4318432803572284e-05,
+      "loss": 0.9015,
+      "step": 6582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.431603207375042e-05,
+      "loss": 0.2417,
+      "step": 6583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4313631343928554e-05,
+      "loss": 0.3449,
+      "step": 6584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4311230614106686e-05,
+      "loss": 0.2791,
+      "step": 6585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4308829884284825e-05,
+      "loss": 0.2936,
+      "step": 6586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.430642915446296e-05,
+      "loss": 0.3056,
+      "step": 6587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4304028424641095e-05,
+      "loss": 0.1215,
+      "step": 6588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.430162769481923e-05,
+      "loss": 0.1153,
+      "step": 6589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.429922696499736e-05,
+      "loss": 0.0528,
+      "step": 6590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42968262351755e-05,
+      "loss": 0.2903,
+      "step": 6591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.429442550535363e-05,
+      "loss": 0.3276,
+      "step": 6592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.429202477553176e-05,
+      "loss": 0.2579,
+      "step": 6593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428962404570989e-05,
+      "loss": 0.2335,
+      "step": 6594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428722331588803e-05,
+      "loss": 0.6195,
+      "step": 6595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428482258606617e-05,
+      "loss": 0.3246,
+      "step": 6596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42824218562443e-05,
+      "loss": 0.6132,
+      "step": 6597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.428002112642243e-05,
+      "loss": 0.2461,
+      "step": 6598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4277620396600565e-05,
+      "loss": 0.2492,
+      "step": 6599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4275219666778704e-05,
+      "loss": 0.0768,
+      "step": 6600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4272818936956836e-05,
+      "loss": 0.4911,
+      "step": 6601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.427041820713497e-05,
+      "loss": 0.2596,
+      "step": 6602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.42680174773131e-05,
+      "loss": 0.1065,
+      "step": 6603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4265616747491244e-05,
+      "loss": 0.4772,
+      "step": 6604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4263216017669376e-05,
+      "loss": 0.8901,
+      "step": 6605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.426081528784751e-05,
+      "loss": 0.3553,
+      "step": 6606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425841455802564e-05,
+      "loss": 0.078,
+      "step": 6607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425601382820378e-05,
+      "loss": 0.1109,
+      "step": 6608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425361309838191e-05,
+      "loss": 0.123,
+      "step": 6609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.425121236856004e-05,
+      "loss": 0.5093,
+      "step": 6610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4248811638738174e-05,
+      "loss": 0.303,
+      "step": 6611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424641090891631e-05,
+      "loss": 0.6215,
+      "step": 6612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424401017909445e-05,
+      "loss": 0.2955,
+      "step": 6613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.424160944927258e-05,
+      "loss": 0.5415,
+      "step": 6614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4239208719450715e-05,
+      "loss": 0.5211,
+      "step": 6615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4236807989628846e-05,
+      "loss": 0.136,
+      "step": 6616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4234407259806985e-05,
+      "loss": 0.3281,
+      "step": 6617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423200652998512e-05,
+      "loss": 0.1001,
+      "step": 6618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422960580016325e-05,
+      "loss": 0.5793,
+      "step": 6619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422720507034139e-05,
+      "loss": 0.7325,
+      "step": 6620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422480434051952e-05,
+      "loss": 0.3055,
+      "step": 6621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422240361069766e-05,
+      "loss": 0.096,
+      "step": 6622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422000288087579e-05,
+      "loss": 0.501,
+      "step": 6623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421760215105392e-05,
+      "loss": 0.3785,
+      "step": 6624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421520142123206e-05,
+      "loss": 0.1785,
+      "step": 6625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421280069141019e-05,
+      "loss": 0.1345,
+      "step": 6626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421039996158832e-05,
+      "loss": 0.1021,
+      "step": 6627
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4207999231766455e-05,
+      "loss": 0.1602,
+      "step": 6628
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4205598501944594e-05,
+      "loss": 0.072,
+      "step": 6629
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4203197772122725e-05,
+      "loss": 0.3363,
+      "step": 6630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420079704230086e-05,
+      "loss": 0.2762,
+      "step": 6631
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4198396312478996e-05,
+      "loss": 0.1132,
+      "step": 6632
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419599558265713e-05,
+      "loss": 0.0741,
+      "step": 6633
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4193594852835266e-05,
+      "loss": 0.0696,
+      "step": 6634
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41911941230134e-05,
+      "loss": 0.0319,
+      "step": 6635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418879339319153e-05,
+      "loss": 0.074,
+      "step": 6636
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418639266336967e-05,
+      "loss": 0.0297,
+      "step": 6637
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41839919335478e-05,
+      "loss": 0.0933,
+      "step": 6638
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.418159120372593e-05,
+      "loss": 0.0348,
+      "step": 6639
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4179190473904064e-05,
+      "loss": 0.3328,
+      "step": 6640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41767897440822e-05,
+      "loss": 0.0626,
+      "step": 6641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417438901426034e-05,
+      "loss": 0.0737,
+      "step": 6642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417198828443847e-05,
+      "loss": 0.3283,
+      "step": 6643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4169587554616605e-05,
+      "loss": 0.0563,
+      "step": 6644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4167186824794736e-05,
+      "loss": 0.3609,
+      "step": 6645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4164786094972875e-05,
+      "loss": 0.0573,
+      "step": 6646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.416238536515101e-05,
+      "loss": 0.3091,
+      "step": 6647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415998463532914e-05,
+      "loss": 0.0261,
+      "step": 6648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415758390550727e-05,
+      "loss": 0.0393,
+      "step": 6649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4155183175685416e-05,
+      "loss": 0.0461,
+      "step": 6650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415278244586355e-05,
+      "loss": 0.0362,
+      "step": 6651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.415038171604168e-05,
+      "loss": 0.056,
+      "step": 6652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414798098621981e-05,
+      "loss": 0.0225,
+      "step": 6653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414558025639795e-05,
+      "loss": 0.0529,
+      "step": 6654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414317952657608e-05,
+      "loss": 0.0441,
+      "step": 6655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414077879675421e-05,
+      "loss": 0.0199,
+      "step": 6656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4138378066932345e-05,
+      "loss": 0.4226,
+      "step": 6657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4135977337110484e-05,
+      "loss": 0.0427,
+      "step": 6658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.413357660728862e-05,
+      "loss": 0.0177,
+      "step": 6659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4131175877466754e-05,
+      "loss": 0.0206,
+      "step": 6660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4128775147644886e-05,
+      "loss": 0.0117,
+      "step": 6661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.412637441782302e-05,
+      "loss": 0.0216,
+      "step": 6662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4123973688001156e-05,
+      "loss": 0.0191,
+      "step": 6663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.412157295817929e-05,
+      "loss": 0.0107,
+      "step": 6664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411917222835742e-05,
+      "loss": 0.0162,
+      "step": 6665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411677149853555e-05,
+      "loss": 0.8124,
+      "step": 6666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411437076871369e-05,
+      "loss": 0.0102,
+      "step": 6667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.411197003889183e-05,
+      "loss": 0.498,
+      "step": 6668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410956930906996e-05,
+      "loss": 0.0101,
+      "step": 6669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410716857924809e-05,
+      "loss": 0.0072,
+      "step": 6670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410476784942623e-05,
+      "loss": 0.0096,
+      "step": 6671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410236711960436e-05,
+      "loss": 0.0122,
+      "step": 6672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4099966389782495e-05,
+      "loss": 0.01,
+      "step": 6673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4097565659960626e-05,
+      "loss": 0.0102,
+      "step": 6674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4095164930138765e-05,
+      "loss": 0.0067,
+      "step": 6675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40927642003169e-05,
+      "loss": 0.0044,
+      "step": 6676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4090363470495035e-05,
+      "loss": 0.0091,
+      "step": 6677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408796274067317e-05,
+      "loss": 0.0033,
+      "step": 6678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40855620108513e-05,
+      "loss": 0.0041,
+      "step": 6679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408316128102944e-05,
+      "loss": 0.01,
+      "step": 6680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408076055120757e-05,
+      "loss": 0.0054,
+      "step": 6681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40783598213857e-05,
+      "loss": 0.8352,
+      "step": 6682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407595909156383e-05,
+      "loss": 0.0032,
+      "step": 6683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407355836174197e-05,
+      "loss": 0.0047,
+      "step": 6684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40711576319201e-05,
+      "loss": 1.2961,
+      "step": 6685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406875690209824e-05,
+      "loss": 0.0052,
+      "step": 6686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4066356172276374e-05,
+      "loss": 0.0034,
+      "step": 6687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406395544245451e-05,
+      "loss": 0.0068,
+      "step": 6688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4061554712632644e-05,
+      "loss": 0.004,
+      "step": 6689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4059153982810776e-05,
+      "loss": 0.0038,
+      "step": 6690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405675325298891e-05,
+      "loss": 0.6281,
+      "step": 6691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4054352523167046e-05,
+      "loss": 0.0038,
+      "step": 6692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405195179334518e-05,
+      "loss": 0.0054,
+      "step": 6693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404955106352331e-05,
+      "loss": 0.0024,
+      "step": 6694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404715033370144e-05,
+      "loss": 0.7519,
+      "step": 6695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404474960387958e-05,
+      "loss": 0.0038,
+      "step": 6696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404234887405772e-05,
+      "loss": 0.0042,
+      "step": 6697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403994814423585e-05,
+      "loss": 0.0047,
+      "step": 6698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403754741441398e-05,
+      "loss": 0.6878,
+      "step": 6699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4035146684592114e-05,
+      "loss": 0.0045,
+      "step": 6700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403274595477025e-05,
+      "loss": 0.0049,
+      "step": 6701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4030345224948385e-05,
+      "loss": 0.0042,
+      "step": 6702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4027944495126516e-05,
+      "loss": 0.0044,
+      "step": 6703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4025543765304655e-05,
+      "loss": 0.0052,
+      "step": 6704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4023143035482793e-05,
+      "loss": 0.0041,
+      "step": 6705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4020742305660925e-05,
+      "loss": 0.0055,
+      "step": 6706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401834157583906e-05,
+      "loss": 0.7267,
+      "step": 6707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401594084601719e-05,
+      "loss": 0.6971,
+      "step": 6708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401354011619533e-05,
+      "loss": 1.3214,
+      "step": 6709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401113938637346e-05,
+      "loss": 0.0045,
+      "step": 6710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400873865655159e-05,
+      "loss": 0.0054,
+      "step": 6711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400633792672972e-05,
+      "loss": 0.6504,
+      "step": 6712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400393719690786e-05,
+      "loss": 0.0067,
+      "step": 6713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4001536467086e-05,
+      "loss": 0.0056,
+      "step": 6714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399913573726413e-05,
+      "loss": 0.0053,
+      "step": 6715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3996735007442264e-05,
+      "loss": 0.0051,
+      "step": 6716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39943342776204e-05,
+      "loss": 0.0058,
+      "step": 6717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3991933547798534e-05,
+      "loss": 0.0042,
+      "step": 6718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3989532817976666e-05,
+      "loss": 0.0058,
+      "step": 6719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39871320881548e-05,
+      "loss": 0.0042,
+      "step": 6720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3984731358332936e-05,
+      "loss": 0.0055,
+      "step": 6721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.398233062851107e-05,
+      "loss": 0.0049,
+      "step": 6722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3979929898689207e-05,
+      "loss": 0.0037,
+      "step": 6723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397752916886734e-05,
+      "loss": 0.0045,
+      "step": 6724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397512843904547e-05,
+      "loss": 0.0053,
+      "step": 6725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397272770922361e-05,
+      "loss": 0.005,
+      "step": 6726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.397032697940174e-05,
+      "loss": 0.0058,
+      "step": 6727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.396792624957987e-05,
+      "loss": 0.0055,
+      "step": 6728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3965525519758004e-05,
+      "loss": 0.0049,
+      "step": 6729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.396312478993614e-05,
+      "loss": 0.0048,
+      "step": 6730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3960724060114274e-05,
+      "loss": 0.0033,
+      "step": 6731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.395832333029241e-05,
+      "loss": 0.0057,
+      "step": 6732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3955922600470545e-05,
+      "loss": 0.0047,
+      "step": 6733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3953521870648683e-05,
+      "loss": 0.0043,
+      "step": 6734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3951121140826815e-05,
+      "loss": 0.003,
+      "step": 6735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394872041100495e-05,
+      "loss": 0.0033,
+      "step": 6736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394631968118308e-05,
+      "loss": 0.0044,
+      "step": 6737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394391895136122e-05,
+      "loss": 0.0037,
+      "step": 6738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.394151822153935e-05,
+      "loss": 0.003,
+      "step": 6739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393911749171748e-05,
+      "loss": 0.0042,
+      "step": 6740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393671676189562e-05,
+      "loss": 0.0032,
+      "step": 6741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393431603207375e-05,
+      "loss": 0.003,
+      "step": 6742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393191530225189e-05,
+      "loss": 0.0036,
+      "step": 6743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392951457243002e-05,
+      "loss": 0.0037,
+      "step": 6744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3927113842608154e-05,
+      "loss": 0.0039,
+      "step": 6745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3924713112786285e-05,
+      "loss": 0.0038,
+      "step": 6746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3922312382964424e-05,
+      "loss": 0.0028,
+      "step": 6747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3919911653142556e-05,
+      "loss": 0.0034,
+      "step": 6748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391751092332069e-05,
+      "loss": 0.0022,
+      "step": 6749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3915110193498826e-05,
+      "loss": 0.0026,
+      "step": 6750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3912709463676965e-05,
+      "loss": 0.0022,
+      "step": 6751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3910308733855096e-05,
+      "loss": 0.5431,
+      "step": 6752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390790800403323e-05,
+      "loss": 0.0025,
+      "step": 6753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390550727421136e-05,
+      "loss": 0.0025,
+      "step": 6754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.39031065443895e-05,
+      "loss": 0.7669,
+      "step": 6755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390070581456763e-05,
+      "loss": 0.0027,
+      "step": 6756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.0025,
+      "step": 6757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3895904354923894e-05,
+      "loss": 0.8139,
+      "step": 6758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389350362510203e-05,
+      "loss": 0.0023,
+      "step": 6759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389110289528017e-05,
+      "loss": 0.0029,
+      "step": 6760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38887021654583e-05,
+      "loss": 0.0019,
+      "step": 6761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3886301435636435e-05,
+      "loss": 0.0029,
+      "step": 6762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388390070581457e-05,
+      "loss": 0.7487,
+      "step": 6763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3881499975992705e-05,
+      "loss": 0.0032,
+      "step": 6764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387909924617084e-05,
+      "loss": 0.0028,
+      "step": 6765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387669851634897e-05,
+      "loss": 0.0023,
+      "step": 6766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38742977865271e-05,
+      "loss": 0.0024,
+      "step": 6767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387189705670524e-05,
+      "loss": 0.0026,
+      "step": 6768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386949632688338e-05,
+      "loss": 0.0029,
+      "step": 6769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386709559706151e-05,
+      "loss": 0.003,
+      "step": 6770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386469486723964e-05,
+      "loss": 0.003,
+      "step": 6771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386229413741778e-05,
+      "loss": 0.5324,
+      "step": 6772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385989340759591e-05,
+      "loss": 0.0029,
+      "step": 6773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3857492677774044e-05,
+      "loss": 0.0028,
+      "step": 6774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3855091947952175e-05,
+      "loss": 0.0031,
+      "step": 6775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3852691218130314e-05,
+      "loss": 0.0029,
+      "step": 6776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3850290488308446e-05,
+      "loss": 0.0022,
+      "step": 6777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3847889758486584e-05,
+      "loss": 0.0057,
+      "step": 6778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3845489028664716e-05,
+      "loss": 0.7498,
+      "step": 6779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384308829884285e-05,
+      "loss": 0.0031,
+      "step": 6780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3840687569020986e-05,
+      "loss": 1.3159,
+      "step": 6781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383828683919912e-05,
+      "loss": 0.0031,
+      "step": 6782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383588610937725e-05,
+      "loss": 0.0027,
+      "step": 6783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383348537955538e-05,
+      "loss": 0.0031,
+      "step": 6784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383108464973352e-05,
+      "loss": 0.0032,
+      "step": 6785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382868391991165e-05,
+      "loss": 0.0029,
+      "step": 6786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382628319008979e-05,
+      "loss": 0.7146,
+      "step": 6787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382388246026792e-05,
+      "loss": 0.0039,
+      "step": 6788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.382148173044606e-05,
+      "loss": 0.0031,
+      "step": 6789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.381908100062419e-05,
+      "loss": 0.0037,
+      "step": 6790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3816680270802325e-05,
+      "loss": 0.0037,
+      "step": 6791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3814279540980457e-05,
+      "loss": 0.0032,
+      "step": 6792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3811878811158595e-05,
+      "loss": 0.0032,
+      "step": 6793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380947808133673e-05,
+      "loss": 0.0044,
+      "step": 6794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380707735151486e-05,
+      "loss": 0.0039,
+      "step": 6795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3804676621693e-05,
+      "loss": 0.7264,
+      "step": 6796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.380227589187113e-05,
+      "loss": 0.658,
+      "step": 6797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.379987516204927e-05,
+      "loss": 0.646,
+      "step": 6798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37974744322274e-05,
+      "loss": 0.0054,
+      "step": 6799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.379507370240553e-05,
+      "loss": 0.0047,
+      "step": 6800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.379267297258367e-05,
+      "loss": 0.0068,
+      "step": 6801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37902722427618e-05,
+      "loss": 0.015,
+      "step": 6802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3787871512939934e-05,
+      "loss": 0.4237,
+      "step": 6803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3785470783118065e-05,
+      "loss": 0.008,
+      "step": 6804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3783070053296204e-05,
+      "loss": 0.0117,
+      "step": 6805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.378066932347434e-05,
+      "loss": 0.4848,
+      "step": 6806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3778268593652474e-05,
+      "loss": 0.0184,
+      "step": 6807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3775867863830606e-05,
+      "loss": 0.0152,
+      "step": 6808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.377346713400874e-05,
+      "loss": 0.0104,
+      "step": 6809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3771066404186876e-05,
+      "loss": 0.4279,
+      "step": 6810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376866567436501e-05,
+      "loss": 0.4933,
+      "step": 6811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376626494454314e-05,
+      "loss": 0.0151,
+      "step": 6812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376386421472127e-05,
+      "loss": 0.0098,
+      "step": 6813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.376146348489941e-05,
+      "loss": 0.0112,
+      "step": 6814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375906275507755e-05,
+      "loss": 0.0065,
+      "step": 6815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375666202525568e-05,
+      "loss": 0.8242,
+      "step": 6816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375426129543381e-05,
+      "loss": 0.0114,
+      "step": 6817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.375186056561195e-05,
+      "loss": 0.0057,
+      "step": 6818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.374945983579008e-05,
+      "loss": 0.0045,
+      "step": 6819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3747059105968215e-05,
+      "loss": 0.0044,
+      "step": 6820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3744658376146347e-05,
+      "loss": 0.7368,
+      "step": 6821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3742257646324485e-05,
+      "loss": 1.4565,
+      "step": 6822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373985691650262e-05,
+      "loss": 0.7687,
+      "step": 6823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3737456186680756e-05,
+      "loss": 0.0041,
+      "step": 6824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373505545685889e-05,
+      "loss": 1.4484,
+      "step": 6825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373265472703702e-05,
+      "loss": 0.0037,
+      "step": 6826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373025399721516e-05,
+      "loss": 2.0849,
+      "step": 6827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372785326739329e-05,
+      "loss": 0.73,
+      "step": 6828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372545253757142e-05,
+      "loss": 0.0044,
+      "step": 6829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372305180774955e-05,
+      "loss": 0.7342,
+      "step": 6830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372065107792769e-05,
+      "loss": 0.004,
+      "step": 6831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3718250348105823e-05,
+      "loss": 0.0043,
+      "step": 6832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371584961828396e-05,
+      "loss": 0.0043,
+      "step": 6833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3713448888462094e-05,
+      "loss": 0.0038,
+      "step": 6834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371104815864023e-05,
+      "loss": 0.2954,
+      "step": 6835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3708647428818364e-05,
+      "loss": 0.0052,
+      "step": 6836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3706246698996496e-05,
+      "loss": 0.0042,
+      "step": 6837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.370384596917463e-05,
+      "loss": 1.4252,
+      "step": 6838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3701445239352766e-05,
+      "loss": 0.6572,
+      "step": 6839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36990445095309e-05,
+      "loss": 0.006,
+      "step": 6840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369664377970903e-05,
+      "loss": 0.0043,
+      "step": 6841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.369424304988717e-05,
+      "loss": 0.5742,
+      "step": 6842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36918423200653e-05,
+      "loss": 0.0062,
+      "step": 6843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368944159024344e-05,
+      "loss": 0.0044,
+      "step": 6844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368704086042157e-05,
+      "loss": 0.6616,
+      "step": 6845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.36846401305997e-05,
+      "loss": 0.6607,
+      "step": 6846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3682239400777834e-05,
+      "loss": 0.558,
+      "step": 6847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.367983867095597e-05,
+      "loss": 0.5893,
+      "step": 6848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3677437941134105e-05,
+      "loss": 0.6167,
+      "step": 6849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3675037211312237e-05,
+      "loss": 0.0071,
+      "step": 6850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3672636481490375e-05,
+      "loss": 0.0109,
+      "step": 6851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3670235751668514e-05,
+      "loss": 1.6789,
+      "step": 6852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3667835021846645e-05,
+      "loss": 0.5318,
+      "step": 6853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366543429202478e-05,
+      "loss": 0.5305,
+      "step": 6854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366303356220291e-05,
+      "loss": 0.0105,
+      "step": 6855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.366063283238105e-05,
+      "loss": 0.5394,
+      "step": 6856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365823210255918e-05,
+      "loss": 0.0155,
+      "step": 6857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365583137273731e-05,
+      "loss": 0.5154,
+      "step": 6858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365343064291544e-05,
+      "loss": 0.0326,
+      "step": 6859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.365102991309358e-05,
+      "loss": 1.0185,
+      "step": 6860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.364862918327172e-05,
+      "loss": 0.4689,
+      "step": 6861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.364622845344985e-05,
+      "loss": 0.0196,
+      "step": 6862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3643827723627984e-05,
+      "loss": 0.0221,
+      "step": 6863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3641426993806116e-05,
+      "loss": 0.4634,
+      "step": 6864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3639026263984254e-05,
+      "loss": 0.0161,
+      "step": 6865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3636625534162386e-05,
+      "loss": 0.5046,
+      "step": 6866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.363422480434052e-05,
+      "loss": 0.027,
+      "step": 6867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3631824074518656e-05,
+      "loss": 0.009,
+      "step": 6868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3629423344696795e-05,
+      "loss": 0.0211,
+      "step": 6869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362702261487493e-05,
+      "loss": 0.4291,
+      "step": 6870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362462188505306e-05,
+      "loss": 0.4259,
+      "step": 6871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.362222115523119e-05,
+      "loss": 0.0281,
+      "step": 6872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361982042540933e-05,
+      "loss": 0.4437,
+      "step": 6873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361741969558746e-05,
+      "loss": 0.0249,
+      "step": 6874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361501896576559e-05,
+      "loss": 0.0298,
+      "step": 6875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3612618235943724e-05,
+      "loss": 0.0107,
+      "step": 6876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.361021750612186e-05,
+      "loss": 0.4511,
+      "step": 6877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36078167763e-05,
+      "loss": 0.0237,
+      "step": 6878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.360541604647813e-05,
+      "loss": 0.0265,
+      "step": 6879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3603015316656265e-05,
+      "loss": 0.3802,
+      "step": 6880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36006145868344e-05,
+      "loss": 0.4092,
+      "step": 6881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3598213857012535e-05,
+      "loss": 0.0128,
+      "step": 6882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359581312719067e-05,
+      "loss": 0.0247,
+      "step": 6883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35934123973688e-05,
+      "loss": 0.0209,
+      "step": 6884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359101166754694e-05,
+      "loss": 0.0221,
+      "step": 6885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358861093772507e-05,
+      "loss": 0.0179,
+      "step": 6886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35862102079032e-05,
+      "loss": 0.021,
+      "step": 6887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358380947808134e-05,
+      "loss": 0.0225,
+      "step": 6888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358140874825947e-05,
+      "loss": 0.0274,
+      "step": 6889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357900801843761e-05,
+      "loss": 0.979,
+      "step": 6890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357660728861574e-05,
+      "loss": 0.511,
+      "step": 6891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3574206558793874e-05,
+      "loss": 0.0154,
+      "step": 6892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3571805828972006e-05,
+      "loss": 0.0131,
+      "step": 6893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3569405099150144e-05,
+      "loss": 0.0107,
+      "step": 6894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3567004369328276e-05,
+      "loss": 0.0038,
+      "step": 6895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.356460363950641e-05,
+      "loss": 0.0115,
+      "step": 6896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3562202909684546e-05,
+      "loss": 0.0086,
+      "step": 6897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3559802179862685e-05,
+      "loss": 0.01,
+      "step": 6898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355740145004082e-05,
+      "loss": 0.6193,
+      "step": 6899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355500072021895e-05,
+      "loss": 0.0074,
+      "step": 6900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355259999039708e-05,
+      "loss": 0.006,
+      "step": 6901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.355019926057522e-05,
+      "loss": 0.0056,
+      "step": 6902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354779853075335e-05,
+      "loss": 0.0064,
+      "step": 6903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354539780093148e-05,
+      "loss": 0.0082,
+      "step": 6904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3542997071109614e-05,
+      "loss": 0.0083,
+      "step": 6905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354059634128775e-05,
+      "loss": 0.0041,
+      "step": 6906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353819561146589e-05,
+      "loss": 0.0041,
+      "step": 6907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353579488164402e-05,
+      "loss": 0.0075,
+      "step": 6908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3533394151822155e-05,
+      "loss": 0.0064,
+      "step": 6909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.353099342200029e-05,
+      "loss": 0.002,
+      "step": 6910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3528592692178425e-05,
+      "loss": 0.0037,
+      "step": 6911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352619196235656e-05,
+      "loss": 0.5576,
+      "step": 6912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352379123253469e-05,
+      "loss": 0.0055,
+      "step": 6913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.352139050271282e-05,
+      "loss": 0.0034,
+      "step": 6914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3518989772890966e-05,
+      "loss": 0.0038,
+      "step": 6915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35165890430691e-05,
+      "loss": 0.0042,
+      "step": 6916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351418831324723e-05,
+      "loss": 0.0031,
+      "step": 6917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.351178758342536e-05,
+      "loss": 0.8211,
+      "step": 6918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.35093868536035e-05,
+      "loss": 0.0035,
+      "step": 6919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350698612378163e-05,
+      "loss": 0.0021,
+      "step": 6920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3504585393959764e-05,
+      "loss": 0.0022,
+      "step": 6921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3502184664137896e-05,
+      "loss": 0.0025,
+      "step": 6922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3499783934316034e-05,
+      "loss": 0.0031,
+      "step": 6923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349738320449417e-05,
+      "loss": 0.6883,
+      "step": 6924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3494982474672305e-05,
+      "loss": 0.0042,
+      "step": 6925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3492581744850436e-05,
+      "loss": 0.0029,
+      "step": 6926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.349018101502857e-05,
+      "loss": 0.0037,
+      "step": 6927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348778028520671e-05,
+      "loss": 0.003,
+      "step": 6928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348537955538484e-05,
+      "loss": 0.0026,
+      "step": 6929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.348297882556297e-05,
+      "loss": 0.0027,
+      "step": 6930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34805780957411e-05,
+      "loss": 0.0034,
+      "step": 6931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.347817736591924e-05,
+      "loss": 0.0038,
+      "step": 6932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.347577663609738e-05,
+      "loss": 0.0028,
+      "step": 6933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.347337590627551e-05,
+      "loss": 0.0038,
+      "step": 6934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.347097517645364e-05,
+      "loss": 0.002,
+      "step": 6935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346857444663178e-05,
+      "loss": 0.0021,
+      "step": 6936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346617371680991e-05,
+      "loss": 0.0029,
+      "step": 6937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3463772986988045e-05,
+      "loss": 0.7799,
+      "step": 6938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.346137225716618e-05,
+      "loss": 0.5454,
+      "step": 6939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3458971527344315e-05,
+      "loss": 0.002,
+      "step": 6940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.345657079752245e-05,
+      "loss": 0.0021,
+      "step": 6941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.345417006770058e-05,
+      "loss": 0.0025,
+      "step": 6942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.345176933787872e-05,
+      "loss": 0.0019,
+      "step": 6943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344936860805685e-05,
+      "loss": 0.0024,
+      "step": 6944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344696787823499e-05,
+      "loss": 0.0023,
+      "step": 6945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344456714841312e-05,
+      "loss": 0.0024,
+      "step": 6946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.344216641859125e-05,
+      "loss": 0.002,
+      "step": 6947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343976568876938e-05,
+      "loss": 0.0019,
+      "step": 6948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.343736495894752e-05,
+      "loss": 0.0024,
+      "step": 6949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3434964229125654e-05,
+      "loss": 0.0025,
+      "step": 6950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3432563499303786e-05,
+      "loss": 0.0019,
+      "step": 6951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3430162769481924e-05,
+      "loss": 0.0022,
+      "step": 6952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342776203966006e-05,
+      "loss": 0.0019,
+      "step": 6953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3425361309838194e-05,
+      "loss": 0.0021,
+      "step": 6954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3422960580016326e-05,
+      "loss": 0.0021,
+      "step": 6955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.342055985019446e-05,
+      "loss": 0.002,
+      "step": 6956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34181591203726e-05,
+      "loss": 0.0018,
+      "step": 6957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341575839055073e-05,
+      "loss": 0.0018,
+      "step": 6958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341335766072886e-05,
+      "loss": 0.0017,
+      "step": 6959
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9844075591003808,
+      "eval_loss": 0.08995544165372849,
+      "eval_runtime": 265.1723,
+      "eval_samples_per_second": 52.483,
+      "eval_steps_per_second": 1.64,
+      "step": 6959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.341095693090699e-05,
+      "loss": 0.0016,
+      "step": 6960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340855620108513e-05,
+      "loss": 0.0021,
+      "step": 6961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340615547126327e-05,
+      "loss": 0.0017,
+      "step": 6962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.34037547414414e-05,
+      "loss": 0.002,
+      "step": 6963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.340135401161953e-05,
+      "loss": 0.002,
+      "step": 6964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339895328179767e-05,
+      "loss": 0.0017,
+      "step": 6965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.33965525519758e-05,
+      "loss": 0.0018,
+      "step": 6966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3394151822153935e-05,
+      "loss": 0.0018,
+      "step": 6967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339175109233207e-05,
+      "loss": 0.4502,
+      "step": 6968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3389350362510205e-05,
+      "loss": 0.0019,
+      "step": 6969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3386949632688344e-05,
+      "loss": 0.0017,
+      "step": 6970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3384548902866476e-05,
+      "loss": 0.0018,
+      "step": 6971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.338214817304461e-05,
+      "loss": 0.0016,
+      "step": 6972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337974744322274e-05,
+      "loss": 0.0016,
+      "step": 6973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337734671340088e-05,
+      "loss": 1.1734,
+      "step": 6974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337494598357901e-05,
+      "loss": 0.0016,
+      "step": 6975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337254525375714e-05,
+      "loss": 0.0017,
+      "step": 6976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337014452393527e-05,
+      "loss": 0.0017,
+      "step": 6977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336774379411341e-05,
+      "loss": 0.8499,
+      "step": 6978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336534306429155e-05,
+      "loss": 0.0017,
+      "step": 6979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.336294233446968e-05,
+      "loss": 0.0017,
+      "step": 6980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3360541604647814e-05,
+      "loss": 0.0019,
+      "step": 6981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335814087482595e-05,
+      "loss": 0.0016,
+      "step": 6982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3355740145004084e-05,
+      "loss": 0.0016,
+      "step": 6983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3353339415182216e-05,
+      "loss": 0.798,
+      "step": 6984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.335093868536035e-05,
+      "loss": 0.7901,
+      "step": 6985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.334853795553849e-05,
+      "loss": 1.2733,
+      "step": 6986
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.334613722571662e-05,
+      "loss": 0.0016,
+      "step": 6987
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.334373649589476e-05,
+      "loss": 0.0019,
+      "step": 6988
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.334133576607289e-05,
+      "loss": 0.002,
+      "step": 6989
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.333893503625102e-05,
+      "loss": 0.0022,
+      "step": 6990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.333653430642916e-05,
+      "loss": 0.0016,
+      "step": 6991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.333413357660729e-05,
+      "loss": 0.0021,
+      "step": 6992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.333173284678542e-05,
+      "loss": 0.0017,
+      "step": 6993
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3329332116963555e-05,
+      "loss": 0.0021,
+      "step": 6994
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.332693138714169e-05,
+      "loss": 0.0021,
+      "step": 6995
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3324530657319825e-05,
+      "loss": 0.0021,
+      "step": 6996
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3322129927497964e-05,
+      "loss": 0.0029,
+      "step": 6997
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3319729197676095e-05,
+      "loss": 0.0019,
+      "step": 6998
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3317328467854234e-05,
+      "loss": 0.0024,
+      "step": 6999
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3314927738032366e-05,
+      "loss": 0.6818,
+      "step": 7000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.33125270082105e-05,
+      "loss": 0.0024,
+      "step": 7001
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.331012627838863e-05,
+      "loss": 0.0024,
+      "step": 7002
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330772554856677e-05,
+      "loss": 0.0086,
+      "step": 7003
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.33053248187449e-05,
+      "loss": 0.0024,
+      "step": 7004
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330292408892303e-05,
+      "loss": 0.002,
+      "step": 7005
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330052335910116e-05,
+      "loss": 0.002,
+      "step": 7006
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32981226292793e-05,
+      "loss": 0.0022,
+      "step": 7007
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329572189945744e-05,
+      "loss": 0.0024,
+      "step": 7008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.329332116963557e-05,
+      "loss": 0.0028,
+      "step": 7009
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3290920439813704e-05,
+      "loss": 0.478,
+      "step": 7010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3288519709991836e-05,
+      "loss": 0.0022,
+      "step": 7011
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3286118980169974e-05,
+      "loss": 0.5323,
+      "step": 7012
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3283718250348106e-05,
+      "loss": 0.0025,
+      "step": 7013
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328131752052624e-05,
+      "loss": 0.0029,
+      "step": 7014
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327891679070437e-05,
+      "loss": 0.0025,
+      "step": 7015
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3276516060882515e-05,
+      "loss": 0.002,
+      "step": 7016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327411533106065e-05,
+      "loss": 0.0718,
+      "step": 7017
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327171460123878e-05,
+      "loss": 0.0026,
+      "step": 7018
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326931387141691e-05,
+      "loss": 0.0021,
+      "step": 7019
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326691314159505e-05,
+      "loss": 0.0028,
+      "step": 7020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326451241177318e-05,
+      "loss": 0.0028,
+      "step": 7021
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326211168195131e-05,
+      "loss": 0.0023,
+      "step": 7022
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3259710952129445e-05,
+      "loss": 0.0024,
+      "step": 7023
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325731022230758e-05,
+      "loss": 0.0025,
+      "step": 7024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325490949248572e-05,
+      "loss": 0.002,
+      "step": 7025
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3252508762663854e-05,
+      "loss": 0.0027,
+      "step": 7026
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3250108032841985e-05,
+      "loss": 0.6544,
+      "step": 7027
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324770730302012e-05,
+      "loss": 0.0025,
+      "step": 7028
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3245306573198256e-05,
+      "loss": 0.002,
+      "step": 7029
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324290584337639e-05,
+      "loss": 0.0023,
+      "step": 7030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324050511355452e-05,
+      "loss": 0.0024,
+      "step": 7031
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323810438373265e-05,
+      "loss": 0.0019,
+      "step": 7032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323570365391079e-05,
+      "loss": 0.0023,
+      "step": 7033
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323330292408893e-05,
+      "loss": 0.0019,
+      "step": 7034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323090219426706e-05,
+      "loss": 0.0021,
+      "step": 7035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322850146444519e-05,
+      "loss": 0.4446,
+      "step": 7036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322610073462333e-05,
+      "loss": 0.0021,
+      "step": 7037
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322370000480146e-05,
+      "loss": 0.0025,
+      "step": 7038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3221299274979594e-05,
+      "loss": 0.0026,
+      "step": 7039
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3218898545157726e-05,
+      "loss": 0.0022,
+      "step": 7040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3216497815335864e-05,
+      "loss": 0.0083,
+      "step": 7041
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3214097085513996e-05,
+      "loss": 0.002,
+      "step": 7042
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3211696355692135e-05,
+      "loss": 0.0024,
+      "step": 7043
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3209295625870267e-05,
+      "loss": 0.0022,
+      "step": 7044
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32068948960484e-05,
+      "loss": 0.3969,
+      "step": 7045
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320449416622654e-05,
+      "loss": 0.0021,
+      "step": 7046
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320209343640467e-05,
+      "loss": 0.0018,
+      "step": 7047
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31996927065828e-05,
+      "loss": 0.8296,
+      "step": 7048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319729197676094e-05,
+      "loss": 0.0019,
+      "step": 7049
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319489124693907e-05,
+      "loss": 0.0017,
+      "step": 7050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31924905171172e-05,
+      "loss": 0.0019,
+      "step": 7051
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319008978729534e-05,
+      "loss": 0.0022,
+      "step": 7052
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.318768905747347e-05,
+      "loss": 0.0021,
+      "step": 7053
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.318528832765161e-05,
+      "loss": 0.002,
+      "step": 7054
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3182887597829743e-05,
+      "loss": 0.0022,
+      "step": 7055
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3180486868007875e-05,
+      "loss": 0.0017,
+      "step": 7056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317808613818601e-05,
+      "loss": 0.7791,
+      "step": 7057
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3175685408364146e-05,
+      "loss": 0.0023,
+      "step": 7058
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317328467854228e-05,
+      "loss": 0.0019,
+      "step": 7059
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317088394872041e-05,
+      "loss": 0.0022,
+      "step": 7060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316848321889855e-05,
+      "loss": 0.0019,
+      "step": 7061
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316608248907668e-05,
+      "loss": 0.0022,
+      "step": 7062
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316368175925482e-05,
+      "loss": 0.0021,
+      "step": 7063
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.316128102943295e-05,
+      "loss": 0.0019,
+      "step": 7064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315888029961108e-05,
+      "loss": 0.0017,
+      "step": 7065
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315647956978922e-05,
+      "loss": 0.7669,
+      "step": 7066
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.315407883996735e-05,
+      "loss": 0.0023,
+      "step": 7067
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3151678110145484e-05,
+      "loss": 0.002,
+      "step": 7068
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3149277380323616e-05,
+      "loss": 0.0023,
+      "step": 7069
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3146876650501754e-05,
+      "loss": 0.0021,
+      "step": 7070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.314447592067989e-05,
+      "loss": 0.0025,
+      "step": 7071
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3142075190858025e-05,
+      "loss": 0.0019,
+      "step": 7072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3139674461036157e-05,
+      "loss": 0.0023,
+      "step": 7073
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.313727373121429e-05,
+      "loss": 0.0021,
+      "step": 7074
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.313487300139243e-05,
+      "loss": 0.0022,
+      "step": 7075
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.313247227157056e-05,
+      "loss": 0.0021,
+      "step": 7076
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.313007154174869e-05,
+      "loss": 0.002,
+      "step": 7077
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.312767081192682e-05,
+      "loss": 0.0019,
+      "step": 7078
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.312527008210496e-05,
+      "loss": 0.0024,
+      "step": 7079
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.31228693522831e-05,
+      "loss": 0.0018,
+      "step": 7080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.312046862246123e-05,
+      "loss": 0.0017,
+      "step": 7081
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.311806789263936e-05,
+      "loss": 0.0019,
+      "step": 7082
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.31156671628175e-05,
+      "loss": 0.002,
+      "step": 7083
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3113266432995633e-05,
+      "loss": 0.0022,
+      "step": 7084
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3110865703173765e-05,
+      "loss": 0.0023,
+      "step": 7085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.31084649733519e-05,
+      "loss": 0.0017,
+      "step": 7086
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3106064243530036e-05,
+      "loss": 0.0021,
+      "step": 7087
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.310366351370817e-05,
+      "loss": 0.366,
+      "step": 7088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3101262783886306e-05,
+      "loss": 0.0018,
+      "step": 7089
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309886205406444e-05,
+      "loss": 0.7871,
+      "step": 7090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309646132424257e-05,
+      "loss": 0.0021,
+      "step": 7091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309406059442071e-05,
+      "loss": 0.0017,
+      "step": 7092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.309165986459884e-05,
+      "loss": 0.0021,
+      "step": 7093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.308925913477697e-05,
+      "loss": 0.0019,
+      "step": 7094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3086858404955104e-05,
+      "loss": 0.0018,
+      "step": 7095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.308445767513324e-05,
+      "loss": 0.0018,
+      "step": 7096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3082056945311374e-05,
+      "loss": 0.0017,
+      "step": 7097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.307965621548951e-05,
+      "loss": 0.0019,
+      "step": 7098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3077255485667644e-05,
+      "loss": 0.002,
+      "step": 7099
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.307485475584578e-05,
+      "loss": 0.1564,
+      "step": 7100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3072454026023915e-05,
+      "loss": 0.0016,
+      "step": 7101
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3070053296202047e-05,
+      "loss": 0.0021,
+      "step": 7102
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.306765256638018e-05,
+      "loss": 0.0017,
+      "step": 7103
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.306525183655832e-05,
+      "loss": 0.0019,
+      "step": 7104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.306285110673645e-05,
+      "loss": 0.7854,
+      "step": 7105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.306045037691458e-05,
+      "loss": 0.0021,
+      "step": 7106
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305804964709272e-05,
+      "loss": 0.0019,
+      "step": 7107
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305564891727085e-05,
+      "loss": 0.3019,
+      "step": 7108
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305324818744899e-05,
+      "loss": 0.0018,
+      "step": 7109
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305084745762712e-05,
+      "loss": 0.0019,
+      "step": 7110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.304844672780525e-05,
+      "loss": 0.0018,
+      "step": 7111
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3046045997983385e-05,
+      "loss": 0.0019,
+      "step": 7112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3043645268161523e-05,
+      "loss": 0.002,
+      "step": 7113
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3041244538339655e-05,
+      "loss": 0.0021,
+      "step": 7114
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.303884380851779e-05,
+      "loss": 0.0023,
+      "step": 7115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3036443078695926e-05,
+      "loss": 0.0021,
+      "step": 7116
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3034042348874064e-05,
+      "loss": 0.301,
+      "step": 7117
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3031641619052196e-05,
+      "loss": 0.7082,
+      "step": 7118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302924088923033e-05,
+      "loss": 0.0019,
+      "step": 7119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302684015940846e-05,
+      "loss": 0.0022,
+      "step": 7120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.30244394295866e-05,
+      "loss": 0.002,
+      "step": 7121
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302203869976473e-05,
+      "loss": 0.0018,
+      "step": 7122
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.301963796994286e-05,
+      "loss": 0.0023,
+      "step": 7123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3017237240120994e-05,
+      "loss": 0.0017,
+      "step": 7124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.301483651029913e-05,
+      "loss": 0.0023,
+      "step": 7125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.301243578047727e-05,
+      "loss": 0.0019,
+      "step": 7126
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.30100350506554e-05,
+      "loss": 0.0018,
+      "step": 7127
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3007634320833534e-05,
+      "loss": 0.0015,
+      "step": 7128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3005233591011666e-05,
+      "loss": 0.0018,
+      "step": 7129
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3002832861189805e-05,
+      "loss": 0.0019,
+      "step": 7130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3000432131367937e-05,
+      "loss": 0.002,
+      "step": 7131
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.299803140154607e-05,
+      "loss": 0.0021,
+      "step": 7132
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.299563067172421e-05,
+      "loss": 0.0019,
+      "step": 7133
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.299322994190234e-05,
+      "loss": 0.0021,
+      "step": 7134
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.299082921208048e-05,
+      "loss": 0.0018,
+      "step": 7135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298842848225861e-05,
+      "loss": 0.002,
+      "step": 7136
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298602775243674e-05,
+      "loss": 0.0018,
+      "step": 7137
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298362702261488e-05,
+      "loss": 0.0022,
+      "step": 7138
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.298122629279301e-05,
+      "loss": 0.438,
+      "step": 7139
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.297882556297114e-05,
+      "loss": 0.002,
+      "step": 7140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2976424833149275e-05,
+      "loss": 0.0017,
+      "step": 7141
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2974024103327413e-05,
+      "loss": 0.0016,
+      "step": 7142
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2971623373505545e-05,
+      "loss": 0.0018,
+      "step": 7143
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2969222643683684e-05,
+      "loss": 0.002,
+      "step": 7144
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2966821913861816e-05,
+      "loss": 0.2842,
+      "step": 7145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2964421184039954e-05,
+      "loss": 0.0019,
+      "step": 7146
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2962020454218086e-05,
+      "loss": 0.0724,
+      "step": 7147
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295961972439622e-05,
+      "loss": 0.0023,
+      "step": 7148
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295721899457435e-05,
+      "loss": 0.0017,
+      "step": 7149
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295481826475249e-05,
+      "loss": 0.0018,
+      "step": 7150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295241753493062e-05,
+      "loss": 0.8443,
+      "step": 7151
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.295001680510875e-05,
+      "loss": 0.0016,
+      "step": 7152
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294761607528689e-05,
+      "loss": 0.0017,
+      "step": 7153
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294521534546502e-05,
+      "loss": 0.0021,
+      "step": 7154
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294281461564316e-05,
+      "loss": 0.0015,
+      "step": 7155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.294041388582129e-05,
+      "loss": 0.0018,
+      "step": 7156
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2938013155999424e-05,
+      "loss": 0.0019,
+      "step": 7157
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2935612426177556e-05,
+      "loss": 0.0019,
+      "step": 7158
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2933211696355695e-05,
+      "loss": 0.0018,
+      "step": 7159
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2930810966533826e-05,
+      "loss": 0.002,
+      "step": 7160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.292841023671196e-05,
+      "loss": 0.0019,
+      "step": 7161
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29260095068901e-05,
+      "loss": 0.0018,
+      "step": 7162
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2923608777068235e-05,
+      "loss": 0.0019,
+      "step": 7163
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.292120804724637e-05,
+      "loss": 0.7627,
+      "step": 7164
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29188073174245e-05,
+      "loss": 0.3805,
+      "step": 7165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.291640658760263e-05,
+      "loss": 0.0024,
+      "step": 7166
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.291400585778077e-05,
+      "loss": 1.1445,
+      "step": 7167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29116051279589e-05,
+      "loss": 1.5502,
+      "step": 7168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.290920439813703e-05,
+      "loss": 0.0023,
+      "step": 7169
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2906803668315165e-05,
+      "loss": 1.1185,
+      "step": 7170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.29044029384933e-05,
+      "loss": 0.0028,
+      "step": 7171
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.290200220867144e-05,
+      "loss": 0.0023,
+      "step": 7172
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2899601478849574e-05,
+      "loss": 0.0025,
+      "step": 7173
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2897200749027706e-05,
+      "loss": 1.0357,
+      "step": 7174
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.289480001920584e-05,
+      "loss": 0.0024,
+      "step": 7175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2892399289383976e-05,
+      "loss": 0.0024,
+      "step": 7176
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288999855956211e-05,
+      "loss": 0.0029,
+      "step": 7177
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288759782974024e-05,
+      "loss": 0.0032,
+      "step": 7178
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288519709991837e-05,
+      "loss": 0.4733,
+      "step": 7179
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288279637009652e-05,
+      "loss": 1.5064,
+      "step": 7180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.288039564027465e-05,
+      "loss": 2.1819,
+      "step": 7181
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287799491045278e-05,
+      "loss": 1.0395,
+      "step": 7182
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287559418063091e-05,
+      "loss": 0.7555,
+      "step": 7183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287319345080905e-05,
+      "loss": 0.0031,
+      "step": 7184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.287079272098718e-05,
+      "loss": 0.0087,
+      "step": 7185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2868391991165314e-05,
+      "loss": 0.0037,
+      "step": 7186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2865991261343446e-05,
+      "loss": 0.0043,
+      "step": 7187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2863590531521585e-05,
+      "loss": 0.0043,
+      "step": 7188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.286118980169972e-05,
+      "loss": 0.0036,
+      "step": 7189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2858789071877855e-05,
+      "loss": 0.003,
+      "step": 7190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.285638834205599e-05,
+      "loss": 0.0037,
+      "step": 7191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.285398761223412e-05,
+      "loss": 0.0047,
+      "step": 7192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.285158688241226e-05,
+      "loss": 0.0039,
+      "step": 7193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.284918615259039e-05,
+      "loss": 0.0047,
+      "step": 7194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.284678542276852e-05,
+      "loss": 0.0746,
+      "step": 7195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.284438469294665e-05,
+      "loss": 0.003,
+      "step": 7196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.284198396312479e-05,
+      "loss": 0.0027,
+      "step": 7197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.283958323330292e-05,
+      "loss": 0.0038,
+      "step": 7198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.283718250348106e-05,
+      "loss": 0.0029,
+      "step": 7199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.283478177365919e-05,
+      "loss": 0.0044,
+      "step": 7200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.283238104383733e-05,
+      "loss": 0.0053,
+      "step": 7201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2829980314015464e-05,
+      "loss": 0.6121,
+      "step": 7202
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2827579584193596e-05,
+      "loss": 0.0038,
+      "step": 7203
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.282517885437173e-05,
+      "loss": 0.0021,
+      "step": 7204
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2822778124549866e-05,
+      "loss": 0.0035,
+      "step": 7205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2820377394728e-05,
+      "loss": 0.0048,
+      "step": 7206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281797666490613e-05,
+      "loss": 0.0028,
+      "step": 7207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281557593508427e-05,
+      "loss": 0.0048,
+      "step": 7208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.28131752052624e-05,
+      "loss": 0.005,
+      "step": 7209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.281077447544054e-05,
+      "loss": 0.0034,
+      "step": 7210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.280837374561867e-05,
+      "loss": 0.0033,
+      "step": 7211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.28059730157968e-05,
+      "loss": 0.0033,
+      "step": 7212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.280357228597494e-05,
+      "loss": 0.0022,
+      "step": 7213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.280117155615307e-05,
+      "loss": 0.1432,
+      "step": 7214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2798770826331204e-05,
+      "loss": 0.004,
+      "step": 7215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2796370096509336e-05,
+      "loss": 0.1042,
+      "step": 7216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2793969366687475e-05,
+      "loss": 0.0033,
+      "step": 7217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.279156863686561e-05,
+      "loss": 0.0033,
+      "step": 7218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2789167907043745e-05,
+      "loss": 0.0042,
+      "step": 7219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278676717722188e-05,
+      "loss": 0.0038,
+      "step": 7220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278436644740001e-05,
+      "loss": 0.2056,
+      "step": 7221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.278196571757815e-05,
+      "loss": 0.003,
+      "step": 7222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277956498775628e-05,
+      "loss": 0.0026,
+      "step": 7223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277716425793441e-05,
+      "loss": 0.0018,
+      "step": 7224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277476352811254e-05,
+      "loss": 0.0031,
+      "step": 7225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.277236279829068e-05,
+      "loss": 0.0025,
+      "step": 7226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.276996206846882e-05,
+      "loss": 0.0032,
+      "step": 7227
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.276756133864695e-05,
+      "loss": 0.0024,
+      "step": 7228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.276516060882508e-05,
+      "loss": 0.0026,
+      "step": 7229
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.276275987900322e-05,
+      "loss": 0.0026,
+      "step": 7230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2760359149181354e-05,
+      "loss": 0.0026,
+      "step": 7231
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2757958419359486e-05,
+      "loss": 0.0021,
+      "step": 7232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.275555768953762e-05,
+      "loss": 0.003,
+      "step": 7233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2753156959715756e-05,
+      "loss": 0.0028,
+      "step": 7234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2750756229893894e-05,
+      "loss": 0.0017,
+      "step": 7235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2748355500072026e-05,
+      "loss": 0.0016,
+      "step": 7236
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.274595477025016e-05,
+      "loss": 0.0022,
+      "step": 7237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.274355404042829e-05,
+      "loss": 0.0018,
+      "step": 7238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.274115331060643e-05,
+      "loss": 0.0017,
+      "step": 7239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273875258078456e-05,
+      "loss": 0.0031,
+      "step": 7240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273635185096269e-05,
+      "loss": 0.0026,
+      "step": 7241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2733951121140824e-05,
+      "loss": 0.0022,
+      "step": 7242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.273155039131896e-05,
+      "loss": 0.002,
+      "step": 7243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27291496614971e-05,
+      "loss": 0.0017,
+      "step": 7244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.272674893167523e-05,
+      "loss": 0.0022,
+      "step": 7245
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2724348201853365e-05,
+      "loss": 0.0014,
+      "step": 7246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27219474720315e-05,
+      "loss": 0.0017,
+      "step": 7247
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2719546742209635e-05,
+      "loss": 0.0014,
+      "step": 7248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.271714601238777e-05,
+      "loss": 0.0024,
+      "step": 7249
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27147452825659e-05,
+      "loss": 0.0019,
+      "step": 7250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.271234455274404e-05,
+      "loss": 0.0211,
+      "step": 7251
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270994382292217e-05,
+      "loss": 0.0015,
+      "step": 7252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27075430931003e-05,
+      "loss": 0.0017,
+      "step": 7253
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270514236327844e-05,
+      "loss": 0.0017,
+      "step": 7254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270274163345657e-05,
+      "loss": 0.0019,
+      "step": 7255
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.270034090363471e-05,
+      "loss": 0.0018,
+      "step": 7256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269794017381284e-05,
+      "loss": 0.002,
+      "step": 7257
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.269553944399097e-05,
+      "loss": 0.0019,
+      "step": 7258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2693138714169105e-05,
+      "loss": 0.0017,
+      "step": 7259
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2690737984347244e-05,
+      "loss": 0.0015,
+      "step": 7260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2688337254525375e-05,
+      "loss": 0.0013,
+      "step": 7261
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.268593652470351e-05,
+      "loss": 0.0019,
+      "step": 7262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2683535794881646e-05,
+      "loss": 0.0014,
+      "step": 7263
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2681135065059784e-05,
+      "loss": 0.0019,
+      "step": 7264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2678734335237916e-05,
+      "loss": 0.0012,
+      "step": 7265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.267633360541605e-05,
+      "loss": 0.0019,
+      "step": 7266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.267393287559418e-05,
+      "loss": 0.0015,
+      "step": 7267
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.267153214577232e-05,
+      "loss": 0.0017,
+      "step": 7268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266913141595045e-05,
+      "loss": 0.0019,
+      "step": 7269
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266673068612858e-05,
+      "loss": 0.0017,
+      "step": 7270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2664329956306714e-05,
+      "loss": 0.0012,
+      "step": 7271
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266192922648485e-05,
+      "loss": 0.0015,
+      "step": 7272
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.265952849666299e-05,
+      "loss": 0.0015,
+      "step": 7273
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.265712776684112e-05,
+      "loss": 0.0017,
+      "step": 7274
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2654727037019255e-05,
+      "loss": 0.8689,
+      "step": 7275
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2652326307197386e-05,
+      "loss": 0.0016,
+      "step": 7276
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2649925577375525e-05,
+      "loss": 0.0016,
+      "step": 7277
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264752484755366e-05,
+      "loss": 0.0013,
+      "step": 7278
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264512411773179e-05,
+      "loss": 0.0015,
+      "step": 7279
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.264272338790992e-05,
+      "loss": 0.0015,
+      "step": 7280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2640322658088066e-05,
+      "loss": 0.0019,
+      "step": 7281
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.26379219282662e-05,
+      "loss": 0.0012,
+      "step": 7282
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.263552119844433e-05,
+      "loss": 0.0014,
+      "step": 7283
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.263312046862246e-05,
+      "loss": 0.0015,
+      "step": 7284
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.26307197388006e-05,
+      "loss": 0.0012,
+      "step": 7285
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.262831900897873e-05,
+      "loss": 0.0016,
+      "step": 7286
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.262591827915686e-05,
+      "loss": 0.0014,
+      "step": 7287
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2623517549334995e-05,
+      "loss": 0.0015,
+      "step": 7288
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2621116819513134e-05,
+      "loss": 0.0015,
+      "step": 7289
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.261871608969127e-05,
+      "loss": 0.0014,
+      "step": 7290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2616315359869404e-05,
+      "loss": 0.0012,
+      "step": 7291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2613914630047536e-05,
+      "loss": 0.0015,
+      "step": 7292
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.261151390022567e-05,
+      "loss": 0.0012,
+      "step": 7293
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2609113170403806e-05,
+      "loss": 0.0015,
+      "step": 7294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260671244058194e-05,
+      "loss": 0.0012,
+      "step": 7295
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260431171076007e-05,
+      "loss": 0.0014,
+      "step": 7296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.260191098093821e-05,
+      "loss": 0.0014,
+      "step": 7297
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.259951025111634e-05,
+      "loss": 0.0012,
+      "step": 7298
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.259710952129448e-05,
+      "loss": 0.0013,
+      "step": 7299
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.259470879147261e-05,
+      "loss": 0.0011,
+      "step": 7300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.259230806165074e-05,
+      "loss": 0.0011,
+      "step": 7301
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.258990733182888e-05,
+      "loss": 0.001,
+      "step": 7302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.258750660200701e-05,
+      "loss": 0.0013,
+      "step": 7303
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2585105872185145e-05,
+      "loss": 0.0013,
+      "step": 7304
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2582705142363276e-05,
+      "loss": 0.001,
+      "step": 7305
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2580304412541415e-05,
+      "loss": 0.0012,
+      "step": 7306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257790368271955e-05,
+      "loss": 0.8538,
+      "step": 7307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2575502952897685e-05,
+      "loss": 0.0012,
+      "step": 7308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257310222307582e-05,
+      "loss": 0.0012,
+      "step": 7309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.257070149325395e-05,
+      "loss": 0.8156,
+      "step": 7310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256830076343209e-05,
+      "loss": 0.0012,
+      "step": 7311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256590003361022e-05,
+      "loss": 0.0014,
+      "step": 7312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256349930378835e-05,
+      "loss": 0.0011,
+      "step": 7313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.256109857396649e-05,
+      "loss": 0.0012,
+      "step": 7314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.255869784414462e-05,
+      "loss": 0.0017,
+      "step": 7315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.255629711432275e-05,
+      "loss": 0.0014,
+      "step": 7316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2553896384500885e-05,
+      "loss": 0.0014,
+      "step": 7317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2551495654679024e-05,
+      "loss": 0.4154,
+      "step": 7318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.254909492485716e-05,
+      "loss": 0.0014,
+      "step": 7319
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2546694195035294e-05,
+      "loss": 0.0012,
+      "step": 7320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2544293465213426e-05,
+      "loss": 0.0013,
+      "step": 7321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.254189273539156e-05,
+      "loss": 0.0018,
+      "step": 7322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2539492005569696e-05,
+      "loss": 0.7713,
+      "step": 7323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253709127574783e-05,
+      "loss": 0.0015,
+      "step": 7324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253469054592596e-05,
+      "loss": 0.0014,
+      "step": 7325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.253228981610409e-05,
+      "loss": 0.0018,
+      "step": 7326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.252988908628224e-05,
+      "loss": 0.0016,
+      "step": 7327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.252748835646037e-05,
+      "loss": 0.0015,
+      "step": 7328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.25250876266385e-05,
+      "loss": 0.0019,
+      "step": 7329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.252268689681663e-05,
+      "loss": 0.0018,
+      "step": 7330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.252028616699477e-05,
+      "loss": 0.0011,
+      "step": 7331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.25178854371729e-05,
+      "loss": 0.0016,
+      "step": 7332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2515484707351035e-05,
+      "loss": 0.0014,
+      "step": 7333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2513083977529166e-05,
+      "loss": 0.0014,
+      "step": 7334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2510683247707305e-05,
+      "loss": 0.0016,
+      "step": 7335
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2508282517885443e-05,
+      "loss": 0.0016,
+      "step": 7336
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2505881788063575e-05,
+      "loss": 0.0015,
+      "step": 7337
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.250348105824171e-05,
+      "loss": 0.0017,
+      "step": 7338
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.250108032841984e-05,
+      "loss": 0.8162,
+      "step": 7339
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.249867959859798e-05,
+      "loss": 0.0014,
+      "step": 7340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.249627886877611e-05,
+      "loss": 0.0016,
+      "step": 7341
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.249387813895424e-05,
+      "loss": 0.0019,
+      "step": 7342
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.249147740913237e-05,
+      "loss": 0.7418,
+      "step": 7343
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.248907667931051e-05,
+      "loss": 0.0021,
+      "step": 7344
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.248667594948865e-05,
+      "loss": 0.0016,
+      "step": 7345
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.248427521966678e-05,
+      "loss": 0.0016,
+      "step": 7346
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2481874489844914e-05,
+      "loss": 0.0016,
+      "step": 7347
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.247947376002305e-05,
+      "loss": 0.0014,
+      "step": 7348
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2477073030201184e-05,
+      "loss": 0.0025,
+      "step": 7349
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2474672300379316e-05,
+      "loss": 0.7901,
+      "step": 7350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.247227157055745e-05,
+      "loss": 0.7654,
+      "step": 7351
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2469870840735586e-05,
+      "loss": 0.4061,
+      "step": 7352
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.246747011091372e-05,
+      "loss": 0.476,
+      "step": 7353
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2465069381091857e-05,
+      "loss": 0.0128,
+      "step": 7354
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.246266865126999e-05,
+      "loss": 0.381,
+      "step": 7355
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.246026792144812e-05,
+      "loss": 0.3157,
+      "step": 7356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.245786719162626e-05,
+      "loss": 0.0013,
+      "step": 7357
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.245546646180439e-05,
+      "loss": 0.0142,
+      "step": 7358
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.245306573198252e-05,
+      "loss": 0.002,
+      "step": 7359
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2450665002160654e-05,
+      "loss": 0.0028,
+      "step": 7360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.244826427233879e-05,
+      "loss": 0.0027,
+      "step": 7361
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2445863542516924e-05,
+      "loss": 0.0023,
+      "step": 7362
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.244346281269506e-05,
+      "loss": 0.3517,
+      "step": 7363
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2441062082873195e-05,
+      "loss": 0.0034,
+      "step": 7364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2438661353051333e-05,
+      "loss": 0.0028,
+      "step": 7365
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2436260623229465e-05,
+      "loss": 0.0018,
+      "step": 7366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.24338598934076e-05,
+      "loss": 0.0026,
+      "step": 7367
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.243145916358573e-05,
+      "loss": 0.0032,
+      "step": 7368
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.242905843376387e-05,
+      "loss": 0.0034,
+      "step": 7369
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2426657703942e-05,
+      "loss": 0.0035,
+      "step": 7370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.242425697412013e-05,
+      "loss": 0.0025,
+      "step": 7371
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.242185624429827e-05,
+      "loss": 0.0029,
+      "step": 7372
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.24194555144764e-05,
+      "loss": 0.0022,
+      "step": 7373
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.241705478465454e-05,
+      "loss": 0.0017,
+      "step": 7374
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.241465405483267e-05,
+      "loss": 0.0032,
+      "step": 7375
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2412253325010804e-05,
+      "loss": 0.1131,
+      "step": 7376
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2409852595188935e-05,
+      "loss": 0.0029,
+      "step": 7377
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2407451865367074e-05,
+      "loss": 0.8261,
+      "step": 7378
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2405051135545206e-05,
+      "loss": 0.0023,
+      "step": 7379
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.240265040572334e-05,
+      "loss": 0.0028,
+      "step": 7380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2400249675901476e-05,
+      "loss": 0.0096,
+      "step": 7381
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2397848946079615e-05,
+      "loss": 0.0033,
+      "step": 7382
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2395448216257746e-05,
+      "loss": 0.0031,
+      "step": 7383
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239304748643588e-05,
+      "loss": 0.0025,
+      "step": 7384
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239064675661401e-05,
+      "loss": 0.0026,
+      "step": 7385
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238824602679215e-05,
+      "loss": 0.0027,
+      "step": 7386
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238584529697028e-05,
+      "loss": 0.0068,
+      "step": 7387
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.238344456714841e-05,
+      "loss": 0.0056,
+      "step": 7388
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2381043837326544e-05,
+      "loss": 0.0032,
+      "step": 7389
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237864310750468e-05,
+      "loss": 0.0032,
+      "step": 7390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237624237768282e-05,
+      "loss": 0.0028,
+      "step": 7391
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.237384164786095e-05,
+      "loss": 0.0013,
+      "step": 7392
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2371440918039085e-05,
+      "loss": 0.0024,
+      "step": 7393
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2369040188217223e-05,
+      "loss": 0.0071,
+      "step": 7394
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2366639458395355e-05,
+      "loss": 0.0027,
+      "step": 7395
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.236423872857349e-05,
+      "loss": 0.0019,
+      "step": 7396
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.236183799875162e-05,
+      "loss": 0.0021,
+      "step": 7397
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.235943726892976e-05,
+      "loss": 0.0012,
+      "step": 7398
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.235703653910789e-05,
+      "loss": 0.0017,
+      "step": 7399
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.235463580928603e-05,
+      "loss": 0.0028,
+      "step": 7400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.235223507946416e-05,
+      "loss": 0.0021,
+      "step": 7401
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.234983434964229e-05,
+      "loss": 0.0018,
+      "step": 7402
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.234743361982043e-05,
+      "loss": 0.0017,
+      "step": 7403
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.234503288999856e-05,
+      "loss": 0.8907,
+      "step": 7404
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2342632160176694e-05,
+      "loss": 0.0015,
+      "step": 7405
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2340231430354825e-05,
+      "loss": 0.0023,
+      "step": 7406
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2337830700532964e-05,
+      "loss": 0.0018,
+      "step": 7407
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2335429970711096e-05,
+      "loss": 0.0021,
+      "step": 7408
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2333029240889234e-05,
+      "loss": 0.0022,
+      "step": 7409
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2330628511067366e-05,
+      "loss": 0.8777,
+      "step": 7410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2328227781245505e-05,
+      "loss": 0.0016,
+      "step": 7411
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2325827051423636e-05,
+      "loss": 0.0041,
+      "step": 7412
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.232342632160177e-05,
+      "loss": 0.0021,
+      "step": 7413
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.23210255917799e-05,
+      "loss": 0.0015,
+      "step": 7414
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.231862486195804e-05,
+      "loss": 0.0021,
+      "step": 7415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.231622413213617e-05,
+      "loss": 0.0012,
+      "step": 7416
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.23138234023143e-05,
+      "loss": 0.0023,
+      "step": 7417
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.231142267249244e-05,
+      "loss": 0.0022,
+      "step": 7418
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.230902194267057e-05,
+      "loss": 0.0016,
+      "step": 7419
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.230662121284871e-05,
+      "loss": 0.002,
+      "step": 7420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.230422048302684e-05,
+      "loss": 0.0023,
+      "step": 7421
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2301819753204975e-05,
+      "loss": 0.0021,
+      "step": 7422
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2299419023383107e-05,
+      "loss": 0.0015,
+      "step": 7423
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2297018293561245e-05,
+      "loss": 0.0016,
+      "step": 7424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.229461756373938e-05,
+      "loss": 0.0017,
+      "step": 7425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.229221683391751e-05,
+      "loss": 0.85,
+      "step": 7426
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228981610409565e-05,
+      "loss": 0.0019,
+      "step": 7427
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2287415374273786e-05,
+      "loss": 0.002,
+      "step": 7428
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228501464445192e-05,
+      "loss": 0.0021,
+      "step": 7429
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228261391463005e-05,
+      "loss": 0.002,
+      "step": 7430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.228021318480818e-05,
+      "loss": 0.0068,
+      "step": 7431
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.227781245498632e-05,
+      "loss": 0.0015,
+      "step": 7432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.227541172516445e-05,
+      "loss": 0.0016,
+      "step": 7433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2273010995342584e-05,
+      "loss": 0.0015,
+      "step": 7434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2270610265520715e-05,
+      "loss": 0.002,
+      "step": 7435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2268209535698854e-05,
+      "loss": 0.0016,
+      "step": 7436
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.226580880587699e-05,
+      "loss": 0.0017,
+      "step": 7437
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2263408076055124e-05,
+      "loss": 0.0019,
+      "step": 7438
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2261007346233256e-05,
+      "loss": 0.7953,
+      "step": 7439
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.225860661641139e-05,
+      "loss": 0.0015,
+      "step": 7440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2256205886589526e-05,
+      "loss": 0.0016,
+      "step": 7441
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.225380515676766e-05,
+      "loss": 0.0018,
+      "step": 7442
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.225140442694579e-05,
+      "loss": 0.0018,
+      "step": 7443
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.224900369712392e-05,
+      "loss": 0.0017,
+      "step": 7444
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.224660296730206e-05,
+      "loss": 0.0017,
+      "step": 7445
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22442022374802e-05,
+      "loss": 0.8375,
+      "step": 7446
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.224180150765833e-05,
+      "loss": 0.0023,
+      "step": 7447
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.223940077783646e-05,
+      "loss": 0.0017,
+      "step": 7448
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22370000480146e-05,
+      "loss": 0.002,
+      "step": 7449
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.223459931819273e-05,
+      "loss": 0.8522,
+      "step": 7450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2232198588370865e-05,
+      "loss": 0.0018,
+      "step": 7451
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2229797858548997e-05,
+      "loss": 0.0015,
+      "step": 7452
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2227397128727135e-05,
+      "loss": 0.0047,
+      "step": 7453
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222499639890527e-05,
+      "loss": 0.0019,
+      "step": 7454
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2222595669083406e-05,
+      "loss": 0.002,
+      "step": 7455
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.222019493926154e-05,
+      "loss": 0.002,
+      "step": 7456
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221779420943967e-05,
+      "loss": 0.0069,
+      "step": 7457
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221539347961781e-05,
+      "loss": 0.0022,
+      "step": 7458
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221299274979594e-05,
+      "loss": 0.0019,
+      "step": 7459
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221059201997407e-05,
+      "loss": 0.7887,
+      "step": 7460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.22081912901522e-05,
+      "loss": 0.0041,
+      "step": 7461
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.220579056033034e-05,
+      "loss": 0.0019,
+      "step": 7462
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2203389830508473e-05,
+      "loss": 0.0023,
+      "step": 7463
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.220098910068661e-05,
+      "loss": 0.005,
+      "step": 7464
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2198588370864744e-05,
+      "loss": 0.0041,
+      "step": 7465
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.219618764104288e-05,
+      "loss": 0.0027,
+      "step": 7466
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2193786911221014e-05,
+      "loss": 0.0046,
+      "step": 7467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2191386181399146e-05,
+      "loss": 0.0023,
+      "step": 7468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.218898545157728e-05,
+      "loss": 0.004,
+      "step": 7469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2186584721755416e-05,
+      "loss": 0.0021,
+      "step": 7470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.218418399193355e-05,
+      "loss": 0.0023,
+      "step": 7471
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.218178326211168e-05,
+      "loss": 0.0039,
+      "step": 7472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217938253228982e-05,
+      "loss": 0.0037,
+      "step": 7473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217698180246795e-05,
+      "loss": 0.0022,
+      "step": 7474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217458107264609e-05,
+      "loss": 0.002,
+      "step": 7475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217218034282422e-05,
+      "loss": 0.0056,
+      "step": 7476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.216977961300235e-05,
+      "loss": 0.0022,
+      "step": 7477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.216737888318049e-05,
+      "loss": 0.0022,
+      "step": 7478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.216497815335862e-05,
+      "loss": 0.7495,
+      "step": 7479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2162577423536755e-05,
+      "loss": 0.0046,
+      "step": 7480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2160176693714887e-05,
+      "loss": 0.7958,
+      "step": 7481
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2157775963893025e-05,
+      "loss": 0.0022,
+      "step": 7482
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2155375234071164e-05,
+      "loss": 0.003,
+      "step": 7483
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2152974504249295e-05,
+      "loss": 0.0023,
+      "step": 7484
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.215057377442743e-05,
+      "loss": 0.0025,
+      "step": 7485
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.214817304460556e-05,
+      "loss": 0.0022,
+      "step": 7486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.21457723147837e-05,
+      "loss": 0.0023,
+      "step": 7487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.214337158496183e-05,
+      "loss": 0.004,
+      "step": 7488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.214097085513996e-05,
+      "loss": 0.0023,
+      "step": 7489
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.213857012531809e-05,
+      "loss": 0.0031,
+      "step": 7490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.213616939549624e-05,
+      "loss": 0.7261,
+      "step": 7491
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.213376866567437e-05,
+      "loss": 0.0032,
+      "step": 7492
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.21313679358525e-05,
+      "loss": 0.0048,
+      "step": 7493
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2128967206030634e-05,
+      "loss": 0.5486,
+      "step": 7494
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.212656647620877e-05,
+      "loss": 0.0018,
+      "step": 7495
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2124165746386904e-05,
+      "loss": 0.0031,
+      "step": 7496
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2121765016565036e-05,
+      "loss": 0.0027,
+      "step": 7497
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.211936428674317e-05,
+      "loss": 0.0026,
+      "step": 7498
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2116963556921306e-05,
+      "loss": 0.0041,
+      "step": 7499
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2114562827099445e-05,
+      "loss": 0.0029,
+      "step": 7500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.211216209727758e-05,
+      "loss": 0.6809,
+      "step": 7501
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210976136745571e-05,
+      "loss": 0.0028,
+      "step": 7502
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210736063763384e-05,
+      "loss": 0.0039,
+      "step": 7503
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210495990781198e-05,
+      "loss": 0.0053,
+      "step": 7504
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210255917799011e-05,
+      "loss": 0.0037,
+      "step": 7505
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.210015844816824e-05,
+      "loss": 0.0039,
+      "step": 7506
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2097757718346374e-05,
+      "loss": 0.0034,
+      "step": 7507
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.209535698852451e-05,
+      "loss": 0.0051,
+      "step": 7508
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2092956258702645e-05,
+      "loss": 0.7135,
+      "step": 7509
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.209055552888078e-05,
+      "loss": 0.0045,
+      "step": 7510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2088154799058915e-05,
+      "loss": 0.0034,
+      "step": 7511
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2085754069237054e-05,
+      "loss": 0.0039,
+      "step": 7512
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2083353339415185e-05,
+      "loss": 0.0052,
+      "step": 7513
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.208095260959332e-05,
+      "loss": 0.0065,
+      "step": 7514
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207855187977145e-05,
+      "loss": 0.0052,
+      "step": 7515
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207615114994959e-05,
+      "loss": 0.0085,
+      "step": 7516
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207375042012772e-05,
+      "loss": 0.0022,
+      "step": 7517
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.207134969030585e-05,
+      "loss": 0.6493,
+      "step": 7518
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206894896048399e-05,
+      "loss": 1.3959,
+      "step": 7519
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206654823066212e-05,
+      "loss": 0.0049,
+      "step": 7520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206414750084026e-05,
+      "loss": 0.0026,
+      "step": 7521
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206174677101839e-05,
+      "loss": 0.6465,
+      "step": 7522
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2059346041196524e-05,
+      "loss": 0.6325,
+      "step": 7523
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2056945311374656e-05,
+      "loss": 0.0059,
+      "step": 7524
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2054544581552794e-05,
+      "loss": 0.0047,
+      "step": 7525
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2052143851730926e-05,
+      "loss": 0.5634,
+      "step": 7526
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.204974312190906e-05,
+      "loss": 0.9244,
+      "step": 7527
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2047342392087196e-05,
+      "loss": 0.4682,
+      "step": 7528
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2044941662265335e-05,
+      "loss": 0.4849,
+      "step": 7529
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.204254093244347e-05,
+      "loss": 0.0647,
+      "step": 7530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.20401402026216e-05,
+      "loss": 0.3262,
+      "step": 7531
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203773947279973e-05,
+      "loss": 0.355,
+      "step": 7532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203533874297787e-05,
+      "loss": 0.3261,
+      "step": 7533
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2032938013156e-05,
+      "loss": 0.0357,
+      "step": 7534
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203053728333413e-05,
+      "loss": 0.3341,
+      "step": 7535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2028136553512264e-05,
+      "loss": 0.0353,
+      "step": 7536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.20257358236904e-05,
+      "loss": 0.0463,
+      "step": 7537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.202333509386854e-05,
+      "loss": 0.2932,
+      "step": 7538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.202093436404667e-05,
+      "loss": 0.1012,
+      "step": 7539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2018533634224805e-05,
+      "loss": 0.3453,
+      "step": 7540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.201613290440294e-05,
+      "loss": 0.3439,
+      "step": 7541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2013732174581075e-05,
+      "loss": 0.0956,
+      "step": 7542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.201133144475921e-05,
+      "loss": 0.3394,
+      "step": 7543
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.200893071493734e-05,
+      "loss": 0.3396,
+      "step": 7544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.200652998511548e-05,
+      "loss": 0.0657,
+      "step": 7545
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2004129255293616e-05,
+      "loss": 0.0659,
+      "step": 7546
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.200172852547175e-05,
+      "loss": 0.0889,
+      "step": 7547
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.199932779564988e-05,
+      "loss": 0.0565,
+      "step": 7548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.199692706582801e-05,
+      "loss": 0.0656,
+      "step": 7549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.199452633600615e-05,
+      "loss": 0.0202,
+      "step": 7550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.199212560618428e-05,
+      "loss": 0.0571,
+      "step": 7551
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1989724876362414e-05,
+      "loss": 0.0334,
+      "step": 7552
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1987324146540546e-05,
+      "loss": 0.0239,
+      "step": 7553
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1984923416718684e-05,
+      "loss": 0.024,
+      "step": 7554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.198252268689682e-05,
+      "loss": 0.0105,
+      "step": 7555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1980121957074955e-05,
+      "loss": 0.007,
+      "step": 7556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1977721227253086e-05,
+      "loss": 0.6545,
+      "step": 7557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.197532049743122e-05,
+      "loss": 0.0045,
+      "step": 7558
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.197291976760936e-05,
+      "loss": 0.0091,
+      "step": 7559
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.197051903778749e-05,
+      "loss": 0.0085,
+      "step": 7560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196811830796562e-05,
+      "loss": 0.0029,
+      "step": 7561
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196571757814376e-05,
+      "loss": 0.0051,
+      "step": 7562
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196331684832189e-05,
+      "loss": 0.0042,
+      "step": 7563
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.196091611850002e-05,
+      "loss": 0.0034,
+      "step": 7564
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195851538867816e-05,
+      "loss": 0.0025,
+      "step": 7565
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195611465885629e-05,
+      "loss": 0.6751,
+      "step": 7566
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195371392903443e-05,
+      "loss": 0.7458,
+      "step": 7567
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.195131319921256e-05,
+      "loss": 0.0029,
+      "step": 7568
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1948912469390695e-05,
+      "loss": 0.5751,
+      "step": 7569
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.194651173956883e-05,
+      "loss": 0.0038,
+      "step": 7570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1944111009746965e-05,
+      "loss": 0.0056,
+      "step": 7571
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.19417102799251e-05,
+      "loss": 0.0024,
+      "step": 7572
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193930955010323e-05,
+      "loss": 0.0053,
+      "step": 7573
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193690882028137e-05,
+      "loss": 0.0041,
+      "step": 7574
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1934508090459506e-05,
+      "loss": 0.0033,
+      "step": 7575
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.193210736063764e-05,
+      "loss": 0.0036,
+      "step": 7576
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.192970663081577e-05,
+      "loss": 0.0046,
+      "step": 7577
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.19273059009939e-05,
+      "loss": 0.5255,
+      "step": 7578
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.192490517117204e-05,
+      "loss": 0.003,
+      "step": 7579
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.192250444135017e-05,
+      "loss": 0.0031,
+      "step": 7580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1920103711528304e-05,
+      "loss": 0.0047,
+      "step": 7581
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1917702981706436e-05,
+      "loss": 0.0047,
+      "step": 7582
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1915302251884574e-05,
+      "loss": 0.681,
+      "step": 7583
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.191290152206271e-05,
+      "loss": 0.6243,
+      "step": 7584
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1910500792240844e-05,
+      "loss": 0.0028,
+      "step": 7585
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1908100062418976e-05,
+      "loss": 0.7142,
+      "step": 7586
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.190569933259711e-05,
+      "loss": 0.0053,
+      "step": 7587
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.190329860277525e-05,
+      "loss": 0.6327,
+      "step": 7588
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.190089787295338e-05,
+      "loss": 0.3077,
+      "step": 7589
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189849714313151e-05,
+      "loss": 0.0032,
+      "step": 7590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189609641330964e-05,
+      "loss": 0.0081,
+      "step": 7591
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189369568348779e-05,
+      "loss": 0.0042,
+      "step": 7592
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189129495366592e-05,
+      "loss": 0.6139,
+      "step": 7593
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188889422384405e-05,
+      "loss": 0.006,
+      "step": 7594
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188649349402218e-05,
+      "loss": 0.006,
+      "step": 7595
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188409276420032e-05,
+      "loss": 0.0069,
+      "step": 7596
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.188169203437845e-05,
+      "loss": 0.0077,
+      "step": 7597
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1879291304556585e-05,
+      "loss": 0.0068,
+      "step": 7598
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.187689057473472e-05,
+      "loss": 0.0101,
+      "step": 7599
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1874489844912855e-05,
+      "loss": 0.0081,
+      "step": 7600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1872089115090994e-05,
+      "loss": 0.0072,
+      "step": 7601
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1869688385269126e-05,
+      "loss": 0.0094,
+      "step": 7602
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186728765544726e-05,
+      "loss": 0.008,
+      "step": 7603
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186488692562539e-05,
+      "loss": 0.0049,
+      "step": 7604
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186248619580353e-05,
+      "loss": 0.0045,
+      "step": 7605
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186008546598166e-05,
+      "loss": 0.0041,
+      "step": 7606
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.185768473615979e-05,
+      "loss": 0.0046,
+      "step": 7607
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.185528400633792e-05,
+      "loss": 0.0031,
+      "step": 7608
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.185288327651606e-05,
+      "loss": 0.0026,
+      "step": 7609
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.18504825466942e-05,
+      "loss": 0.006,
+      "step": 7610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.184808181687233e-05,
+      "loss": 0.0054,
+      "step": 7611
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1845681087050464e-05,
+      "loss": 0.0017,
+      "step": 7612
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.18432803572286e-05,
+      "loss": 0.0042,
+      "step": 7613
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1840879627406734e-05,
+      "loss": 0.0054,
+      "step": 7614
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1838478897584866e-05,
+      "loss": 0.0018,
+      "step": 7615
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1836078167763e-05,
+      "loss": 0.004,
+      "step": 7616
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.183367743794114e-05,
+      "loss": 0.0042,
+      "step": 7617
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.183127670811927e-05,
+      "loss": 0.0021,
+      "step": 7618
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.182887597829741e-05,
+      "loss": 0.0055,
+      "step": 7619
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.182647524847554e-05,
+      "loss": 0.0034,
+      "step": 7620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.182407451865367e-05,
+      "loss": 0.0027,
+      "step": 7621
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.182167378883181e-05,
+      "loss": 0.0036,
+      "step": 7622
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.181927305900994e-05,
+      "loss": 0.0039,
+      "step": 7623
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.181687232918807e-05,
+      "loss": 0.0019,
+      "step": 7624
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1814471599366205e-05,
+      "loss": 0.6784,
+      "step": 7625
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.181207086954434e-05,
+      "loss": 0.5487,
+      "step": 7626
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1809670139722475e-05,
+      "loss": 0.0047,
+      "step": 7627
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.180726940990061e-05,
+      "loss": 0.0035,
+      "step": 7628
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1804868680078745e-05,
+      "loss": 1.3344,
+      "step": 7629
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1802467950256884e-05,
+      "loss": 0.0032,
+      "step": 7630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1800067220435016e-05,
+      "loss": 0.0068,
+      "step": 7631
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.179766649061315e-05,
+      "loss": 0.0037,
+      "step": 7632
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.179526576079128e-05,
+      "loss": 0.0032,
+      "step": 7633
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.179286503096942e-05,
+      "loss": 0.0023,
+      "step": 7634
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.179046430114755e-05,
+      "loss": 0.0055,
+      "step": 7635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178806357132568e-05,
+      "loss": 0.0019,
+      "step": 7636
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178566284150381e-05,
+      "loss": 0.6858,
+      "step": 7637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178326211168195e-05,
+      "loss": 0.0034,
+      "step": 7638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.178086138186009e-05,
+      "loss": 0.0025,
+      "step": 7639
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.177846065203822e-05,
+      "loss": 0.0036,
+      "step": 7640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1776059922216354e-05,
+      "loss": 0.0024,
+      "step": 7641
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.177365919239449e-05,
+      "loss": 0.0039,
+      "step": 7642
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1771258462572624e-05,
+      "loss": 0.7135,
+      "step": 7643
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1768857732750756e-05,
+      "loss": 0.0048,
+      "step": 7644
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.176645700292889e-05,
+      "loss": 0.0053,
+      "step": 7645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.176405627310703e-05,
+      "loss": 0.6461,
+      "step": 7646
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1761655543285165e-05,
+      "loss": 0.0043,
+      "step": 7647
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.17592548134633e-05,
+      "loss": 0.0036,
+      "step": 7648
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.175685408364143e-05,
+      "loss": 0.6706,
+      "step": 7649
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.175445335381956e-05,
+      "loss": 0.6379,
+      "step": 7650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.17520526239977e-05,
+      "loss": 0.0039,
+      "step": 7651
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174965189417583e-05,
+      "loss": 0.0025,
+      "step": 7652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174725116435396e-05,
+      "loss": 0.0028,
+      "step": 7653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1744850434532095e-05,
+      "loss": 0.0041,
+      "step": 7654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174244970471023e-05,
+      "loss": 0.0036,
+      "step": 7655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174004897488837e-05,
+      "loss": 0.0051,
+      "step": 7656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1737648245066504e-05,
+      "loss": 0.6354,
+      "step": 7657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1735247515244635e-05,
+      "loss": 0.0042,
+      "step": 7658
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1732846785422774e-05,
+      "loss": 0.0057,
+      "step": 7659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1730446055600906e-05,
+      "loss": 0.0026,
+      "step": 7660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172804532577904e-05,
+      "loss": 0.0044,
+      "step": 7661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172564459595717e-05,
+      "loss": 0.0046,
+      "step": 7662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172324386613531e-05,
+      "loss": 0.5905,
+      "step": 7663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.172084313631344e-05,
+      "loss": 0.0091,
+      "step": 7664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171844240649158e-05,
+      "loss": 0.0066,
+      "step": 7665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171604167666971e-05,
+      "loss": 0.0054,
+      "step": 7666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171364094684784e-05,
+      "loss": 0.0038,
+      "step": 7667
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.171124021702598e-05,
+      "loss": 0.0058,
+      "step": 7668
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.170883948720411e-05,
+      "loss": 0.568,
+      "step": 7669
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1706438757382244e-05,
+      "loss": 0.008,
+      "step": 7670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1704038027560376e-05,
+      "loss": 0.0073,
+      "step": 7671
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1701637297738514e-05,
+      "loss": 0.0031,
+      "step": 7672
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1699236567916646e-05,
+      "loss": 0.0062,
+      "step": 7673
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1696835838094785e-05,
+      "loss": 0.0088,
+      "step": 7674
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1694435108272917e-05,
+      "loss": 0.0041,
+      "step": 7675
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1692034378451055e-05,
+      "loss": 0.007,
+      "step": 7676
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168963364862919e-05,
+      "loss": 0.0053,
+      "step": 7677
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168723291880732e-05,
+      "loss": 0.0083,
+      "step": 7678
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168483218898545e-05,
+      "loss": 0.0055,
+      "step": 7679
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168243145916359e-05,
+      "loss": 0.6204,
+      "step": 7680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.168003072934172e-05,
+      "loss": 0.0056,
+      "step": 7681
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167762999951985e-05,
+      "loss": 0.568,
+      "step": 7682
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167522926969799e-05,
+      "loss": 0.6349,
+      "step": 7683
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167282853987612e-05,
+      "loss": 0.0072,
+      "step": 7684
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167042781005426e-05,
+      "loss": 0.0078,
+      "step": 7685
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1668027080232393e-05,
+      "loss": 0.5252,
+      "step": 7686
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1665626350410525e-05,
+      "loss": 0.0082,
+      "step": 7687
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.166322562058866e-05,
+      "loss": 1.2613,
+      "step": 7688
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1660824890766796e-05,
+      "loss": 0.0092,
+      "step": 7689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165842416094493e-05,
+      "loss": 0.0093,
+      "step": 7690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165602343112306e-05,
+      "loss": 0.0069,
+      "step": 7691
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.165362270130119e-05,
+      "loss": 0.0066,
+      "step": 7692
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1651221971479336e-05,
+      "loss": 0.0068,
+      "step": 7693
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.164882124165747e-05,
+      "loss": 0.4037,
+      "step": 7694
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.16464205118356e-05,
+      "loss": 0.0083,
+      "step": 7695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.164401978201373e-05,
+      "loss": 0.0089,
+      "step": 7696
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.164161905219187e-05,
+      "loss": 0.0054,
+      "step": 7697
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.163921832237e-05,
+      "loss": 0.0078,
+      "step": 7698
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1636817592548134e-05,
+      "loss": 0.5465,
+      "step": 7699
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1634416862726266e-05,
+      "loss": 0.0086,
+      "step": 7700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1632016132904404e-05,
+      "loss": 0.0049,
+      "step": 7701
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.162961540308254e-05,
+      "loss": 0.0072,
+      "step": 7702
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1627214673260675e-05,
+      "loss": 0.0053,
+      "step": 7703
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1624813943438807e-05,
+      "loss": 0.0077,
+      "step": 7704
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.162241321361694e-05,
+      "loss": 0.0046,
+      "step": 7705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.162001248379508e-05,
+      "loss": 0.0044,
+      "step": 7706
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161761175397321e-05,
+      "loss": 0.0061,
+      "step": 7707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161521102415134e-05,
+      "loss": 0.0049,
+      "step": 7708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161281029432947e-05,
+      "loss": 0.0068,
+      "step": 7709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.161040956450761e-05,
+      "loss": 0.0021,
+      "step": 7710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160800883468575e-05,
+      "loss": 0.0031,
+      "step": 7711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160560810486388e-05,
+      "loss": 0.0033,
+      "step": 7712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160320737504201e-05,
+      "loss": 0.0048,
+      "step": 7713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.160080664522015e-05,
+      "loss": 0.6285,
+      "step": 7714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1598405915398283e-05,
+      "loss": 0.0045,
+      "step": 7715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1596005185576415e-05,
+      "loss": 0.0048,
+      "step": 7716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.159360445575455e-05,
+      "loss": 0.6552,
+      "step": 7717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1591203725932686e-05,
+      "loss": 0.5603,
+      "step": 7718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.158880299611082e-05,
+      "loss": 0.0038,
+      "step": 7719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1586402266288956e-05,
+      "loss": 0.003,
+      "step": 7720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.158400153646709e-05,
+      "loss": 0.0451,
+      "step": 7721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.158160080664522e-05,
+      "loss": 0.0022,
+      "step": 7722
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157920007682336e-05,
+      "loss": 0.0041,
+      "step": 7723
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157679934700149e-05,
+      "loss": 0.0045,
+      "step": 7724
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157439861717962e-05,
+      "loss": 0.0032,
+      "step": 7725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.157199788735776e-05,
+      "loss": 0.004,
+      "step": 7726
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.156959715753589e-05,
+      "loss": 0.5907,
+      "step": 7727
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1567196427714024e-05,
+      "loss": 0.0045,
+      "step": 7728
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.156479569789216e-05,
+      "loss": 0.004,
+      "step": 7729
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1562394968070294e-05,
+      "loss": 0.0024,
+      "step": 7730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155999423824843e-05,
+      "loss": 0.0022,
+      "step": 7731
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1557593508426565e-05,
+      "loss": 0.0042,
+      "step": 7732
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1555192778604697e-05,
+      "loss": 0.0039,
+      "step": 7733
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155279204878283e-05,
+      "loss": 0.6408,
+      "step": 7734
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.155039131896097e-05,
+      "loss": 0.0042,
+      "step": 7735
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.15479905891391e-05,
+      "loss": 0.0018,
+      "step": 7736
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.154558985931723e-05,
+      "loss": 0.0043,
+      "step": 7737
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.154318912949537e-05,
+      "loss": 0.0054,
+      "step": 7738
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.154078839967351e-05,
+      "loss": 0.0044,
+      "step": 7739
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.153838766985164e-05,
+      "loss": 0.0042,
+      "step": 7740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.153598694002977e-05,
+      "loss": 0.323,
+      "step": 7741
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.15335862102079e-05,
+      "loss": 0.0026,
+      "step": 7742
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.153118548038604e-05,
+      "loss": 0.8769,
+      "step": 7743
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1528784750564173e-05,
+      "loss": 0.002,
+      "step": 7744
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1526384020742305e-05,
+      "loss": 0.5942,
+      "step": 7745
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.152398329092044e-05,
+      "loss": 0.0022,
+      "step": 7746
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1521582561098576e-05,
+      "loss": 0.0027,
+      "step": 7747
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1519181831276714e-05,
+      "loss": 0.6661,
+      "step": 7748
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1516781101454846e-05,
+      "loss": 0.6971,
+      "step": 7749
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.151438037163298e-05,
+      "loss": 0.0028,
+      "step": 7750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.151197964181111e-05,
+      "loss": 0.0042,
+      "step": 7751
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.150957891198925e-05,
+      "loss": 0.0044,
+      "step": 7752
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.150717818216738e-05,
+      "loss": 0.0041,
+      "step": 7753
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.150477745234551e-05,
+      "loss": 0.0068,
+      "step": 7754
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1502376722523644e-05,
+      "loss": 0.0037,
+      "step": 7755
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.149997599270178e-05,
+      "loss": 0.0164,
+      "step": 7756
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.149757526287992e-05,
+      "loss": 0.0033,
+      "step": 7757
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.149517453305805e-05,
+      "loss": 0.0042,
+      "step": 7758
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1492773803236184e-05,
+      "loss": 0.0038,
+      "step": 7759
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149037307341432e-05,
+      "loss": 0.0027,
+      "step": 7760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1487972343592455e-05,
+      "loss": 0.0049,
+      "step": 7761
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1485571613770587e-05,
+      "loss": 0.0051,
+      "step": 7762
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.148317088394872e-05,
+      "loss": 0.0027,
+      "step": 7763
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.148077015412686e-05,
+      "loss": 0.0049,
+      "step": 7764
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147836942430499e-05,
+      "loss": 0.002,
+      "step": 7765
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147596869448313e-05,
+      "loss": 0.0044,
+      "step": 7766
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147356796466126e-05,
+      "loss": 0.0043,
+      "step": 7767
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147116723483939e-05,
+      "loss": 0.0025,
+      "step": 7768
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146876650501753e-05,
+      "loss": 0.0026,
+      "step": 7769
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146636577519566e-05,
+      "loss": 0.0038,
+      "step": 7770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.146396504537379e-05,
+      "loss": 0.0044,
+      "step": 7771
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1461564315551925e-05,
+      "loss": 0.0039,
+      "step": 7772
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1459163585730063e-05,
+      "loss": 0.0039,
+      "step": 7773
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1456762855908195e-05,
+      "loss": 0.003,
+      "step": 7774
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1454362126086334e-05,
+      "loss": 0.0028,
+      "step": 7775
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1451961396264466e-05,
+      "loss": 0.0047,
+      "step": 7776
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1449560666442604e-05,
+      "loss": 0.0024,
+      "step": 7777
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1447159936620736e-05,
+      "loss": 0.0036,
+      "step": 7778
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.144475920679887e-05,
+      "loss": 0.0029,
+      "step": 7779
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1442358476977e-05,
+      "loss": 0.0041,
+      "step": 7780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.143995774715514e-05,
+      "loss": 0.0026,
+      "step": 7781
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.143755701733327e-05,
+      "loss": 0.0038,
+      "step": 7782
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.14351562875114e-05,
+      "loss": 0.0032,
+      "step": 7783
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.143275555768954e-05,
+      "loss": 0.6832,
+      "step": 7784
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.143035482786767e-05,
+      "loss": 0.0038,
+      "step": 7785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.142795409804581e-05,
+      "loss": 0.7217,
+      "step": 7786
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.142555336822394e-05,
+      "loss": 0.0026,
+      "step": 7787
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1423152638402074e-05,
+      "loss": 0.0023,
+      "step": 7788
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1420751908580206e-05,
+      "loss": 0.0034,
+      "step": 7789
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1418351178758345e-05,
+      "loss": 0.0027,
+      "step": 7790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1415950448936476e-05,
+      "loss": 0.0042,
+      "step": 7791
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.141354971911461e-05,
+      "loss": 0.0024,
+      "step": 7792
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.141114898929275e-05,
+      "loss": 0.0017,
+      "step": 7793
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1408748259470885e-05,
+      "loss": 0.0043,
+      "step": 7794
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140634752964902e-05,
+      "loss": 0.0029,
+      "step": 7795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140394679982715e-05,
+      "loss": 0.0034,
+      "step": 7796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.140154607000528e-05,
+      "loss": 0.0023,
+      "step": 7797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.139914534018342e-05,
+      "loss": 0.0027,
+      "step": 7798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.139674461036155e-05,
+      "loss": 0.0023,
+      "step": 7799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.139434388053968e-05,
+      "loss": 0.0033,
+      "step": 7800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1391943150717815e-05,
+      "loss": 0.0035,
+      "step": 7801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.138954242089595e-05,
+      "loss": 0.5804,
+      "step": 7802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.138714169107409e-05,
+      "loss": 0.0031,
+      "step": 7803
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1384740961252224e-05,
+      "loss": 0.0029,
+      "step": 7804
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1382340231430356e-05,
+      "loss": 0.0035,
+      "step": 7805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137993950160849e-05,
+      "loss": 0.0024,
+      "step": 7806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1377538771786626e-05,
+      "loss": 0.0028,
+      "step": 7807
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137513804196476e-05,
+      "loss": 0.0018,
+      "step": 7808
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137273731214289e-05,
+      "loss": 0.7071,
+      "step": 7809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137033658232103e-05,
+      "loss": 0.0038,
+      "step": 7810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136793585249917e-05,
+      "loss": 0.0021,
+      "step": 7811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.13655351226773e-05,
+      "loss": 0.0027,
+      "step": 7812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136313439285543e-05,
+      "loss": 0.0027,
+      "step": 7813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136073366303356e-05,
+      "loss": 0.0035,
+      "step": 7814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.13583329332117e-05,
+      "loss": 0.6754,
+      "step": 7815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.135593220338983e-05,
+      "loss": 0.0041,
+      "step": 7816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1353531473567964e-05,
+      "loss": 0.0025,
+      "step": 7817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1351130743746096e-05,
+      "loss": 0.6413,
+      "step": 7818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1348730013924235e-05,
+      "loss": 0.0041,
+      "step": 7819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1346329284102366e-05,
+      "loss": 0.0043,
+      "step": 7820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1343928554280505e-05,
+      "loss": 0.0044,
+      "step": 7821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.134152782445864e-05,
+      "loss": 0.0029,
+      "step": 7822
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1339127094636775e-05,
+      "loss": 0.0031,
+      "step": 7823
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.133672636481491e-05,
+      "loss": 0.0034,
+      "step": 7824
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.133432563499304e-05,
+      "loss": 0.003,
+      "step": 7825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.133192490517117e-05,
+      "loss": 0.0048,
+      "step": 7826
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.132952417534931e-05,
+      "loss": 0.0028,
+      "step": 7827
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.132712344552744e-05,
+      "loss": 0.0022,
+      "step": 7828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.132472271570557e-05,
+      "loss": 0.0028,
+      "step": 7829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.132232198588371e-05,
+      "loss": 0.0032,
+      "step": 7830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131992125606184e-05,
+      "loss": 0.0013,
+      "step": 7831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131752052623998e-05,
+      "loss": 0.0036,
+      "step": 7832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1315119796418114e-05,
+      "loss": 0.0051,
+      "step": 7833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1312719066596246e-05,
+      "loss": 0.005,
+      "step": 7834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.131031833677438e-05,
+      "loss": 0.0029,
+      "step": 7835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1307917606952516e-05,
+      "loss": 0.0039,
+      "step": 7836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130551687713065e-05,
+      "loss": 0.0018,
+      "step": 7837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130311614730878e-05,
+      "loss": 0.0031,
+      "step": 7838
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130071541748692e-05,
+      "loss": 0.0026,
+      "step": 7839
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129831468766506e-05,
+      "loss": 0.557,
+      "step": 7840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129591395784319e-05,
+      "loss": 0.0023,
+      "step": 7841
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129351322802132e-05,
+      "loss": 0.0029,
+      "step": 7842
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.129111249819945e-05,
+      "loss": 0.0055,
+      "step": 7843
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.128871176837759e-05,
+      "loss": 0.0049,
+      "step": 7844
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.128631103855572e-05,
+      "loss": 0.0022,
+      "step": 7845
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1283910308733854e-05,
+      "loss": 0.0033,
+      "step": 7846
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1281509578911986e-05,
+      "loss": 0.0023,
+      "step": 7847
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1279108849090125e-05,
+      "loss": 0.0033,
+      "step": 7848
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.127670811926826e-05,
+      "loss": 0.003,
+      "step": 7849
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1274307389446395e-05,
+      "loss": 0.0018,
+      "step": 7850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.127190665962453e-05,
+      "loss": 0.6701,
+      "step": 7851
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.126950592980266e-05,
+      "loss": 0.003,
+      "step": 7852
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.12671051999808e-05,
+      "loss": 0.0031,
+      "step": 7853
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.126470447015893e-05,
+      "loss": 0.6328,
+      "step": 7854
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.126230374033706e-05,
+      "loss": 0.0037,
+      "step": 7855
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125990301051519e-05,
+      "loss": 0.0048,
+      "step": 7856
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125750228069334e-05,
+      "loss": 0.0018,
+      "step": 7857
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125510155087147e-05,
+      "loss": 0.0042,
+      "step": 7858
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.12527008210496e-05,
+      "loss": 0.0024,
+      "step": 7859
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.125030009122773e-05,
+      "loss": 0.0034,
+      "step": 7860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.124789936140587e-05,
+      "loss": 0.0025,
+      "step": 7861
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1245498631584004e-05,
+      "loss": 0.0028,
+      "step": 7862
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1243097901762136e-05,
+      "loss": 0.0039,
+      "step": 7863
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.124069717194027e-05,
+      "loss": 0.0037,
+      "step": 7864
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1238296442118406e-05,
+      "loss": 0.0038,
+      "step": 7865
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1235895712296544e-05,
+      "loss": 0.002,
+      "step": 7866
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1233494982474676e-05,
+      "loss": 0.0017,
+      "step": 7867
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.123109425265281e-05,
+      "loss": 0.0028,
+      "step": 7868
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122869352283094e-05,
+      "loss": 0.0019,
+      "step": 7869
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122629279300908e-05,
+      "loss": 0.0023,
+      "step": 7870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122389206318721e-05,
+      "loss": 0.0034,
+      "step": 7871
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122149133336534e-05,
+      "loss": 0.0029,
+      "step": 7872
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1219090603543474e-05,
+      "loss": 0.0024,
+      "step": 7873
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121668987372161e-05,
+      "loss": 0.5531,
+      "step": 7874
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121428914389975e-05,
+      "loss": 0.0026,
+      "step": 7875
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121188841407788e-05,
+      "loss": 0.6997,
+      "step": 7876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1209487684256015e-05,
+      "loss": 0.0036,
+      "step": 7877
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.120708695443415e-05,
+      "loss": 0.0027,
+      "step": 7878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1204686224612285e-05,
+      "loss": 0.0027,
+      "step": 7879
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.120228549479042e-05,
+      "loss": 0.6285,
+      "step": 7880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119988476496855e-05,
+      "loss": 0.003,
+      "step": 7881
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119748403514669e-05,
+      "loss": 0.0027,
+      "step": 7882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119508330532482e-05,
+      "loss": 0.0029,
+      "step": 7883
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119268257550295e-05,
+      "loss": 0.0031,
+      "step": 7884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119028184568109e-05,
+      "loss": 0.0024,
+      "step": 7885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.118788111585922e-05,
+      "loss": 0.0037,
+      "step": 7886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.118548038603736e-05,
+      "loss": 0.0039,
+      "step": 7887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.118307965621549e-05,
+      "loss": 0.0018,
+      "step": 7888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.118067892639362e-05,
+      "loss": 0.0015,
+      "step": 7889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.117827819657176e-05,
+      "loss": 0.0027,
+      "step": 7890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1175877466749894e-05,
+      "loss": 0.0045,
+      "step": 7891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1173476736928025e-05,
+      "loss": 0.0023,
+      "step": 7892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.117107600710616e-05,
+      "loss": 0.0032,
+      "step": 7893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1168675277284296e-05,
+      "loss": 0.0035,
+      "step": 7894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1166274547462434e-05,
+      "loss": 0.0016,
+      "step": 7895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1163873817640566e-05,
+      "loss": 0.0031,
+      "step": 7896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.11614730878187e-05,
+      "loss": 0.0025,
+      "step": 7897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.115907235799683e-05,
+      "loss": 0.0012,
+      "step": 7898
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.115667162817497e-05,
+      "loss": 0.0024,
+      "step": 7899
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11542708983531e-05,
+      "loss": 0.002,
+      "step": 7900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.115187016853123e-05,
+      "loss": 0.0019,
+      "step": 7901
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1149469438709364e-05,
+      "loss": 0.5575,
+      "step": 7902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.11470687088875e-05,
+      "loss": 0.0047,
+      "step": 7903
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114466797906564e-05,
+      "loss": 0.0015,
+      "step": 7904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.114226724924377e-05,
+      "loss": 0.0023,
+      "step": 7905
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1139866519421905e-05,
+      "loss": 0.0023,
+      "step": 7906
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113746578960004e-05,
+      "loss": 0.0016,
+      "step": 7907
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1135065059778175e-05,
+      "loss": 0.0058,
+      "step": 7908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113266432995631e-05,
+      "loss": 0.0035,
+      "step": 7909
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113026360013444e-05,
+      "loss": 0.0017,
+      "step": 7910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112786287031258e-05,
+      "loss": 0.0022,
+      "step": 7911
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1125462140490716e-05,
+      "loss": 0.0034,
+      "step": 7912
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112306141066885e-05,
+      "loss": 0.0042,
+      "step": 7913
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.112066068084698e-05,
+      "loss": 0.0022,
+      "step": 7914
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111825995102511e-05,
+      "loss": 0.0017,
+      "step": 7915
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111585922120325e-05,
+      "loss": 0.0025,
+      "step": 7916
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111345849138138e-05,
+      "loss": 0.0043,
+      "step": 7917
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.111105776155951e-05,
+      "loss": 0.0021,
+      "step": 7918
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1108657031737645e-05,
+      "loss": 0.0046,
+      "step": 7919
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1106256301915784e-05,
+      "loss": 0.0014,
+      "step": 7920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.110385557209392e-05,
+      "loss": 0.7086,
+      "step": 7921
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1101454842272054e-05,
+      "loss": 0.0016,
+      "step": 7922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1099054112450186e-05,
+      "loss": 0.0057,
+      "step": 7923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1096653382628324e-05,
+      "loss": 0.9671,
+      "step": 7924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1094252652806456e-05,
+      "loss": 0.0027,
+      "step": 7925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.109185192298459e-05,
+      "loss": 0.0021,
+      "step": 7926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108945119316272e-05,
+      "loss": 0.0016,
+      "step": 7927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108705046334086e-05,
+      "loss": 0.9344,
+      "step": 7928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108464973351899e-05,
+      "loss": 0.0018,
+      "step": 7929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108224900369713e-05,
+      "loss": 0.0052,
+      "step": 7930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107984827387526e-05,
+      "loss": 0.0018,
+      "step": 7931
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107744754405339e-05,
+      "loss": 0.0038,
+      "step": 7932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107504681423153e-05,
+      "loss": 0.003,
+      "step": 7933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.107264608440966e-05,
+      "loss": 0.0013,
+      "step": 7934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1070245354587795e-05,
+      "loss": 0.0057,
+      "step": 7935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1067844624765926e-05,
+      "loss": 0.0023,
+      "step": 7936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1065443894944065e-05,
+      "loss": 0.6724,
+      "step": 7937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.10630431651222e-05,
+      "loss": 0.0029,
+      "step": 7938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.106064243530033e-05,
+      "loss": 0.0013,
+      "step": 7939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105824170547847e-05,
+      "loss": 0.002,
+      "step": 7940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1055840975656606e-05,
+      "loss": 0.0039,
+      "step": 7941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105344024583474e-05,
+      "loss": 0.0023,
+      "step": 7942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105103951601287e-05,
+      "loss": 0.0019,
+      "step": 7943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1048638786191e-05,
+      "loss": 0.0016,
+      "step": 7944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.104623805636914e-05,
+      "loss": 0.0032,
+      "step": 7945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.104383732654727e-05,
+      "loss": 0.0023,
+      "step": 7946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.10414365967254e-05,
+      "loss": 0.5552,
+      "step": 7947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1039035866903535e-05,
+      "loss": 0.0027,
+      "step": 7948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1036635137081674e-05,
+      "loss": 0.0019,
+      "step": 7949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.103423440725981e-05,
+      "loss": 0.0018,
+      "step": 7950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1031833677437944e-05,
+      "loss": 0.3772,
+      "step": 7951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1029432947616076e-05,
+      "loss": 0.0025,
+      "step": 7952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102703221779421e-05,
+      "loss": 0.0023,
+      "step": 7953
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1024631487972346e-05,
+      "loss": 0.0036,
+      "step": 7954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102223075815048e-05,
+      "loss": 0.0031,
+      "step": 7955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101983002832861e-05,
+      "loss": 0.0035,
+      "step": 7956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101742929850674e-05,
+      "loss": 0.0032,
+      "step": 7957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101502856868489e-05,
+      "loss": 0.0043,
+      "step": 7958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101262783886302e-05,
+      "loss": 0.6683,
+      "step": 7959
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101022710904115e-05,
+      "loss": 0.002,
+      "step": 7960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.100782637921928e-05,
+      "loss": 0.0028,
+      "step": 7961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.100542564939742e-05,
+      "loss": 0.0026,
+      "step": 7962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.100302491957555e-05,
+      "loss": 0.0025,
+      "step": 7963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1000624189753685e-05,
+      "loss": 0.0033,
+      "step": 7964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0998223459931816e-05,
+      "loss": 0.0019,
+      "step": 7965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0995822730109955e-05,
+      "loss": 0.0012,
+      "step": 7966
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0993422000288093e-05,
+      "loss": 0.4657,
+      "step": 7967
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0991021270466225e-05,
+      "loss": 0.7492,
+      "step": 7968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098862054064436e-05,
+      "loss": 0.0014,
+      "step": 7969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098621981082249e-05,
+      "loss": 0.0039,
+      "step": 7970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098381908100063e-05,
+      "loss": 0.0015,
+      "step": 7971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.098141835117876e-05,
+      "loss": 0.0082,
+      "step": 7972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097901762135689e-05,
+      "loss": 0.0026,
+      "step": 7973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097661689153503e-05,
+      "loss": 0.002,
+      "step": 7974
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.097421616171316e-05,
+      "loss": 0.0027,
+      "step": 7975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09718154318913e-05,
+      "loss": 0.0076,
+      "step": 7976
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.096941470206943e-05,
+      "loss": 0.7423,
+      "step": 7977
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0967013972247564e-05,
+      "loss": 0.0041,
+      "step": 7978
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09646132424257e-05,
+      "loss": 0.0051,
+      "step": 7979
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0962212512603834e-05,
+      "loss": 0.0098,
+      "step": 7980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0959811782781966e-05,
+      "loss": 0.0091,
+      "step": 7981
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.09574110529601e-05,
+      "loss": 0.0087,
+      "step": 7982
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0955010323138236e-05,
+      "loss": 0.0064,
+      "step": 7983
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.095260959331637e-05,
+      "loss": 0.0108,
+      "step": 7984
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0950208863494507e-05,
+      "loss": 0.0147,
+      "step": 7985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094780813367264e-05,
+      "loss": 0.0103,
+      "step": 7986
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094540740385078e-05,
+      "loss": 0.6801,
+      "step": 7987
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094300667402891e-05,
+      "loss": 0.0055,
+      "step": 7988
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.094060594420704e-05,
+      "loss": 0.0063,
+      "step": 7989
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093820521438517e-05,
+      "loss": 0.0056,
+      "step": 7990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093580448456331e-05,
+      "loss": 0.0049,
+      "step": 7991
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093340375474144e-05,
+      "loss": 0.0052,
+      "step": 7992
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0931003024919574e-05,
+      "loss": 0.0048,
+      "step": 7993
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.092860229509771e-05,
+      "loss": 0.004,
+      "step": 7994
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0926201565275845e-05,
+      "loss": 0.0043,
+      "step": 7995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0923800835453983e-05,
+      "loss": 0.0051,
+      "step": 7996
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0921400105632115e-05,
+      "loss": 0.0044,
+      "step": 7997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.091899937581025e-05,
+      "loss": 0.004,
+      "step": 7998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.091659864598838e-05,
+      "loss": 0.0038,
+      "step": 7999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.091419791616652e-05,
+      "loss": 0.0038,
+      "step": 8000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.091179718634465e-05,
+      "loss": 0.0042,
+      "step": 8001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090939645652278e-05,
+      "loss": 0.0041,
+      "step": 8002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090699572670091e-05,
+      "loss": 0.7455,
+      "step": 8003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090459499687906e-05,
+      "loss": 0.0035,
+      "step": 8004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090219426705719e-05,
+      "loss": 0.0038,
+      "step": 8005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.089979353723532e-05,
+      "loss": 0.0037,
+      "step": 8006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0897392807413454e-05,
+      "loss": 0.0032,
+      "step": 8007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.089499207759159e-05,
+      "loss": 0.003,
+      "step": 8008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0892591347769724e-05,
+      "loss": 0.0031,
+      "step": 8009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0890190617947856e-05,
+      "loss": 0.0028,
+      "step": 8010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.088778988812599e-05,
+      "loss": 0.0026,
+      "step": 8011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0885389158304126e-05,
+      "loss": 0.003,
+      "step": 8012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0882988428482265e-05,
+      "loss": 0.0028,
+      "step": 8013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0880587698660396e-05,
+      "loss": 0.0031,
+      "step": 8014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087818696883853e-05,
+      "loss": 0.0026,
+      "step": 8015
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087578623901666e-05,
+      "loss": 0.0027,
+      "step": 8016
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.08733855091948e-05,
+      "loss": 0.0024,
+      "step": 8017
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087098477937293e-05,
+      "loss": 0.0024,
+      "step": 8018
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086858404955106e-05,
+      "loss": 0.0022,
+      "step": 8019
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0866183319729194e-05,
+      "loss": 0.0023,
+      "step": 8020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086378258990733e-05,
+      "loss": 0.0024,
+      "step": 8021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086138186008547e-05,
+      "loss": 0.0023,
+      "step": 8022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.08589811302636e-05,
+      "loss": 0.0024,
+      "step": 8023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0856580400441735e-05,
+      "loss": 0.0022,
+      "step": 8024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0854179670619873e-05,
+      "loss": 0.7162,
+      "step": 8025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0851778940798005e-05,
+      "loss": 0.0019,
+      "step": 8026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.084937821097614e-05,
+      "loss": 0.0021,
+      "step": 8027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.084697748115427e-05,
+      "loss": 0.0015,
+      "step": 8028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.084457675133241e-05,
+      "loss": 0.0018,
+      "step": 8029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.084217602151054e-05,
+      "loss": 0.0024,
+      "step": 8030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083977529168868e-05,
+      "loss": 0.0018,
+      "step": 8031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083737456186681e-05,
+      "loss": 0.0022,
+      "step": 8032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083497383204494e-05,
+      "loss": 0.0017,
+      "step": 8033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083257310222308e-05,
+      "loss": 0.0019,
+      "step": 8034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083017237240121e-05,
+      "loss": 0.0019,
+      "step": 8035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0827771642579344e-05,
+      "loss": 0.0015,
+      "step": 8036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0825370912757475e-05,
+      "loss": 0.0017,
+      "step": 8037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0822970182935614e-05,
+      "loss": 0.0019,
+      "step": 8038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0820569453113746e-05,
+      "loss": 0.0015,
+      "step": 8039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0818168723291884e-05,
+      "loss": 0.0013,
+      "step": 8040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0815767993470016e-05,
+      "loss": 0.0018,
+      "step": 8041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0813367263648155e-05,
+      "loss": 0.0016,
+      "step": 8042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0810966533826286e-05,
+      "loss": 0.0013,
+      "step": 8043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080856580400442e-05,
+      "loss": 0.8058,
+      "step": 8044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080616507418255e-05,
+      "loss": 0.0018,
+      "step": 8045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080376434436069e-05,
+      "loss": 0.0016,
+      "step": 8046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.080136361453882e-05,
+      "loss": 0.0018,
+      "step": 8047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079896288471695e-05,
+      "loss": 0.6247,
+      "step": 8048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079656215489509e-05,
+      "loss": 0.0015,
+      "step": 8049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079416142507322e-05,
+      "loss": 0.0018,
+      "step": 8050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.079176069525136e-05,
+      "loss": 1.6042,
+      "step": 8051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.078935996542949e-05,
+      "loss": 0.0018,
+      "step": 8052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0786959235607625e-05,
+      "loss": 0.0019,
+      "step": 8053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0784558505785757e-05,
+      "loss": 0.0018,
+      "step": 8054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0782157775963895e-05,
+      "loss": 0.0019,
+      "step": 8055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.077975704614203e-05,
+      "loss": 0.0025,
+      "step": 8056
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.077735631632016e-05,
+      "loss": 0.0022,
+      "step": 8057
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.07749555864983e-05,
+      "loss": 0.0021,
+      "step": 8058
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0772554856676436e-05,
+      "loss": 0.002,
+      "step": 8059
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.077015412685457e-05,
+      "loss": 0.0015,
+      "step": 8060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.07677533970327e-05,
+      "loss": 0.0021,
+      "step": 8061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.076535266721083e-05,
+      "loss": 0.0018,
+      "step": 8062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.076295193738897e-05,
+      "loss": 0.0023,
+      "step": 8063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.07605512075671e-05,
+      "loss": 0.0017,
+      "step": 8064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0758150477745234e-05,
+      "loss": 0.0019,
+      "step": 8065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0755749747923365e-05,
+      "loss": 0.002,
+      "step": 8066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0753349018101504e-05,
+      "loss": 0.0017,
+      "step": 8067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.075094828827964e-05,
+      "loss": 0.0024,
+      "step": 8068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0748547558457774e-05,
+      "loss": 0.0021,
+      "step": 8069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0746146828635906e-05,
+      "loss": 0.0019,
+      "step": 8070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0743746098814045e-05,
+      "loss": 0.0017,
+      "step": 8071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0741345368992176e-05,
+      "loss": 0.0013,
+      "step": 8072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073894463917031e-05,
+      "loss": 0.0016,
+      "step": 8073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073654390934844e-05,
+      "loss": 0.0019,
+      "step": 8074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073414317952658e-05,
+      "loss": 0.0017,
+      "step": 8075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073174244970471e-05,
+      "loss": 0.0014,
+      "step": 8076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072934171988285e-05,
+      "loss": 0.0017,
+      "step": 8077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072694099006098e-05,
+      "loss": 0.0018,
+      "step": 8078
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072454026023911e-05,
+      "loss": 0.0014,
+      "step": 8079
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.072213953041725e-05,
+      "loss": 0.0016,
+      "step": 8080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071973880059538e-05,
+      "loss": 0.0015,
+      "step": 8081
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0717338070773515e-05,
+      "loss": 0.0015,
+      "step": 8082
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0714937340951647e-05,
+      "loss": 0.0018,
+      "step": 8083
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0712536611129785e-05,
+      "loss": 0.0014,
+      "step": 8084
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071013588130792e-05,
+      "loss": 0.0016,
+      "step": 8085
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0707735151486056e-05,
+      "loss": 0.0012,
+      "step": 8086
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070533442166419e-05,
+      "loss": 0.0012,
+      "step": 8087
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0702933691842326e-05,
+      "loss": 0.0011,
+      "step": 8088
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070053296202046e-05,
+      "loss": 0.0013,
+      "step": 8089
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069813223219859e-05,
+      "loss": 0.0013,
+      "step": 8090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069573150237672e-05,
+      "loss": 0.0016,
+      "step": 8091
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069333077255486e-05,
+      "loss": 0.0015,
+      "step": 8092
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.069093004273299e-05,
+      "loss": 0.0011,
+      "step": 8093
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0688529312911123e-05,
+      "loss": 0.001,
+      "step": 8094
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.068612858308926e-05,
+      "loss": 0.0014,
+      "step": 8095
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0683727853267394e-05,
+      "loss": 0.0012,
+      "step": 8096
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.068132712344553e-05,
+      "loss": 0.0013,
+      "step": 8097
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0678926393623664e-05,
+      "loss": 0.001,
+      "step": 8098
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0676525663801796e-05,
+      "loss": 0.0015,
+      "step": 8099
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.067412493397993e-05,
+      "loss": 0.0011,
+      "step": 8100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0671724204158066e-05,
+      "loss": 0.0013,
+      "step": 8101
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.06693234743362e-05,
+      "loss": 0.8234,
+      "step": 8102
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.066692274451433e-05,
+      "loss": 0.0012,
+      "step": 8103
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.066452201469247e-05,
+      "loss": 0.0011,
+      "step": 8104
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.066212128487061e-05,
+      "loss": 0.001,
+      "step": 8105
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.065972055504874e-05,
+      "loss": 0.0012,
+      "step": 8106
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.065731982522687e-05,
+      "loss": 0.0055,
+      "step": 8107
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0654919095405e-05,
+      "loss": 0.0011,
+      "step": 8108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.065251836558314e-05,
+      "loss": 0.001,
+      "step": 8109
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.065011763576127e-05,
+      "loss": 0.0014,
+      "step": 8110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0647716905939405e-05,
+      "loss": 0.0011,
+      "step": 8111
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0645316176117537e-05,
+      "loss": 0.001,
+      "step": 8112
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0642915446295675e-05,
+      "loss": 0.0014,
+      "step": 8113
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0640514716473814e-05,
+      "loss": 0.0011,
+      "step": 8114
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0638113986651945e-05,
+      "loss": 0.0007,
+      "step": 8115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063571325683008e-05,
+      "loss": 0.001,
+      "step": 8116
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063331252700821e-05,
+      "loss": 0.0009,
+      "step": 8117
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.063091179718635e-05,
+      "loss": 0.001,
+      "step": 8118
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062851106736448e-05,
+      "loss": 0.0014,
+      "step": 8119
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062611033754261e-05,
+      "loss": 0.0015,
+      "step": 8120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062370960772074e-05,
+      "loss": 0.0013,
+      "step": 8121
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.062130887789889e-05,
+      "loss": 0.0014,
+      "step": 8122
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.061890814807702e-05,
+      "loss": 0.0008,
+      "step": 8123
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.061650741825515e-05,
+      "loss": 0.0008,
+      "step": 8124
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0614106688433284e-05,
+      "loss": 0.0008,
+      "step": 8125
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.061170595861142e-05,
+      "loss": 1.4751,
+      "step": 8126
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0609305228789554e-05,
+      "loss": 0.0009,
+      "step": 8127
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0606904498967686e-05,
+      "loss": 0.0009,
+      "step": 8128
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.060450376914582e-05,
+      "loss": 0.0008,
+      "step": 8129
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0602103039323956e-05,
+      "loss": 0.0009,
+      "step": 8130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059970230950209e-05,
+      "loss": 0.0009,
+      "step": 8131
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059730157968023e-05,
+      "loss": 0.751,
+      "step": 8132
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059490084985836e-05,
+      "loss": 0.0009,
+      "step": 8133
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059250012003649e-05,
+      "loss": 0.0011,
+      "step": 8134
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.059009939021463e-05,
+      "loss": 0.001,
+      "step": 8135
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058769866039276e-05,
+      "loss": 0.001,
+      "step": 8136
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058529793057089e-05,
+      "loss": 0.0011,
+      "step": 8137
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058289720074903e-05,
+      "loss": 0.0012,
+      "step": 8138
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.058049647092716e-05,
+      "loss": 0.7016,
+      "step": 8139
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0578095741105295e-05,
+      "loss": 0.6719,
+      "step": 8140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057569501128343e-05,
+      "loss": 0.001,
+      "step": 8141
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0573294281461565e-05,
+      "loss": 0.0013,
+      "step": 8142
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0570893551639704e-05,
+      "loss": 0.0008,
+      "step": 8143
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0568492821817835e-05,
+      "loss": 0.001,
+      "step": 8144
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056609209199597e-05,
+      "loss": 0.0008,
+      "step": 8145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.05636913621741e-05,
+      "loss": 0.0019,
+      "step": 8146
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056129063235224e-05,
+      "loss": 0.0017,
+      "step": 8147
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.055888990253037e-05,
+      "loss": 0.001,
+      "step": 8148
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.05564891727085e-05,
+      "loss": 0.001,
+      "step": 8149
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.055408844288664e-05,
+      "loss": 0.0011,
+      "step": 8150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.055168771306477e-05,
+      "loss": 0.0014,
+      "step": 8151
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054928698324291e-05,
+      "loss": 0.001,
+      "step": 8152
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054688625342104e-05,
+      "loss": 0.0009,
+      "step": 8153
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0544485523599174e-05,
+      "loss": 0.0011,
+      "step": 8154
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.054208479377731e-05,
+      "loss": 0.0011,
+      "step": 8155
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0539684063955444e-05,
+      "loss": 0.0016,
+      "step": 8156
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0537283334133576e-05,
+      "loss": 0.0018,
+      "step": 8157
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.053488260431171e-05,
+      "loss": 0.0007,
+      "step": 8158
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0532481874489846e-05,
+      "loss": 0.001,
+      "step": 8159
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0530081144667985e-05,
+      "loss": 0.0018,
+      "step": 8160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052768041484612e-05,
+      "loss": 0.001,
+      "step": 8161
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052527968502425e-05,
+      "loss": 0.0014,
+      "step": 8162
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052287895520238e-05,
+      "loss": 0.6158,
+      "step": 8163
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052047822538052e-05,
+      "loss": 0.0015,
+      "step": 8164
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.051807749555865e-05,
+      "loss": 0.0009,
+      "step": 8165
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0515676765736783e-05,
+      "loss": 0.002,
+      "step": 8166
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0513276035914918e-05,
+      "loss": 0.0009,
+      "step": 8167
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0510875306093056e-05,
+      "loss": 0.0011,
+      "step": 8168
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.050847457627119e-05,
+      "loss": 0.0009,
+      "step": 8169
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0506073846449323e-05,
+      "loss": 0.0011,
+      "step": 8170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.050367311662746e-05,
+      "loss": 0.0009,
+      "step": 8171
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.050127238680559e-05,
+      "loss": 0.0007,
+      "step": 8172
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0498871656983725e-05,
+      "loss": 0.001,
+      "step": 8173
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0496470927161857e-05,
+      "loss": 0.0008,
+      "step": 8174
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0494070197339992e-05,
+      "loss": 0.0006,
+      "step": 8175
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0491669467518124e-05,
+      "loss": 0.0009,
+      "step": 8176
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0489268737696263e-05,
+      "loss": 0.0006,
+      "step": 8177
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0486868007874398e-05,
+      "loss": 0.0016,
+      "step": 8178
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.048446727805253e-05,
+      "loss": 0.001,
+      "step": 8179
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0482066548230665e-05,
+      "loss": 0.0007,
+      "step": 8180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0479665818408797e-05,
+      "loss": 0.0009,
+      "step": 8181
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0477265088586932e-05,
+      "loss": 0.0007,
+      "step": 8182
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0474864358765064e-05,
+      "loss": 0.0007,
+      "step": 8183
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.04724636289432e-05,
+      "loss": 0.0007,
+      "step": 8184
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.047006289912133e-05,
+      "loss": 0.0007,
+      "step": 8185
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0467662169299473e-05,
+      "loss": 0.0009,
+      "step": 8186
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0465261439477605e-05,
+      "loss": 0.0007,
+      "step": 8187
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.046286070965574e-05,
+      "loss": 0.0009,
+      "step": 8188
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.046045997983387e-05,
+      "loss": 0.0006,
+      "step": 8189
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0458059250012007e-05,
+      "loss": 0.0011,
+      "step": 8190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.045565852019014e-05,
+      "loss": 0.0007,
+      "step": 8191
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0453257790368274e-05,
+      "loss": 0.0007,
+      "step": 8192
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0450857060546405e-05,
+      "loss": 0.0007,
+      "step": 8193
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.044845633072454e-05,
+      "loss": 0.0006,
+      "step": 8194
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0446055600902672e-05,
+      "loss": 0.0006,
+      "step": 8195
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.044365487108081e-05,
+      "loss": 0.0006,
+      "step": 8196
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0441254141258946e-05,
+      "loss": 0.0006,
+      "step": 8197
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0438853411437078e-05,
+      "loss": 0.0006,
+      "step": 8198
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0436452681615213e-05,
+      "loss": 0.6723,
+      "step": 8199
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0434051951793345e-05,
+      "loss": 0.0006,
+      "step": 8200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.043165122197148e-05,
+      "loss": 0.0005,
+      "step": 8201
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0429250492149612e-05,
+      "loss": 0.0006,
+      "step": 8202
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0426849762327747e-05,
+      "loss": 0.0005,
+      "step": 8203
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.042444903250588e-05,
+      "loss": 0.0007,
+      "step": 8204
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.042204830268402e-05,
+      "loss": 0.0005,
+      "step": 8205
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0419647572862153e-05,
+      "loss": 0.0006,
+      "step": 8206
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0417246843040288e-05,
+      "loss": 0.0005,
+      "step": 8207
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.041484611321842e-05,
+      "loss": 0.0005,
+      "step": 8208
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0412445383396555e-05,
+      "loss": 0.0007,
+      "step": 8209
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0410044653574687e-05,
+      "loss": 0.0006,
+      "step": 8210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0407643923752822e-05,
+      "loss": 0.0006,
+      "step": 8211
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0405243193930954e-05,
+      "loss": 0.9446,
+      "step": 8212
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.040284246410909e-05,
+      "loss": 0.0006,
+      "step": 8213
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0400441734287228e-05,
+      "loss": 0.9965,
+      "step": 8214
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.039804100446536e-05,
+      "loss": 0.0005,
+      "step": 8215
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0395640274643494e-05,
+      "loss": 0.0009,
+      "step": 8216
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0393239544821626e-05,
+      "loss": 0.0227,
+      "step": 8217
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.039083881499976e-05,
+      "loss": 0.0008,
+      "step": 8218
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0388438085177893e-05,
+      "loss": 0.0007,
+      "step": 8219
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.038603735535603e-05,
+      "loss": 0.0004,
+      "step": 8220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0383636625534164e-05,
+      "loss": 0.0014,
+      "step": 8221
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0381235895712295e-05,
+      "loss": 0.0011,
+      "step": 8222
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0378835165890434e-05,
+      "loss": 0.0037,
+      "step": 8223
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.037643443606857e-05,
+      "loss": 0.0015,
+      "step": 8224
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.03740337062467e-05,
+      "loss": 0.0015,
+      "step": 8225
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0371632976424836e-05,
+      "loss": 0.0016,
+      "step": 8226
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0369232246602968e-05,
+      "loss": 0.0011,
+      "step": 8227
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0366831516781103e-05,
+      "loss": 0.0006,
+      "step": 8228
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0364430786959235e-05,
+      "loss": 0.0006,
+      "step": 8229
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.036203005713737e-05,
+      "loss": 0.001,
+      "step": 8230
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0359629327315502e-05,
+      "loss": 0.0007,
+      "step": 8231
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.035722859749364e-05,
+      "loss": 0.001,
+      "step": 8232
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0354827867671776e-05,
+      "loss": 0.6481,
+      "step": 8233
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0352427137849908e-05,
+      "loss": 0.0008,
+      "step": 8234
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0350026408028043e-05,
+      "loss": 0.0014,
+      "step": 8235
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0347625678206178e-05,
+      "loss": 0.0008,
+      "step": 8236
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.034522494838431e-05,
+      "loss": 0.0012,
+      "step": 8237
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0342824218562445e-05,
+      "loss": 0.0007,
+      "step": 8238
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0340423488740577e-05,
+      "loss": 0.0009,
+      "step": 8239
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0338022758918712e-05,
+      "loss": 0.0009,
+      "step": 8240
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.033562202909685e-05,
+      "loss": 0.0008,
+      "step": 8241
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0333221299274982e-05,
+      "loss": 0.0011,
+      "step": 8242
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0330820569453117e-05,
+      "loss": 0.9965,
+      "step": 8243
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.032841983963125e-05,
+      "loss": 0.0017,
+      "step": 8244
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0326019109809384e-05,
+      "loss": 0.0021,
+      "step": 8245
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0323618379987516e-05,
+      "loss": 0.0021,
+      "step": 8246
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.032121765016565e-05,
+      "loss": 0.0031,
+      "step": 8247
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0318816920343783e-05,
+      "loss": 0.0028,
+      "step": 8248
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.031641619052192e-05,
+      "loss": 0.0027,
+      "step": 8249
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.031401546070005e-05,
+      "loss": 0.0031,
+      "step": 8250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0311614730878192e-05,
+      "loss": 0.0026,
+      "step": 8251
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0309214001056324e-05,
+      "loss": 0.0032,
+      "step": 8252
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.030681327123446e-05,
+      "loss": 0.0021,
+      "step": 8253
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.030441254141259e-05,
+      "loss": 0.0026,
+      "step": 8254
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0302011811590726e-05,
+      "loss": 0.0021,
+      "step": 8255
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0299611081768858e-05,
+      "loss": 0.0027,
+      "step": 8256
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0297210351946993e-05,
+      "loss": 0.0027,
+      "step": 8257
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0294809622125125e-05,
+      "loss": 0.0021,
+      "step": 8258
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.029240889230326e-05,
+      "loss": 0.0025,
+      "step": 8259
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.02900081624814e-05,
+      "loss": 0.0021,
+      "step": 8260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.028760743265953e-05,
+      "loss": 0.0027,
+      "step": 8261
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0285206702837666e-05,
+      "loss": 0.0023,
+      "step": 8262
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0282805973015798e-05,
+      "loss": 0.0025,
+      "step": 8263
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0280405243193933e-05,
+      "loss": 0.0022,
+      "step": 8264
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0278004513372065e-05,
+      "loss": 0.0019,
+      "step": 8265
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.02756037835502e-05,
+      "loss": 0.0021,
+      "step": 8266
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.027320305372833e-05,
+      "loss": 0.002,
+      "step": 8267
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0270802323906467e-05,
+      "loss": 0.0019,
+      "step": 8268
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0268401594084605e-05,
+      "loss": 0.0018,
+      "step": 8269
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.026600086426274e-05,
+      "loss": 0.0019,
+      "step": 8270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0263600134440872e-05,
+      "loss": 0.0017,
+      "step": 8271
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0261199404619007e-05,
+      "loss": 0.0016,
+      "step": 8272
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.025879867479714e-05,
+      "loss": 0.002,
+      "step": 8273
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0256397944975274e-05,
+      "loss": 0.0018,
+      "step": 8274
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0253997215153406e-05,
+      "loss": 0.0018,
+      "step": 8275
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.025159648533154e-05,
+      "loss": 0.0016,
+      "step": 8276
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0249195755509673e-05,
+      "loss": 0.0018,
+      "step": 8277
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0246795025687812e-05,
+      "loss": 0.0013,
+      "step": 8278
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0244394295865947e-05,
+      "loss": 0.0013,
+      "step": 8279
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.024199356604408e-05,
+      "loss": 0.0022,
+      "step": 8280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0239592836222214e-05,
+      "loss": 0.0015,
+      "step": 8281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0237192106400346e-05,
+      "loss": 0.7757,
+      "step": 8282
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.023479137657848e-05,
+      "loss": 0.8558,
+      "step": 8283
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0232390646756613e-05,
+      "loss": 0.0016,
+      "step": 8284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0229989916934748e-05,
+      "loss": 0.0014,
+      "step": 8285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022758918711288e-05,
+      "loss": 0.0016,
+      "step": 8286
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0225188457291022e-05,
+      "loss": 0.0013,
+      "step": 8287
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0222787727469154e-05,
+      "loss": 0.0012,
+      "step": 8288
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022038699764729e-05,
+      "loss": 0.0014,
+      "step": 8289
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.021798626782542e-05,
+      "loss": 0.0011,
+      "step": 8290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0215585538003556e-05,
+      "loss": 0.0013,
+      "step": 8291
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0213184808181688e-05,
+      "loss": 0.0014,
+      "step": 8292
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0210784078359823e-05,
+      "loss": 0.0014,
+      "step": 8293
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0208383348537954e-05,
+      "loss": 0.0015,
+      "step": 8294
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.020598261871609e-05,
+      "loss": 0.0015,
+      "step": 8295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0203581888894228e-05,
+      "loss": 0.0017,
+      "step": 8296
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.020118115907236e-05,
+      "loss": 0.0012,
+      "step": 8297
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0198780429250495e-05,
+      "loss": 0.0014,
+      "step": 8298
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0196379699428627e-05,
+      "loss": 0.0016,
+      "step": 8299
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0193978969606762e-05,
+      "loss": 0.0012,
+      "step": 8300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0191578239784894e-05,
+      "loss": 0.0011,
+      "step": 8301
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.018917750996303e-05,
+      "loss": 0.0011,
+      "step": 8302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0186776780141164e-05,
+      "loss": 0.0009,
+      "step": 8303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0184376050319296e-05,
+      "loss": 0.001,
+      "step": 8304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0181975320497435e-05,
+      "loss": 0.8276,
+      "step": 8305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.017957459067557e-05,
+      "loss": 0.0009,
+      "step": 8306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0177173860853702e-05,
+      "loss": 0.001,
+      "step": 8307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0174773131031837e-05,
+      "loss": 0.0012,
+      "step": 8308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.017237240120997e-05,
+      "loss": 0.6457,
+      "step": 8309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0169971671388104e-05,
+      "loss": 0.0012,
+      "step": 8310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0167570941566236e-05,
+      "loss": 0.0013,
+      "step": 8311
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.016517021174437e-05,
+      "loss": 0.0015,
+      "step": 8312
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0162769481922503e-05,
+      "loss": 0.0012,
+      "step": 8313
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0160368752100638e-05,
+      "loss": 0.0012,
+      "step": 8314
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0157968022278777e-05,
+      "loss": 0.6225,
+      "step": 8315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.015556729245691e-05,
+      "loss": 0.0014,
+      "step": 8316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0153166562635043e-05,
+      "loss": 0.0013,
+      "step": 8317
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.015076583281318e-05,
+      "loss": 0.0012,
+      "step": 8318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.014836510299131e-05,
+      "loss": 0.0015,
+      "step": 8319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0145964373169446e-05,
+      "loss": 0.0012,
+      "step": 8320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0143563643347577e-05,
+      "loss": 0.0012,
+      "step": 8321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0141162913525713e-05,
+      "loss": 0.6971,
+      "step": 8322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0138762183703844e-05,
+      "loss": 0.0014,
+      "step": 8323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0136361453881983e-05,
+      "loss": 0.0011,
+      "step": 8324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0133960724060118e-05,
+      "loss": 0.0012,
+      "step": 8325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.013155999423825e-05,
+      "loss": 0.0021,
+      "step": 8326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0129159264416385e-05,
+      "loss": 0.0015,
+      "step": 8327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0126758534594517e-05,
+      "loss": 0.0013,
+      "step": 8328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0124357804772652e-05,
+      "loss": 0.0012,
+      "step": 8329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0121957074950784e-05,
+      "loss": 0.0012,
+      "step": 8330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.011955634512892e-05,
+      "loss": 0.0012,
+      "step": 8331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.011715561530705e-05,
+      "loss": 0.7793,
+      "step": 8332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0114754885485193e-05,
+      "loss": 0.0012,
+      "step": 8333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0112354155663325e-05,
+      "loss": 0.0014,
+      "step": 8334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.010995342584146e-05,
+      "loss": 0.0015,
+      "step": 8335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0107552696019592e-05,
+      "loss": 0.0018,
+      "step": 8336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0105151966197727e-05,
+      "loss": 0.0015,
+      "step": 8337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.010275123637586e-05,
+      "loss": 0.002,
+      "step": 8338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0100350506553994e-05,
+      "loss": 0.0016,
+      "step": 8339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0097949776732126e-05,
+      "loss": 0.8494,
+      "step": 8340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.009554904691026e-05,
+      "loss": 0.0018,
+      "step": 8341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00931483170884e-05,
+      "loss": 0.0013,
+      "step": 8342
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.009074758726653e-05,
+      "loss": 0.774,
+      "step": 8343
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0088346857444666e-05,
+      "loss": 0.0015,
+      "step": 8344
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0085946127622798e-05,
+      "loss": 0.0038,
+      "step": 8345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0083545397800933e-05,
+      "loss": 0.0019,
+      "step": 8346
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0081144667979065e-05,
+      "loss": 0.0023,
+      "step": 8347
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00787439381572e-05,
+      "loss": 0.0017,
+      "step": 8348
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0076343208335332e-05,
+      "loss": 0.0017,
+      "step": 8349
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0073942478513467e-05,
+      "loss": 0.0019,
+      "step": 8350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0071541748691606e-05,
+      "loss": 0.0022,
+      "step": 8351
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.006914101886974e-05,
+      "loss": 0.002,
+      "step": 8352
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0066740289047873e-05,
+      "loss": 0.0022,
+      "step": 8353
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0064339559226008e-05,
+      "loss": 0.0019,
+      "step": 8354
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.006193882940414e-05,
+      "loss": 0.0016,
+      "step": 8355
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0059538099582275e-05,
+      "loss": 0.0021,
+      "step": 8356
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0057137369760407e-05,
+      "loss": 0.0018,
+      "step": 8357
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0054736639938542e-05,
+      "loss": 0.0016,
+      "step": 8358
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0052335910116674e-05,
+      "loss": 0.0016,
+      "step": 8359
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0049935180294813e-05,
+      "loss": 0.0016,
+      "step": 8360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0047534450472948e-05,
+      "loss": 0.0016,
+      "step": 8361
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.004513372065108e-05,
+      "loss": 0.0019,
+      "step": 8362
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0042732990829215e-05,
+      "loss": 0.0016,
+      "step": 8363
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0040332261007347e-05,
+      "loss": 0.0016,
+      "step": 8364
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0037931531185482e-05,
+      "loss": 0.0015,
+      "step": 8365
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0035530801363614e-05,
+      "loss": 0.0015,
+      "step": 8366
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.003313007154175e-05,
+      "loss": 0.0019,
+      "step": 8367
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.003072934171988e-05,
+      "loss": 0.0014,
+      "step": 8368
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0028328611898022e-05,
+      "loss": 0.0013,
+      "step": 8369
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0025927882076154e-05,
+      "loss": 0.0016,
+      "step": 8370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.002352715225429e-05,
+      "loss": 0.0027,
+      "step": 8371
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.002112642243242e-05,
+      "loss": 0.0013,
+      "step": 8372
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0018725692610556e-05,
+      "loss": 0.0012,
+      "step": 8373
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0016324962788688e-05,
+      "loss": 0.0014,
+      "step": 8374
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0013924232966823e-05,
+      "loss": 0.0012,
+      "step": 8375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0011523503144955e-05,
+      "loss": 0.0011,
+      "step": 8376
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.000912277332309e-05,
+      "loss": 0.0013,
+      "step": 8377
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0006722043501222e-05,
+      "loss": 0.0012,
+      "step": 8378
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.000432131367936e-05,
+      "loss": 0.0015,
+      "step": 8379
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0001920583857496e-05,
+      "loss": 0.0013,
+      "step": 8380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9999519854035628e-05,
+      "loss": 0.0013,
+      "step": 8381
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9997119124213763e-05,
+      "loss": 0.0011,
+      "step": 8382
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9994718394391895e-05,
+      "loss": 0.0011,
+      "step": 8383
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.999231766457003e-05,
+      "loss": 0.001,
+      "step": 8384
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9989916934748162e-05,
+      "loss": 0.0012,
+      "step": 8385
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9987516204926297e-05,
+      "loss": 0.0009,
+      "step": 8386
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9985115475104432e-05,
+      "loss": 0.001,
+      "step": 8387
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.998271474528257e-05,
+      "loss": 0.0013,
+      "step": 8388
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9980314015460703e-05,
+      "loss": 0.0012,
+      "step": 8389
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9977913285638838e-05,
+      "loss": 0.001,
+      "step": 8390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.997551255581697e-05,
+      "loss": 0.001,
+      "step": 8391
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9973111825995105e-05,
+      "loss": 0.6543,
+      "step": 8392
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9970711096173237e-05,
+      "loss": 0.001,
+      "step": 8393
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9968310366351372e-05,
+      "loss": 0.001,
+      "step": 8394
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9965909636529503e-05,
+      "loss": 0.0009,
+      "step": 8395
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.996350890670764e-05,
+      "loss": 0.0008,
+      "step": 8396
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9961108176885777e-05,
+      "loss": 0.8426,
+      "step": 8397
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.995870744706391e-05,
+      "loss": 0.0009,
+      "step": 8398
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9956306717242044e-05,
+      "loss": 0.0016,
+      "step": 8399
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9953905987420176e-05,
+      "loss": 0.8332,
+      "step": 8400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.995150525759831e-05,
+      "loss": 0.001,
+      "step": 8401
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9949104527776446e-05,
+      "loss": 0.001,
+      "step": 8402
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9946703797954578e-05,
+      "loss": 0.001,
+      "step": 8403
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9944303068132713e-05,
+      "loss": 0.0014,
+      "step": 8404
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9941902338310845e-05,
+      "loss": 0.0016,
+      "step": 8405
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9939501608488984e-05,
+      "loss": 0.001,
+      "step": 8406
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.993710087866712e-05,
+      "loss": 0.0024,
+      "step": 8407
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.993470014884525e-05,
+      "loss": 0.0011,
+      "step": 8408
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9932299419023386e-05,
+      "loss": 0.0014,
+      "step": 8409
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9929898689201518e-05,
+      "loss": 0.001,
+      "step": 8410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9927497959379653e-05,
+      "loss": 0.7826,
+      "step": 8411
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9925097229557785e-05,
+      "loss": 0.0011,
+      "step": 8412
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.992269649973592e-05,
+      "loss": 0.0014,
+      "step": 8413
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9920295769914052e-05,
+      "loss": 0.002,
+      "step": 8414
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.991789504009219e-05,
+      "loss": 0.0027,
+      "step": 8415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9915494310270326e-05,
+      "loss": 0.0023,
+      "step": 8416
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.991309358044846e-05,
+      "loss": 0.0043,
+      "step": 8417
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9910692850626593e-05,
+      "loss": 0.002,
+      "step": 8418
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9908292120804728e-05,
+      "loss": 0.0015,
+      "step": 8419
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.990589139098286e-05,
+      "loss": 0.0015,
+      "step": 8420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9903490661160995e-05,
+      "loss": 0.0029,
+      "step": 8421
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9901089931339126e-05,
+      "loss": 0.0014,
+      "step": 8422
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.989868920151726e-05,
+      "loss": 0.0016,
+      "step": 8423
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98962884716954e-05,
+      "loss": 0.0014,
+      "step": 8424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9893887741873532e-05,
+      "loss": 0.0012,
+      "step": 8425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9891487012051667e-05,
+      "loss": 0.001,
+      "step": 8426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98890862822298e-05,
+      "loss": 0.0018,
+      "step": 8427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9886685552407934e-05,
+      "loss": 0.0011,
+      "step": 8428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9884284822586066e-05,
+      "loss": 0.0015,
+      "step": 8429
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.98818840927642e-05,
+      "loss": 0.0028,
+      "step": 8430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9879483362942333e-05,
+      "loss": 0.0011,
+      "step": 8431
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9877082633120468e-05,
+      "loss": 0.0011,
+      "step": 8432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9874681903298607e-05,
+      "loss": 0.001,
+      "step": 8433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9872281173476742e-05,
+      "loss": 0.0184,
+      "step": 8434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9869880443654874e-05,
+      "loss": 0.0013,
+      "step": 8435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.986747971383301e-05,
+      "loss": 0.0014,
+      "step": 8436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.986507898401114e-05,
+      "loss": 0.001,
+      "step": 8437
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9862678254189276e-05,
+      "loss": 0.0013,
+      "step": 8438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9860277524367408e-05,
+      "loss": 0.0015,
+      "step": 8439
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9857876794545543e-05,
+      "loss": 0.0008,
+      "step": 8440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9855476064723675e-05,
+      "loss": 0.0005,
+      "step": 8441
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.985307533490181e-05,
+      "loss": 0.0008,
+      "step": 8442
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.985067460507995e-05,
+      "loss": 0.0016,
+      "step": 8443
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.984827387525808e-05,
+      "loss": 0.0009,
+      "step": 8444
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9845873145436215e-05,
+      "loss": 0.0009,
+      "step": 8445
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9843472415614347e-05,
+      "loss": 0.0011,
+      "step": 8446
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9841071685792482e-05,
+      "loss": 0.0025,
+      "step": 8447
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9838670955970614e-05,
+      "loss": 0.001,
+      "step": 8448
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.983627022614875e-05,
+      "loss": 0.0011,
+      "step": 8449
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.983386949632688e-05,
+      "loss": 0.001,
+      "step": 8450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9831468766505016e-05,
+      "loss": 0.0008,
+      "step": 8451
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9829068036683155e-05,
+      "loss": 0.0009,
+      "step": 8452
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.982666730686129e-05,
+      "loss": 0.0008,
+      "step": 8453
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9824266577039422e-05,
+      "loss": 0.0005,
+      "step": 8454
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9821865847217557e-05,
+      "loss": 0.0008,
+      "step": 8455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.981946511739569e-05,
+      "loss": 0.0007,
+      "step": 8456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9817064387573824e-05,
+      "loss": 0.0009,
+      "step": 8457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9814663657751956e-05,
+      "loss": 0.0011,
+      "step": 8458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.981226292793009e-05,
+      "loss": 0.0006,
+      "step": 8459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9809862198108223e-05,
+      "loss": 0.0011,
+      "step": 8460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.980746146828636e-05,
+      "loss": 0.9292,
+      "step": 8461
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9805060738464497e-05,
+      "loss": 0.001,
+      "step": 8462
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.980266000864263e-05,
+      "loss": 0.0009,
+      "step": 8463
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9800259278820764e-05,
+      "loss": 0.0012,
+      "step": 8464
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9797858548998896e-05,
+      "loss": 0.565,
+      "step": 8465
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.979545781917703e-05,
+      "loss": 0.0012,
+      "step": 8466
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9793057089355163e-05,
+      "loss": 0.0008,
+      "step": 8467
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9790656359533298e-05,
+      "loss": 0.0011,
+      "step": 8468
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9788255629711433e-05,
+      "loss": 0.0008,
+      "step": 8469
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.978585489988957e-05,
+      "loss": 0.0008,
+      "step": 8470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9783454170067703e-05,
+      "loss": 0.0007,
+      "step": 8471
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.978105344024584e-05,
+      "loss": 0.0012,
+      "step": 8472
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.977865271042397e-05,
+      "loss": 0.0008,
+      "step": 8473
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9776251980602105e-05,
+      "loss": 0.0009,
+      "step": 8474
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9773851250780237e-05,
+      "loss": 0.0011,
+      "step": 8475
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9771450520958372e-05,
+      "loss": 0.001,
+      "step": 8476
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9769049791136504e-05,
+      "loss": 0.0008,
+      "step": 8477
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.976664906131464e-05,
+      "loss": 0.0009,
+      "step": 8478
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9764248331492778e-05,
+      "loss": 0.001,
+      "step": 8479
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.976184760167091e-05,
+      "loss": 0.0008,
+      "step": 8480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9759446871849045e-05,
+      "loss": 0.4222,
+      "step": 8481
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9757046142027177e-05,
+      "loss": 0.0011,
+      "step": 8482
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9754645412205312e-05,
+      "loss": 0.0007,
+      "step": 8483
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9752244682383447e-05,
+      "loss": 0.0008,
+      "step": 8484
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.974984395256158e-05,
+      "loss": 0.0009,
+      "step": 8485
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9747443222739714e-05,
+      "loss": 0.0009,
+      "step": 8486
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9745042492917846e-05,
+      "loss": 0.0007,
+      "step": 8487
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9742641763095985e-05,
+      "loss": 0.001,
+      "step": 8488
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.974024103327412e-05,
+      "loss": 0.0012,
+      "step": 8489
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.973784030345225e-05,
+      "loss": 0.0006,
+      "step": 8490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9735439573630387e-05,
+      "loss": 0.0009,
+      "step": 8491
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.973303884380852e-05,
+      "loss": 0.0006,
+      "step": 8492
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9730638113986654e-05,
+      "loss": 0.0007,
+      "step": 8493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9728237384164786e-05,
+      "loss": 0.0007,
+      "step": 8494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.972583665434292e-05,
+      "loss": 0.0009,
+      "step": 8495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9723435924521052e-05,
+      "loss": 0.0009,
+      "step": 8496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.972103519469919e-05,
+      "loss": 0.0012,
+      "step": 8497
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9718634464877326e-05,
+      "loss": 1.25,
+      "step": 8498
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.971623373505546e-05,
+      "loss": 0.0005,
+      "step": 8499
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9713833005233593e-05,
+      "loss": 0.0007,
+      "step": 8500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.971143227541173e-05,
+      "loss": 0.0008,
+      "step": 8501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.970903154558986e-05,
+      "loss": 0.0008,
+      "step": 8502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9706630815767995e-05,
+      "loss": 0.0011,
+      "step": 8503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9704230085946127e-05,
+      "loss": 0.001,
+      "step": 8504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9701829356124262e-05,
+      "loss": 0.0007,
+      "step": 8505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9699428626302394e-05,
+      "loss": 0.8965,
+      "step": 8506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9697027896480533e-05,
+      "loss": 0.0004,
+      "step": 8507
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9694627166658668e-05,
+      "loss": 0.0009,
+      "step": 8508
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.96922264368368e-05,
+      "loss": 0.0077,
+      "step": 8509
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9689825707014935e-05,
+      "loss": 0.0011,
+      "step": 8510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9687424977193067e-05,
+      "loss": 0.0009,
+      "step": 8511
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9685024247371202e-05,
+      "loss": 0.0008,
+      "step": 8512
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9682623517549334e-05,
+      "loss": 0.0031,
+      "step": 8513
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.968022278772747e-05,
+      "loss": 0.9018,
+      "step": 8514
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.96778220579056e-05,
+      "loss": 0.0012,
+      "step": 8515
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9675421328083743e-05,
+      "loss": 0.0019,
+      "step": 8516
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9673020598261875e-05,
+      "loss": 0.0074,
+      "step": 8517
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.967061986844001e-05,
+      "loss": 0.0029,
+      "step": 8518
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.966821913861814e-05,
+      "loss": 0.003,
+      "step": 8519
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9665818408796277e-05,
+      "loss": 0.0019,
+      "step": 8520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.966341767897441e-05,
+      "loss": 0.0022,
+      "step": 8521
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9661016949152544e-05,
+      "loss": 0.0021,
+      "step": 8522
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9658616219330675e-05,
+      "loss": 0.0027,
+      "step": 8523
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.965621548950881e-05,
+      "loss": 0.0025,
+      "step": 8524
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.965381475968695e-05,
+      "loss": 0.0014,
+      "step": 8525
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.965141402986508e-05,
+      "loss": 0.0023,
+      "step": 8526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9649013300043216e-05,
+      "loss": 0.0019,
+      "step": 8527
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9646612570221348e-05,
+      "loss": 0.0028,
+      "step": 8528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9644211840399483e-05,
+      "loss": 0.0021,
+      "step": 8529
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9641811110577615e-05,
+      "loss": 0.0019,
+      "step": 8530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.963941038075575e-05,
+      "loss": 0.0023,
+      "step": 8531
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9637009650933882e-05,
+      "loss": 0.8588,
+      "step": 8532
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9634608921112017e-05,
+      "loss": 0.0016,
+      "step": 8533
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9632208191290156e-05,
+      "loss": 0.0019,
+      "step": 8534
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.962980746146829e-05,
+      "loss": 0.0015,
+      "step": 8535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9627406731646423e-05,
+      "loss": 0.0015,
+      "step": 8536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9625006001824558e-05,
+      "loss": 0.0021,
+      "step": 8537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.962260527200269e-05,
+      "loss": 0.0017,
+      "step": 8538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9620204542180825e-05,
+      "loss": 0.002,
+      "step": 8539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9617803812358957e-05,
+      "loss": 0.0018,
+      "step": 8540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9615403082537092e-05,
+      "loss": 0.0014,
+      "step": 8541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9613002352715224e-05,
+      "loss": 0.0018,
+      "step": 8542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9610601622893362e-05,
+      "loss": 0.0015,
+      "step": 8543
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9608200893071497e-05,
+      "loss": 0.0019,
+      "step": 8544
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.960580016324963e-05,
+      "loss": 0.0023,
+      "step": 8545
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9603399433427764e-05,
+      "loss": 0.0016,
+      "step": 8546
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9600998703605896e-05,
+      "loss": 0.0016,
+      "step": 8547
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.959859797378403e-05,
+      "loss": 0.0018,
+      "step": 8548
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9596197243962163e-05,
+      "loss": 0.0014,
+      "step": 8549
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.95937965141403e-05,
+      "loss": 0.0016,
+      "step": 8550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9591395784318434e-05,
+      "loss": 0.0015,
+      "step": 8551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9588995054496572e-05,
+      "loss": 0.8683,
+      "step": 8552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9586594324674704e-05,
+      "loss": 0.0016,
+      "step": 8553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.958419359485284e-05,
+      "loss": 0.0016,
+      "step": 8554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.958179286503097e-05,
+      "loss": 0.0015,
+      "step": 8555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9579392135209106e-05,
+      "loss": 0.0016,
+      "step": 8556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9576991405387238e-05,
+      "loss": 0.0016,
+      "step": 8557
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9574590675565373e-05,
+      "loss": 0.0014,
+      "step": 8558
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9572189945743505e-05,
+      "loss": 0.0015,
+      "step": 8559
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.956978921592164e-05,
+      "loss": 0.0017,
+      "step": 8560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.956738848609978e-05,
+      "loss": 0.0015,
+      "step": 8561
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.956498775627791e-05,
+      "loss": 0.0015,
+      "step": 8562
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9562587026456046e-05,
+      "loss": 0.0018,
+      "step": 8563
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9560186296634178e-05,
+      "loss": 0.0017,
+      "step": 8564
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9557785566812313e-05,
+      "loss": 0.0016,
+      "step": 8565
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9555384836990448e-05,
+      "loss": 0.0012,
+      "step": 8566
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.955298410716858e-05,
+      "loss": 0.0016,
+      "step": 8567
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9550583377346715e-05,
+      "loss": 0.0013,
+      "step": 8568
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9548182647524847e-05,
+      "loss": 0.0018,
+      "step": 8569
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9545781917702982e-05,
+      "loss": 0.0016,
+      "step": 8570
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.954338118788112e-05,
+      "loss": 0.0013,
+      "step": 8571
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9540980458059252e-05,
+      "loss": 0.0013,
+      "step": 8572
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9538579728237387e-05,
+      "loss": 0.0012,
+      "step": 8573
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.953617899841552e-05,
+      "loss": 0.0011,
+      "step": 8574
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9533778268593654e-05,
+      "loss": 0.0015,
+      "step": 8575
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9531377538771786e-05,
+      "loss": 0.0014,
+      "step": 8576
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.952897680894992e-05,
+      "loss": 0.8186,
+      "step": 8577
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9526576079128053e-05,
+      "loss": 0.0012,
+      "step": 8578
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.952417534930619e-05,
+      "loss": 0.0015,
+      "step": 8579
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9521774619484327e-05,
+      "loss": 0.0018,
+      "step": 8580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9519373889662462e-05,
+      "loss": 0.0013,
+      "step": 8581
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9516973159840594e-05,
+      "loss": 0.0014,
+      "step": 8582
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.951457243001873e-05,
+      "loss": 0.0012,
+      "step": 8583
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.951217170019686e-05,
+      "loss": 0.0013,
+      "step": 8584
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9509770970374996e-05,
+      "loss": 0.0012,
+      "step": 8585
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9507370240553128e-05,
+      "loss": 0.0015,
+      "step": 8586
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9504969510731263e-05,
+      "loss": 0.0013,
+      "step": 8587
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9502568780909395e-05,
+      "loss": 0.0012,
+      "step": 8588
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9500168051087534e-05,
+      "loss": 0.0012,
+      "step": 8589
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.949776732126567e-05,
+      "loss": 0.0014,
+      "step": 8590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.94953665914438e-05,
+      "loss": 0.0013,
+      "step": 8591
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9492965861621936e-05,
+      "loss": 0.0013,
+      "step": 8592
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9490565131800068e-05,
+      "loss": 0.0013,
+      "step": 8593
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9488164401978203e-05,
+      "loss": 0.0017,
+      "step": 8594
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9485763672156335e-05,
+      "loss": 0.0076,
+      "step": 8595
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.948336294233447e-05,
+      "loss": 0.8219,
+      "step": 8596
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.94809622125126e-05,
+      "loss": 0.0012,
+      "step": 8597
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9478561482690743e-05,
+      "loss": 0.1032,
+      "step": 8598
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9476160752868875e-05,
+      "loss": 0.0011,
+      "step": 8599
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.947376002304701e-05,
+      "loss": 0.0016,
+      "step": 8600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9471359293225142e-05,
+      "loss": 0.0012,
+      "step": 8601
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9468958563403277e-05,
+      "loss": 0.0013,
+      "step": 8602
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.946655783358141e-05,
+      "loss": 0.0012,
+      "step": 8603
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9464157103759544e-05,
+      "loss": 0.0012,
+      "step": 8604
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9461756373937676e-05,
+      "loss": 0.0014,
+      "step": 8605
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.945935564411581e-05,
+      "loss": 0.0013,
+      "step": 8606
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.945695491429395e-05,
+      "loss": 0.0012,
+      "step": 8607
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9454554184472082e-05,
+      "loss": 0.0012,
+      "step": 8608
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9452153454650217e-05,
+      "loss": 0.0013,
+      "step": 8609
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.944975272482835e-05,
+      "loss": 0.0011,
+      "step": 8610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9447351995006484e-05,
+      "loss": 0.0015,
+      "step": 8611
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9444951265184616e-05,
+      "loss": 0.0012,
+      "step": 8612
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.944255053536275e-05,
+      "loss": 0.0012,
+      "step": 8613
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9440149805540883e-05,
+      "loss": 0.0012,
+      "step": 8614
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9437749075719018e-05,
+      "loss": 0.0012,
+      "step": 8615
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9435348345897157e-05,
+      "loss": 0.0014,
+      "step": 8616
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9432947616075292e-05,
+      "loss": 0.0013,
+      "step": 8617
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9430546886253424e-05,
+      "loss": 0.0013,
+      "step": 8618
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.942814615643156e-05,
+      "loss": 0.0013,
+      "step": 8619
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.942574542660969e-05,
+      "loss": 0.0011,
+      "step": 8620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9423344696787826e-05,
+      "loss": 0.0011,
+      "step": 8621
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9420943966965957e-05,
+      "loss": 0.0012,
+      "step": 8622
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9418543237144093e-05,
+      "loss": 0.001,
+      "step": 8623
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9416142507322224e-05,
+      "loss": 0.0012,
+      "step": 8624
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.941374177750036e-05,
+      "loss": 0.0012,
+      "step": 8625
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9411341047678498e-05,
+      "loss": 0.0011,
+      "step": 8626
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.940894031785663e-05,
+      "loss": 0.0012,
+      "step": 8627
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9406539588034765e-05,
+      "loss": 0.0011,
+      "step": 8628
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9404138858212897e-05,
+      "loss": 0.001,
+      "step": 8629
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9401738128391032e-05,
+      "loss": 0.001,
+      "step": 8630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9399337398569164e-05,
+      "loss": 0.001,
+      "step": 8631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.93969366687473e-05,
+      "loss": 0.0008,
+      "step": 8632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.939453593892543e-05,
+      "loss": 0.0008,
+      "step": 8633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9392135209103566e-05,
+      "loss": 0.001,
+      "step": 8634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9389734479281705e-05,
+      "loss": 0.0009,
+      "step": 8635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.938733374945984e-05,
+      "loss": 0.0013,
+      "step": 8636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9384933019637972e-05,
+      "loss": 0.0009,
+      "step": 8637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9382532289816107e-05,
+      "loss": 0.7774,
+      "step": 8638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.938013155999424e-05,
+      "loss": 0.0009,
+      "step": 8639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9377730830172374e-05,
+      "loss": 0.0009,
+      "step": 8640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9375330100350506e-05,
+      "loss": 0.0009,
+      "step": 8641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.937292937052864e-05,
+      "loss": 0.0009,
+      "step": 8642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9370528640706773e-05,
+      "loss": 0.0008,
+      "step": 8643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.936812791088491e-05,
+      "loss": 0.0009,
+      "step": 8644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9365727181063046e-05,
+      "loss": 0.0007,
+      "step": 8645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9363326451241178e-05,
+      "loss": 0.0007,
+      "step": 8646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9360925721419313e-05,
+      "loss": 0.0013,
+      "step": 8647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9358524991597445e-05,
+      "loss": 0.0008,
+      "step": 8648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.935612426177558e-05,
+      "loss": 0.0008,
+      "step": 8649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9353723531953716e-05,
+      "loss": 0.0008,
+      "step": 8650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9351322802131847e-05,
+      "loss": 0.001,
+      "step": 8651
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9348922072309983e-05,
+      "loss": 0.0008,
+      "step": 8652
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.934652134248812e-05,
+      "loss": 0.0011,
+      "step": 8653
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9344120612666253e-05,
+      "loss": 0.0009,
+      "step": 8654
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9341719882844388e-05,
+      "loss": 0.0009,
+      "step": 8655
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.933931915302252e-05,
+      "loss": 0.0009,
+      "step": 8656
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9336918423200655e-05,
+      "loss": 0.0008,
+      "step": 8657
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9334517693378787e-05,
+      "loss": 0.0007,
+      "step": 8658
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9332116963556922e-05,
+      "loss": 0.0007,
+      "step": 8659
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9329716233735054e-05,
+      "loss": 0.0008,
+      "step": 8660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.932731550391319e-05,
+      "loss": 0.0006,
+      "step": 8661
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9324914774091328e-05,
+      "loss": 0.0007,
+      "step": 8662
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.932251404426946e-05,
+      "loss": 0.4153,
+      "step": 8663
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9320113314447595e-05,
+      "loss": 0.0007,
+      "step": 8664
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.931771258462573e-05,
+      "loss": 0.001,
+      "step": 8665
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9315311854803862e-05,
+      "loss": 0.0006,
+      "step": 8666
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9312911124981997e-05,
+      "loss": 0.0006,
+      "step": 8667
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.931051039516013e-05,
+      "loss": 0.0007,
+      "step": 8668
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9308109665338264e-05,
+      "loss": 0.0005,
+      "step": 8669
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9305708935516396e-05,
+      "loss": 0.0007,
+      "step": 8670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9303308205694534e-05,
+      "loss": 0.0007,
+      "step": 8671
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.930090747587267e-05,
+      "loss": 0.0008,
+      "step": 8672
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.92985067460508e-05,
+      "loss": 0.0005,
+      "step": 8673
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9296106016228936e-05,
+      "loss": 0.0007,
+      "step": 8674
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9293705286407068e-05,
+      "loss": 0.0007,
+      "step": 8675
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9291304556585203e-05,
+      "loss": 0.0005,
+      "step": 8676
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9288903826763335e-05,
+      "loss": 0.0007,
+      "step": 8677
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.928650309694147e-05,
+      "loss": 0.0007,
+      "step": 8678
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9284102367119602e-05,
+      "loss": 0.0006,
+      "step": 8679
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9281701637297744e-05,
+      "loss": 0.8724,
+      "step": 8680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9279300907475876e-05,
+      "loss": 0.0007,
+      "step": 8681
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.927690017765401e-05,
+      "loss": 0.0008,
+      "step": 8682
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9274499447832143e-05,
+      "loss": 0.0006,
+      "step": 8683
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9272098718010278e-05,
+      "loss": 0.0008,
+      "step": 8684
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.926969798818841e-05,
+      "loss": 0.0006,
+      "step": 8685
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9267297258366545e-05,
+      "loss": 0.0005,
+      "step": 8686
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9264896528544677e-05,
+      "loss": 0.0006,
+      "step": 8687
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9262495798722812e-05,
+      "loss": 0.0009,
+      "step": 8688
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9260095068900944e-05,
+      "loss": 0.0006,
+      "step": 8689
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9257694339079083e-05,
+      "loss": 0.0005,
+      "step": 8690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9255293609257218e-05,
+      "loss": 0.0006,
+      "step": 8691
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.925289287943535e-05,
+      "loss": 0.0006,
+      "step": 8692
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9250492149613485e-05,
+      "loss": 0.0009,
+      "step": 8693
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9248091419791617e-05,
+      "loss": 0.0007,
+      "step": 8694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9245690689969752e-05,
+      "loss": 0.0005,
+      "step": 8695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9243289960147884e-05,
+      "loss": 0.0007,
+      "step": 8696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.924088923032602e-05,
+      "loss": 0.0008,
+      "step": 8697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.923848850050415e-05,
+      "loss": 0.0008,
+      "step": 8698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9236087770682292e-05,
+      "loss": 0.0007,
+      "step": 8699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9233687040860424e-05,
+      "loss": 0.0005,
+      "step": 8700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.923128631103856e-05,
+      "loss": 0.0014,
+      "step": 8701
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.922888558121669e-05,
+      "loss": 0.0007,
+      "step": 8702
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9226484851394826e-05,
+      "loss": 0.0011,
+      "step": 8703
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9224084121572958e-05,
+      "loss": 0.0007,
+      "step": 8704
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9221683391751093e-05,
+      "loss": 0.0006,
+      "step": 8705
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9219282661929225e-05,
+      "loss": 0.0006,
+      "step": 8706
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.921688193210736e-05,
+      "loss": 0.0006,
+      "step": 8707
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.92144812022855e-05,
+      "loss": 0.0006,
+      "step": 8708
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.921208047246363e-05,
+      "loss": 0.0005,
+      "step": 8709
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9209679742641766e-05,
+      "loss": 0.0006,
+      "step": 8710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9207279012819898e-05,
+      "loss": 0.0005,
+      "step": 8711
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9204878282998033e-05,
+      "loss": 0.0005,
+      "step": 8712
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9202477553176165e-05,
+      "loss": 0.0005,
+      "step": 8713
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.92000768233543e-05,
+      "loss": 0.0004,
+      "step": 8714
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9197676093532432e-05,
+      "loss": 0.0005,
+      "step": 8715
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9195275363710567e-05,
+      "loss": 0.0005,
+      "step": 8716
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9192874633888706e-05,
+      "loss": 0.0006,
+      "step": 8717
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.919047390406684e-05,
+      "loss": 0.0004,
+      "step": 8718
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9188073174244973e-05,
+      "loss": 0.0004,
+      "step": 8719
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9185672444423108e-05,
+      "loss": 0.0006,
+      "step": 8720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.918327171460124e-05,
+      "loss": 0.0004,
+      "step": 8721
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9180870984779375e-05,
+      "loss": 0.0005,
+      "step": 8722
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9178470254957506e-05,
+      "loss": 0.0006,
+      "step": 8723
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.917606952513564e-05,
+      "loss": 0.0006,
+      "step": 8724
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9173668795313773e-05,
+      "loss": 0.0005,
+      "step": 8725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9171268065491912e-05,
+      "loss": 0.0004,
+      "step": 8726
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9168867335670047e-05,
+      "loss": 0.0005,
+      "step": 8727
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.916646660584818e-05,
+      "loss": 0.0003,
+      "step": 8728
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9164065876026314e-05,
+      "loss": 0.0002,
+      "step": 8729
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9161665146204446e-05,
+      "loss": 0.0004,
+      "step": 8730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.915926441638258e-05,
+      "loss": 0.0006,
+      "step": 8731
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9156863686560716e-05,
+      "loss": 1.0028,
+      "step": 8732
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9154462956738848e-05,
+      "loss": 0.0004,
+      "step": 8733
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9152062226916983e-05,
+      "loss": 0.0006,
+      "step": 8734
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9149661497095122e-05,
+      "loss": 0.0006,
+      "step": 8735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9147260767273254e-05,
+      "loss": 0.0004,
+      "step": 8736
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.914486003745139e-05,
+      "loss": 0.0005,
+      "step": 8737
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.914245930762952e-05,
+      "loss": 0.0005,
+      "step": 8738
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9140058577807656e-05,
+      "loss": 0.0007,
+      "step": 8739
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9137657847985788e-05,
+      "loss": 0.0004,
+      "step": 8740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9135257118163923e-05,
+      "loss": 0.0004,
+      "step": 8741
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9132856388342055e-05,
+      "loss": 0.0005,
+      "step": 8742
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.913045565852019e-05,
+      "loss": 0.0005,
+      "step": 8743
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.912805492869833e-05,
+      "loss": 0.0005,
+      "step": 8744
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.912565419887646e-05,
+      "loss": 0.3739,
+      "step": 8745
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9123253469054595e-05,
+      "loss": 0.0006,
+      "step": 8746
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.912085273923273e-05,
+      "loss": 0.0006,
+      "step": 8747
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9118452009410862e-05,
+      "loss": 0.0006,
+      "step": 8748
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9116051279588998e-05,
+      "loss": 0.0004,
+      "step": 8749
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.911365054976713e-05,
+      "loss": 0.0004,
+      "step": 8750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9111249819945265e-05,
+      "loss": 0.0003,
+      "step": 8751
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9108849090123396e-05,
+      "loss": 0.0005,
+      "step": 8752
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.910644836030153e-05,
+      "loss": 0.0005,
+      "step": 8753
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.910404763047967e-05,
+      "loss": 0.0004,
+      "step": 8754
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9101646900657802e-05,
+      "loss": 0.0005,
+      "step": 8755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9099246170835937e-05,
+      "loss": 0.0006,
+      "step": 8756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.909684544101407e-05,
+      "loss": 0.0006,
+      "step": 8757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9094444711192204e-05,
+      "loss": 0.0006,
+      "step": 8758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9092043981370336e-05,
+      "loss": 0.0004,
+      "step": 8759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.908964325154847e-05,
+      "loss": 0.0004,
+      "step": 8760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9087242521726603e-05,
+      "loss": 0.0007,
+      "step": 8761
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9084841791904738e-05,
+      "loss": 0.0004,
+      "step": 8762
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9082441062082877e-05,
+      "loss": 0.0006,
+      "step": 8763
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9080040332261012e-05,
+      "loss": 0.0004,
+      "step": 8764
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9077639602439144e-05,
+      "loss": 0.0004,
+      "step": 8765
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.907523887261728e-05,
+      "loss": 0.0005,
+      "step": 8766
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.907283814279541e-05,
+      "loss": 0.0006,
+      "step": 8767
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9070437412973546e-05,
+      "loss": 0.0006,
+      "step": 8768
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9068036683151678e-05,
+      "loss": 0.0008,
+      "step": 8769
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9065635953329813e-05,
+      "loss": 0.0005,
+      "step": 8770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9063235223507945e-05,
+      "loss": 0.0005,
+      "step": 8771
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9060834493686083e-05,
+      "loss": 0.0005,
+      "step": 8772
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.905843376386422e-05,
+      "loss": 0.0007,
+      "step": 8773
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.905603303404235e-05,
+      "loss": 0.0007,
+      "step": 8774
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9053632304220485e-05,
+      "loss": 0.0006,
+      "step": 8775
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9051231574398617e-05,
+      "loss": 0.0004,
+      "step": 8776
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9048830844576752e-05,
+      "loss": 0.0005,
+      "step": 8777
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9046430114754884e-05,
+      "loss": 0.0009,
+      "step": 8778
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.904402938493302e-05,
+      "loss": 0.0005,
+      "step": 8779
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.904162865511115e-05,
+      "loss": 0.0005,
+      "step": 8780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9039227925289293e-05,
+      "loss": 0.0005,
+      "step": 8781
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9036827195467425e-05,
+      "loss": 0.0004,
+      "step": 8782
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.903442646564556e-05,
+      "loss": 0.0005,
+      "step": 8783
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9032025735823692e-05,
+      "loss": 0.0006,
+      "step": 8784
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9029625006001827e-05,
+      "loss": 0.0004,
+      "step": 8785
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.902722427617996e-05,
+      "loss": 0.0004,
+      "step": 8786
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9024823546358094e-05,
+      "loss": 0.9484,
+      "step": 8787
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9022422816536226e-05,
+      "loss": 0.0009,
+      "step": 8788
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.902002208671436e-05,
+      "loss": 0.0005,
+      "step": 8789
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.90176213568925e-05,
+      "loss": 0.0004,
+      "step": 8790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.901522062707063e-05,
+      "loss": 0.0005,
+      "step": 8791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9012819897248767e-05,
+      "loss": 0.0004,
+      "step": 8792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.90104191674269e-05,
+      "loss": 0.001,
+      "step": 8793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9008018437605034e-05,
+      "loss": 0.0005,
+      "step": 8794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9005617707783166e-05,
+      "loss": 0.0072,
+      "step": 8795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.90032169779613e-05,
+      "loss": 0.0037,
+      "step": 8796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9000816248139433e-05,
+      "loss": 0.0006,
+      "step": 8797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8998415518317568e-05,
+      "loss": 0.8832,
+      "step": 8798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8996014788495706e-05,
+      "loss": 0.0006,
+      "step": 8799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.899361405867384e-05,
+      "loss": 0.0005,
+      "step": 8800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8991213328851973e-05,
+      "loss": 0.0005,
+      "step": 8801
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.898881259903011e-05,
+      "loss": 0.0004,
+      "step": 8802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.898641186920824e-05,
+      "loss": 0.0006,
+      "step": 8803
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8984011139386375e-05,
+      "loss": 0.9218,
+      "step": 8804
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8981610409564507e-05,
+      "loss": 0.0004,
+      "step": 8805
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8979209679742642e-05,
+      "loss": 0.0007,
+      "step": 8806
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8976808949920774e-05,
+      "loss": 0.0007,
+      "step": 8807
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8974408220098913e-05,
+      "loss": 0.0009,
+      "step": 8808
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8972007490277048e-05,
+      "loss": 0.001,
+      "step": 8809
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.896960676045518e-05,
+      "loss": 0.0006,
+      "step": 8810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8967206030633315e-05,
+      "loss": 0.0012,
+      "step": 8811
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8964805300811447e-05,
+      "loss": 0.0009,
+      "step": 8812
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8962404570989582e-05,
+      "loss": 0.001,
+      "step": 8813
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8960003841167717e-05,
+      "loss": 0.0005,
+      "step": 8814
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.895760311134585e-05,
+      "loss": 0.0005,
+      "step": 8815
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8955202381523984e-05,
+      "loss": 0.0009,
+      "step": 8816
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8952801651702116e-05,
+      "loss": 0.0009,
+      "step": 8817
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8950400921880255e-05,
+      "loss": 0.0009,
+      "step": 8818
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.894800019205839e-05,
+      "loss": 0.6984,
+      "step": 8819
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.894559946223652e-05,
+      "loss": 0.0011,
+      "step": 8820
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8943198732414657e-05,
+      "loss": 0.0006,
+      "step": 8821
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.894079800259279e-05,
+      "loss": 0.0008,
+      "step": 8822
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8938397272770924e-05,
+      "loss": 1.2185,
+      "step": 8823
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8935996542949055e-05,
+      "loss": 0.7967,
+      "step": 8824
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.893359581312719e-05,
+      "loss": 0.0011,
+      "step": 8825
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8931195083305322e-05,
+      "loss": 0.0007,
+      "step": 8826
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.892879435348346e-05,
+      "loss": 0.0008,
+      "step": 8827
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8926393623661596e-05,
+      "loss": 0.001,
+      "step": 8828
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.892399289383973e-05,
+      "loss": 0.0006,
+      "step": 8829
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8921592164017863e-05,
+      "loss": 0.0004,
+      "step": 8830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8919191434196e-05,
+      "loss": 0.0008,
+      "step": 8831
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.891679070437413e-05,
+      "loss": 0.0007,
+      "step": 8832
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8914389974552265e-05,
+      "loss": 0.0004,
+      "step": 8833
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8911989244730397e-05,
+      "loss": 0.0008,
+      "step": 8834
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8909588514908532e-05,
+      "loss": 0.8287,
+      "step": 8835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.890718778508667e-05,
+      "loss": 0.0006,
+      "step": 8836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8904787055264803e-05,
+      "loss": 0.0007,
+      "step": 8837
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8902386325442938e-05,
+      "loss": 0.0011,
+      "step": 8838
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.889998559562107e-05,
+      "loss": 0.0009,
+      "step": 8839
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8897584865799205e-05,
+      "loss": 0.0011,
+      "step": 8840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8895184135977337e-05,
+      "loss": 0.0007,
+      "step": 8841
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8892783406155472e-05,
+      "loss": 0.0009,
+      "step": 8842
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8890382676333604e-05,
+      "loss": 0.001,
+      "step": 8843
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.888798194651174e-05,
+      "loss": 0.0008,
+      "step": 8844
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8885581216689878e-05,
+      "loss": 0.0006,
+      "step": 8845
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8883180486868013e-05,
+      "loss": 0.0007,
+      "step": 8846
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8880779757046145e-05,
+      "loss": 0.001,
+      "step": 8847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.887837902722428e-05,
+      "loss": 0.0011,
+      "step": 8848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.887597829740241e-05,
+      "loss": 0.0005,
+      "step": 8849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8873577567580547e-05,
+      "loss": 0.0005,
+      "step": 8850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.887117683775868e-05,
+      "loss": 0.0007,
+      "step": 8851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8868776107936814e-05,
+      "loss": 0.7715,
+      "step": 8852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8866375378114945e-05,
+      "loss": 0.0009,
+      "step": 8853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8863974648293084e-05,
+      "loss": 0.0004,
+      "step": 8854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.886157391847122e-05,
+      "loss": 0.0015,
+      "step": 8855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.885917318864935e-05,
+      "loss": 0.0017,
+      "step": 8856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8856772458827486e-05,
+      "loss": 0.0022,
+      "step": 8857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8854371729005618e-05,
+      "loss": 0.0005,
+      "step": 8858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8851970999183753e-05,
+      "loss": 0.0009,
+      "step": 8859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8849570269361885e-05,
+      "loss": 0.0018,
+      "step": 8860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.884716953954002e-05,
+      "loss": 0.0015,
+      "step": 8861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8844768809718152e-05,
+      "loss": 0.0006,
+      "step": 8862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8842368079896294e-05,
+      "loss": 0.0018,
+      "step": 8863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8839967350074426e-05,
+      "loss": 0.0008,
+      "step": 8864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.883756662025256e-05,
+      "loss": 0.0009,
+      "step": 8865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8835165890430693e-05,
+      "loss": 0.0005,
+      "step": 8866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8832765160608828e-05,
+      "loss": 0.0008,
+      "step": 8867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.883036443078696e-05,
+      "loss": 0.0006,
+      "step": 8868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8827963700965095e-05,
+      "loss": 0.0006,
+      "step": 8869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8825562971143227e-05,
+      "loss": 0.0008,
+      "step": 8870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8823162241321362e-05,
+      "loss": 0.0008,
+      "step": 8871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.88207615114995e-05,
+      "loss": 0.0005,
+      "step": 8872
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8818360781677632e-05,
+      "loss": 0.0017,
+      "step": 8873
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8815960051855767e-05,
+      "loss": 0.0007,
+      "step": 8874
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.88135593220339e-05,
+      "loss": 0.0127,
+      "step": 8875
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8811158592212034e-05,
+      "loss": 0.0039,
+      "step": 8876
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8808757862390166e-05,
+      "loss": 0.0008,
+      "step": 8877
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.88063571325683e-05,
+      "loss": 0.0004,
+      "step": 8878
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8803956402746433e-05,
+      "loss": 0.0005,
+      "step": 8879
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.880155567292457e-05,
+      "loss": 0.0009,
+      "step": 8880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.87991549431027e-05,
+      "loss": 0.0006,
+      "step": 8881
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8796754213280842e-05,
+      "loss": 0.0006,
+      "step": 8882
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8794353483458974e-05,
+      "loss": 0.6572,
+      "step": 8883
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.879195275363711e-05,
+      "loss": 0.0006,
+      "step": 8884
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.878955202381524e-05,
+      "loss": 0.0008,
+      "step": 8885
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8787151293993376e-05,
+      "loss": 0.0006,
+      "step": 8886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8784750564171508e-05,
+      "loss": 0.0006,
+      "step": 8887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8782349834349643e-05,
+      "loss": 0.0003,
+      "step": 8888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8779949104527775e-05,
+      "loss": 0.0006,
+      "step": 8889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.877754837470591e-05,
+      "loss": 0.0037,
+      "step": 8890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.877514764488405e-05,
+      "loss": 0.0013,
+      "step": 8891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.877274691506218e-05,
+      "loss": 0.001,
+      "step": 8892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8770346185240316e-05,
+      "loss": 0.0011,
+      "step": 8893
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8767945455418448e-05,
+      "loss": 0.0007,
+      "step": 8894
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8765544725596583e-05,
+      "loss": 0.0006,
+      "step": 8895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8763143995774715e-05,
+      "loss": 0.0025,
+      "step": 8896
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.876074326595285e-05,
+      "loss": 0.0006,
+      "step": 8897
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8758342536130985e-05,
+      "loss": 0.0004,
+      "step": 8898
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8755941806309117e-05,
+      "loss": 0.9205,
+      "step": 8899
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8753541076487255e-05,
+      "loss": 0.0003,
+      "step": 8900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.875114034666539e-05,
+      "loss": 0.0006,
+      "step": 8901
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8748739616843522e-05,
+      "loss": 0.0006,
+      "step": 8902
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8746338887021657e-05,
+      "loss": 0.0005,
+      "step": 8903
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.874393815719979e-05,
+      "loss": 0.0006,
+      "step": 8904
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8741537427377924e-05,
+      "loss": 0.0004,
+      "step": 8905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8739136697556056e-05,
+      "loss": 0.0006,
+      "step": 8906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.873673596773419e-05,
+      "loss": 0.0006,
+      "step": 8907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8734335237912323e-05,
+      "loss": 0.0005,
+      "step": 8908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8731934508090462e-05,
+      "loss": 0.6004,
+      "step": 8909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8729533778268597e-05,
+      "loss": 0.0013,
+      "step": 8910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.872713304844673e-05,
+      "loss": 0.0006,
+      "step": 8911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8724732318624864e-05,
+      "loss": 0.0006,
+      "step": 8912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8722331588803e-05,
+      "loss": 0.587,
+      "step": 8913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.871993085898113e-05,
+      "loss": 0.0008,
+      "step": 8914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8717530129159266e-05,
+      "loss": 0.0007,
+      "step": 8915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8715129399337398e-05,
+      "loss": 0.0006,
+      "step": 8916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8712728669515533e-05,
+      "loss": 0.0005,
+      "step": 8917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8710327939693672e-05,
+      "loss": 0.8617,
+      "step": 8918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8707927209871804e-05,
+      "loss": 0.0035,
+      "step": 8919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.870552648004994e-05,
+      "loss": 0.0093,
+      "step": 8920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.870312575022807e-05,
+      "loss": 0.0004,
+      "step": 8921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8700725020406206e-05,
+      "loss": 0.0008,
+      "step": 8922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8698324290584338e-05,
+      "loss": 0.0006,
+      "step": 8923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8695923560762473e-05,
+      "loss": 0.0007,
+      "step": 8924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8693522830940604e-05,
+      "loss": 0.0005,
+      "step": 8925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.869112210111874e-05,
+      "loss": 0.0004,
+      "step": 8926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8688721371296878e-05,
+      "loss": 0.0006,
+      "step": 8927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8686320641475013e-05,
+      "loss": 0.0007,
+      "step": 8928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8683919911653145e-05,
+      "loss": 0.0005,
+      "step": 8929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.868151918183128e-05,
+      "loss": 0.0006,
+      "step": 8930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8679118452009412e-05,
+      "loss": 0.0006,
+      "step": 8931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8676717722187547e-05,
+      "loss": 0.0004,
+      "step": 8932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.867431699236568e-05,
+      "loss": 0.0009,
+      "step": 8933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8671916262543814e-05,
+      "loss": 0.6412,
+      "step": 8934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8669515532721946e-05,
+      "loss": 0.0008,
+      "step": 8935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.866711480290008e-05,
+      "loss": 0.0018,
+      "step": 8936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.866471407307822e-05,
+      "loss": 0.0008,
+      "step": 8937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8662313343256352e-05,
+      "loss": 0.0006,
+      "step": 8938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8659912613434487e-05,
+      "loss": 0.1151,
+      "step": 8939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.865751188361262e-05,
+      "loss": 0.0006,
+      "step": 8940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8655111153790754e-05,
+      "loss": 0.5511,
+      "step": 8941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8652710423968886e-05,
+      "loss": 0.0005,
+      "step": 8942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.865030969414702e-05,
+      "loss": 0.0005,
+      "step": 8943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8647908964325153e-05,
+      "loss": 0.0071,
+      "step": 8944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8645508234503288e-05,
+      "loss": 0.0039,
+      "step": 8945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8643107504681427e-05,
+      "loss": 0.0005,
+      "step": 8946
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8640706774859562e-05,
+      "loss": 0.0026,
+      "step": 8947
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8638306045037694e-05,
+      "loss": 0.0006,
+      "step": 8948
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.863590531521583e-05,
+      "loss": 0.0006,
+      "step": 8949
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.863350458539396e-05,
+      "loss": 0.0006,
+      "step": 8950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8631103855572096e-05,
+      "loss": 0.0008,
+      "step": 8951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8628703125750227e-05,
+      "loss": 0.0026,
+      "step": 8952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8626302395928363e-05,
+      "loss": 0.0007,
+      "step": 8953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8623901666106494e-05,
+      "loss": 0.8106,
+      "step": 8954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8621500936284633e-05,
+      "loss": 0.0007,
+      "step": 8955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8619100206462768e-05,
+      "loss": 0.7797,
+      "step": 8956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.86166994766409e-05,
+      "loss": 0.001,
+      "step": 8957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8614298746819035e-05,
+      "loss": 0.0005,
+      "step": 8958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8611898016997167e-05,
+      "loss": 0.9128,
+      "step": 8959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8609497287175302e-05,
+      "loss": 0.8201,
+      "step": 8960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8607096557353434e-05,
+      "loss": 0.0034,
+      "step": 8961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.860469582753157e-05,
+      "loss": 0.0017,
+      "step": 8962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.86022950977097e-05,
+      "loss": 0.0034,
+      "step": 8963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8599894367887843e-05,
+      "loss": 0.0043,
+      "step": 8964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8597493638065975e-05,
+      "loss": 0.0023,
+      "step": 8965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.859509290824411e-05,
+      "loss": 0.0024,
+      "step": 8966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8592692178422242e-05,
+      "loss": 0.0017,
+      "step": 8967
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8590291448600377e-05,
+      "loss": 0.0017,
+      "step": 8968
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.858789071877851e-05,
+      "loss": 0.1543,
+      "step": 8969
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8585489988956644e-05,
+      "loss": 0.0014,
+      "step": 8970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8583089259134776e-05,
+      "loss": 0.002,
+      "step": 8971
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.858068852931291e-05,
+      "loss": 0.0017,
+      "step": 8972
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.857828779949105e-05,
+      "loss": 0.0016,
+      "step": 8973
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.857588706966918e-05,
+      "loss": 0.0019,
+      "step": 8974
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8573486339847316e-05,
+      "loss": 0.0014,
+      "step": 8975
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8571085610025448e-05,
+      "loss": 0.0013,
+      "step": 8976
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8568684880203583e-05,
+      "loss": 0.0017,
+      "step": 8977
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8566284150381715e-05,
+      "loss": 0.0005,
+      "step": 8978
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.856388342055985e-05,
+      "loss": 0.0012,
+      "step": 8979
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8561482690737986e-05,
+      "loss": 0.0011,
+      "step": 8980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8559081960916117e-05,
+      "loss": 0.001,
+      "step": 8981
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8556681231094256e-05,
+      "loss": 0.0007,
+      "step": 8982
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.855428050127239e-05,
+      "loss": 0.0031,
+      "step": 8983
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8551879771450523e-05,
+      "loss": 0.0014,
+      "step": 8984
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8549479041628658e-05,
+      "loss": 0.0015,
+      "step": 8985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.854707831180679e-05,
+      "loss": 0.0015,
+      "step": 8986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8544677581984925e-05,
+      "loss": 0.0017,
+      "step": 8987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8542276852163057e-05,
+      "loss": 0.0013,
+      "step": 8988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8539876122341192e-05,
+      "loss": 0.0004,
+      "step": 8989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8537475392519324e-05,
+      "loss": 0.0011,
+      "step": 8990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8535074662697463e-05,
+      "loss": 0.001,
+      "step": 8991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8532673932875598e-05,
+      "loss": 0.0009,
+      "step": 8992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.853027320305373e-05,
+      "loss": 0.001,
+      "step": 8993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8527872473231865e-05,
+      "loss": 0.0007,
+      "step": 8994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.852547174341e-05,
+      "loss": 0.7399,
+      "step": 8995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8523071013588132e-05,
+      "loss": 0.0008,
+      "step": 8996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8520670283766267e-05,
+      "loss": 0.0015,
+      "step": 8997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.85182695539444e-05,
+      "loss": 0.0007,
+      "step": 8998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8515868824122534e-05,
+      "loss": 0.0016,
+      "step": 8999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8513468094300666e-05,
+      "loss": 0.0013,
+      "step": 9000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8511067364478804e-05,
+      "loss": 0.0012,
+      "step": 9001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.850866663465694e-05,
+      "loss": 0.7076,
+      "step": 9002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.850626590483507e-05,
+      "loss": 0.5698,
+      "step": 9003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8503865175013206e-05,
+      "loss": 0.0008,
+      "step": 9004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8501464445191338e-05,
+      "loss": 0.0014,
+      "step": 9005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8499063715369473e-05,
+      "loss": 0.0018,
+      "step": 9006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8496662985547605e-05,
+      "loss": 0.0013,
+      "step": 9007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.849426225572574e-05,
+      "loss": 0.0017,
+      "step": 9008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8491861525903872e-05,
+      "loss": 0.0012,
+      "step": 9009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8489460796082014e-05,
+      "loss": 0.0021,
+      "step": 9010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8487060066260146e-05,
+      "loss": 0.002,
+      "step": 9011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.848465933643828e-05,
+      "loss": 0.481,
+      "step": 9012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8482258606616413e-05,
+      "loss": 0.0017,
+      "step": 9013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8479857876794548e-05,
+      "loss": 0.0017,
+      "step": 9014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.847745714697268e-05,
+      "loss": 0.001,
+      "step": 9015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8475056417150815e-05,
+      "loss": 0.0459,
+      "step": 9016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8472655687328947e-05,
+      "loss": 0.0015,
+      "step": 9017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8470254957507082e-05,
+      "loss": 0.0019,
+      "step": 9018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.846785422768522e-05,
+      "loss": 0.0019,
+      "step": 9019
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8465453497863353e-05,
+      "loss": 0.0011,
+      "step": 9020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8463052768041488e-05,
+      "loss": 0.0011,
+      "step": 9021
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.846065203821962e-05,
+      "loss": 0.0023,
+      "step": 9022
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8458251308397755e-05,
+      "loss": 0.0003,
+      "step": 9023
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8455850578575887e-05,
+      "loss": 0.001,
+      "step": 9024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8453449848754022e-05,
+      "loss": 0.0014,
+      "step": 9025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8451049118932153e-05,
+      "loss": 0.0013,
+      "step": 9026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.844864838911029e-05,
+      "loss": 0.0006,
+      "step": 9027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8446247659288427e-05,
+      "loss": 0.0011,
+      "step": 9028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8443846929466562e-05,
+      "loss": 0.0009,
+      "step": 9029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8441446199644694e-05,
+      "loss": 0.0011,
+      "step": 9030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.843904546982283e-05,
+      "loss": 0.001,
+      "step": 9031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.843664474000096e-05,
+      "loss": 0.0008,
+      "step": 9032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8434244010179096e-05,
+      "loss": 0.0007,
+      "step": 9033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8431843280357228e-05,
+      "loss": 0.001,
+      "step": 9034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8429442550535363e-05,
+      "loss": 0.001,
+      "step": 9035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8427041820713495e-05,
+      "loss": 0.0011,
+      "step": 9036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8424641090891634e-05,
+      "loss": 0.0008,
+      "step": 9037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.842224036106977e-05,
+      "loss": 0.3972,
+      "step": 9038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.84198396312479e-05,
+      "loss": 0.0007,
+      "step": 9039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8417438901426036e-05,
+      "loss": 0.0011,
+      "step": 9040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8415038171604168e-05,
+      "loss": 0.0011,
+      "step": 9041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8412637441782303e-05,
+      "loss": 0.0011,
+      "step": 9042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8410236711960435e-05,
+      "loss": 0.0015,
+      "step": 9043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.840783598213857e-05,
+      "loss": 0.0008,
+      "step": 9044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8405435252316702e-05,
+      "loss": 0.0008,
+      "step": 9045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8403034522494844e-05,
+      "loss": 0.0011,
+      "step": 9046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8400633792672976e-05,
+      "loss": 0.0011,
+      "step": 9047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.839823306285111e-05,
+      "loss": 0.001,
+      "step": 9048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8395832333029243e-05,
+      "loss": 0.0008,
+      "step": 9049
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8393431603207378e-05,
+      "loss": 0.0008,
+      "step": 9050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.839103087338551e-05,
+      "loss": 0.0009,
+      "step": 9051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8388630143563645e-05,
+      "loss": 0.0011,
+      "step": 9052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8386229413741776e-05,
+      "loss": 0.0009,
+      "step": 9053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.838382868391991e-05,
+      "loss": 0.0009,
+      "step": 9054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.838142795409805e-05,
+      "loss": 0.0008,
+      "step": 9055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8379027224276182e-05,
+      "loss": 0.0011,
+      "step": 9056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8376626494454317e-05,
+      "loss": 0.0008,
+      "step": 9057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.837422576463245e-05,
+      "loss": 0.0007,
+      "step": 9058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8371825034810584e-05,
+      "loss": 0.0014,
+      "step": 9059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8369424304988716e-05,
+      "loss": 0.0004,
+      "step": 9060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.836702357516685e-05,
+      "loss": 0.001,
+      "step": 9061
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8364622845344986e-05,
+      "loss": 0.0011,
+      "step": 9062
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8362222115523118e-05,
+      "loss": 0.0007,
+      "step": 9063
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8359821385701253e-05,
+      "loss": 0.001,
+      "step": 9064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8357420655879392e-05,
+      "loss": 0.0007,
+      "step": 9065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8355019926057524e-05,
+      "loss": 0.0008,
+      "step": 9066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.835261919623566e-05,
+      "loss": 0.0009,
+      "step": 9067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.835021846641379e-05,
+      "loss": 0.0012,
+      "step": 9068
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8347817736591926e-05,
+      "loss": 0.001,
+      "step": 9069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8345417006770058e-05,
+      "loss": 0.0008,
+      "step": 9070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8343016276948193e-05,
+      "loss": 0.0007,
+      "step": 9071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8340615547126325e-05,
+      "loss": 0.0005,
+      "step": 9072
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.833821481730446e-05,
+      "loss": 0.0008,
+      "step": 9073
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.83358140874826e-05,
+      "loss": 0.0007,
+      "step": 9074
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.833341335766073e-05,
+      "loss": 0.0011,
+      "step": 9075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8331012627838865e-05,
+      "loss": 0.0006,
+      "step": 9076
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8328611898017e-05,
+      "loss": 0.001,
+      "step": 9077
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8326211168195132e-05,
+      "loss": 0.0007,
+      "step": 9078
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8323810438373268e-05,
+      "loss": 0.0006,
+      "step": 9079
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.83214097085514e-05,
+      "loss": 0.0007,
+      "step": 9080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8319008978729535e-05,
+      "loss": 0.0003,
+      "step": 9081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8316608248907666e-05,
+      "loss": 0.0002,
+      "step": 9082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8314207519085805e-05,
+      "loss": 0.0009,
+      "step": 9083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.831180678926394e-05,
+      "loss": 0.0689,
+      "step": 9084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8309406059442072e-05,
+      "loss": 0.0004,
+      "step": 9085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8307005329620207e-05,
+      "loss": 0.0007,
+      "step": 9086
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.830460459979834e-05,
+      "loss": 0.0004,
+      "step": 9087
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8302203869976474e-05,
+      "loss": 0.001,
+      "step": 9088
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8299803140154606e-05,
+      "loss": 0.0008,
+      "step": 9089
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.829740241033274e-05,
+      "loss": 0.001,
+      "step": 9090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8295001680510873e-05,
+      "loss": 0.0007,
+      "step": 9091
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8292600950689015e-05,
+      "loss": 0.0011,
+      "step": 9092
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8290200220867147e-05,
+      "loss": 0.0005,
+      "step": 9093
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8287799491045282e-05,
+      "loss": 0.0007,
+      "step": 9094
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8285398761223414e-05,
+      "loss": 0.0007,
+      "step": 9095
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.828299803140155e-05,
+      "loss": 0.0007,
+      "step": 9096
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.828059730157968e-05,
+      "loss": 0.4042,
+      "step": 9097
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8278196571757816e-05,
+      "loss": 0.3741,
+      "step": 9098
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8275795841935948e-05,
+      "loss": 0.0008,
+      "step": 9099
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8273395112114083e-05,
+      "loss": 0.0008,
+      "step": 9100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.827099438229222e-05,
+      "loss": 0.282,
+      "step": 9101
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8268593652470353e-05,
+      "loss": 0.0009,
+      "step": 9102
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.826619292264849e-05,
+      "loss": 0.0118,
+      "step": 9103
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.826379219282662e-05,
+      "loss": 0.0006,
+      "step": 9104
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8261391463004755e-05,
+      "loss": 0.0007,
+      "step": 9105
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8258990733182887e-05,
+      "loss": 0.0005,
+      "step": 9106
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8256590003361022e-05,
+      "loss": 0.0006,
+      "step": 9107
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8254189273539154e-05,
+      "loss": 0.0004,
+      "step": 9108
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.825178854371729e-05,
+      "loss": 0.0007,
+      "step": 9109
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8249387813895428e-05,
+      "loss": 0.0008,
+      "step": 9110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8246987084073563e-05,
+      "loss": 0.0007,
+      "step": 9111
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8244586354251695e-05,
+      "loss": 0.0007,
+      "step": 9112
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.824218562442983e-05,
+      "loss": 0.0007,
+      "step": 9113
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8239784894607962e-05,
+      "loss": 0.0006,
+      "step": 9114
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8237384164786097e-05,
+      "loss": 0.001,
+      "step": 9115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.823498343496423e-05,
+      "loss": 0.0006,
+      "step": 9116
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8232582705142364e-05,
+      "loss": 0.0009,
+      "step": 9117
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8230181975320496e-05,
+      "loss": 0.0006,
+      "step": 9118
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8227781245498635e-05,
+      "loss": 0.0006,
+      "step": 9119
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.822538051567677e-05,
+      "loss": 0.0005,
+      "step": 9120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.82229797858549e-05,
+      "loss": 0.0007,
+      "step": 9121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8220579056033037e-05,
+      "loss": 0.0007,
+      "step": 9122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.821817832621117e-05,
+      "loss": 0.2074,
+      "step": 9123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8215777596389304e-05,
+      "loss": 0.0008,
+      "step": 9124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8213376866567436e-05,
+      "loss": 0.0006,
+      "step": 9125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.821097613674557e-05,
+      "loss": 0.0006,
+      "step": 9126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8208575406923702e-05,
+      "loss": 0.0004,
+      "step": 9127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8206174677101838e-05,
+      "loss": 0.0008,
+      "step": 9128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8203773947279976e-05,
+      "loss": 0.7524,
+      "step": 9129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.820137321745811e-05,
+      "loss": 0.0009,
+      "step": 9130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8198972487636243e-05,
+      "loss": 0.0006,
+      "step": 9131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.819657175781438e-05,
+      "loss": 0.0008,
+      "step": 9132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.819417102799251e-05,
+      "loss": 0.0003,
+      "step": 9133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8191770298170645e-05,
+      "loss": 0.0007,
+      "step": 9134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8189369568348777e-05,
+      "loss": 0.0005,
+      "step": 9135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8186968838526912e-05,
+      "loss": 0.0007,
+      "step": 9136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8184568108705044e-05,
+      "loss": 0.0006,
+      "step": 9137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8182167378883183e-05,
+      "loss": 0.0019,
+      "step": 9138
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8179766649061318e-05,
+      "loss": 0.0003,
+      "step": 9139
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.817736591923945e-05,
+      "loss": 0.001,
+      "step": 9140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8174965189417585e-05,
+      "loss": 0.0006,
+      "step": 9141
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8172564459595717e-05,
+      "loss": 0.0006,
+      "step": 9142
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8170163729773852e-05,
+      "loss": 0.0003,
+      "step": 9143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8167762999951984e-05,
+      "loss": 0.0006,
+      "step": 9144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.816536227013012e-05,
+      "loss": 0.0005,
+      "step": 9145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8162961540308254e-05,
+      "loss": 0.0007,
+      "step": 9146
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8160560810486393e-05,
+      "loss": 0.0006,
+      "step": 9147
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8158160080664525e-05,
+      "loss": 0.001,
+      "step": 9148
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.815575935084266e-05,
+      "loss": 0.0005,
+      "step": 9149
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.815335862102079e-05,
+      "loss": 0.0009,
+      "step": 9150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8150957891198927e-05,
+      "loss": 0.0006,
+      "step": 9151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.814855716137706e-05,
+      "loss": 0.0006,
+      "step": 9152
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8146156431555194e-05,
+      "loss": 0.0005,
+      "step": 9153
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8143755701733325e-05,
+      "loss": 0.0006,
+      "step": 9154
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.814135497191146e-05,
+      "loss": 0.0004,
+      "step": 9155
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.81389542420896e-05,
+      "loss": 0.0006,
+      "step": 9156
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.813655351226773e-05,
+      "loss": 0.0006,
+      "step": 9157
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8134152782445866e-05,
+      "loss": 0.0003,
+      "step": 9158
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8131752052623998e-05,
+      "loss": 0.0007,
+      "step": 9159
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8129351322802133e-05,
+      "loss": 0.0003,
+      "step": 9160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.812695059298027e-05,
+      "loss": 0.0003,
+      "step": 9161
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.81245498631584e-05,
+      "loss": 0.0006,
+      "step": 9162
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8122149133336535e-05,
+      "loss": 0.0004,
+      "step": 9163
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8119748403514667e-05,
+      "loss": 0.0007,
+      "step": 9164
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8117347673692806e-05,
+      "loss": 0.5744,
+      "step": 9165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.811494694387094e-05,
+      "loss": 0.0005,
+      "step": 9166
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8112546214049073e-05,
+      "loss": 0.0004,
+      "step": 9167
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8110145484227208e-05,
+      "loss": 0.0007,
+      "step": 9168
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.810774475440534e-05,
+      "loss": 0.0006,
+      "step": 9169
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8105344024583475e-05,
+      "loss": 0.0004,
+      "step": 9170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8102943294761607e-05,
+      "loss": 0.0005,
+      "step": 9171
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8100542564939742e-05,
+      "loss": 0.0005,
+      "step": 9172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8098141835117874e-05,
+      "loss": 0.0006,
+      "step": 9173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8095741105296012e-05,
+      "loss": 0.0006,
+      "step": 9174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8093340375474147e-05,
+      "loss": 0.0005,
+      "step": 9175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8090939645652283e-05,
+      "loss": 0.5783,
+      "step": 9176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8088538915830414e-05,
+      "loss": 0.0005,
+      "step": 9177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.808613818600855e-05,
+      "loss": 0.0008,
+      "step": 9178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.808373745618668e-05,
+      "loss": 0.0005,
+      "step": 9179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8081336726364817e-05,
+      "loss": 1.7092,
+      "step": 9180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.807893599654295e-05,
+      "loss": 0.0012,
+      "step": 9181
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8076535266721084e-05,
+      "loss": 0.0006,
+      "step": 9182
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8074134536899222e-05,
+      "loss": 0.0005,
+      "step": 9183
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8071733807077354e-05,
+      "loss": 0.0008,
+      "step": 9184
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.806933307725549e-05,
+      "loss": 0.0005,
+      "step": 9185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.806693234743362e-05,
+      "loss": 0.0014,
+      "step": 9186
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8064531617611756e-05,
+      "loss": 0.0003,
+      "step": 9187
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8062130887789888e-05,
+      "loss": 0.0004,
+      "step": 9188
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8059730157968023e-05,
+      "loss": 0.0007,
+      "step": 9189
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8057329428146155e-05,
+      "loss": 0.001,
+      "step": 9190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.805492869832429e-05,
+      "loss": 0.0006,
+      "step": 9191
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8052527968502422e-05,
+      "loss": 0.0005,
+      "step": 9192
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8050127238680564e-05,
+      "loss": 0.0003,
+      "step": 9193
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8047726508858696e-05,
+      "loss": 0.0003,
+      "step": 9194
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.804532577903683e-05,
+      "loss": 0.0004,
+      "step": 9195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8042925049214963e-05,
+      "loss": 0.0008,
+      "step": 9196
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8040524319393098e-05,
+      "loss": 0.9495,
+      "step": 9197
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.803812358957123e-05,
+      "loss": 0.7455,
+      "step": 9198
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8035722859749365e-05,
+      "loss": 0.0007,
+      "step": 9199
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8033322129927497e-05,
+      "loss": 0.0004,
+      "step": 9200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8030921400105632e-05,
+      "loss": 0.001,
+      "step": 9201
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.802852067028377e-05,
+      "loss": 0.0006,
+      "step": 9202
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8026119940461902e-05,
+      "loss": 0.0007,
+      "step": 9203
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8023719210640037e-05,
+      "loss": 0.0023,
+      "step": 9204
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.802131848081817e-05,
+      "loss": 0.0004,
+      "step": 9205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8018917750996304e-05,
+      "loss": 0.0006,
+      "step": 9206
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8016517021174436e-05,
+      "loss": 0.0006,
+      "step": 9207
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.801411629135257e-05,
+      "loss": 0.0009,
+      "step": 9208
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8011715561530703e-05,
+      "loss": 0.0008,
+      "step": 9209
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.800931483170884e-05,
+      "loss": 0.0005,
+      "step": 9210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8006914101886977e-05,
+      "loss": 0.0005,
+      "step": 9211
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8004513372065112e-05,
+      "loss": 0.0006,
+      "step": 9212
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8002112642243244e-05,
+      "loss": 0.0008,
+      "step": 9213
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.799971191242138e-05,
+      "loss": 0.0006,
+      "step": 9214
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.799731118259951e-05,
+      "loss": 0.0007,
+      "step": 9215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7994910452777646e-05,
+      "loss": 0.0005,
+      "step": 9216
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7992509722955778e-05,
+      "loss": 0.0004,
+      "step": 9217
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7990108993133913e-05,
+      "loss": 0.0005,
+      "step": 9218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7987708263312045e-05,
+      "loss": 0.0018,
+      "step": 9219
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7985307533490184e-05,
+      "loss": 0.0003,
+      "step": 9220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.798290680366832e-05,
+      "loss": 0.0009,
+      "step": 9221
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.798050607384645e-05,
+      "loss": 2.1323,
+      "step": 9222
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7978105344024586e-05,
+      "loss": 0.0007,
+      "step": 9223
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7975704614202718e-05,
+      "loss": 0.0008,
+      "step": 9224
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7973303884380853e-05,
+      "loss": 0.0005,
+      "step": 9225
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7970903154558985e-05,
+      "loss": 0.0006,
+      "step": 9226
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.796850242473712e-05,
+      "loss": 0.0009,
+      "step": 9227
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7966101694915255e-05,
+      "loss": 0.0009,
+      "step": 9228
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7963700965093393e-05,
+      "loss": 0.0012,
+      "step": 9229
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7961300235271525e-05,
+      "loss": 0.0008,
+      "step": 9230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.795889950544966e-05,
+      "loss": 0.0008,
+      "step": 9231
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7956498775627792e-05,
+      "loss": 0.001,
+      "step": 9232
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7954098045805927e-05,
+      "loss": 0.0007,
+      "step": 9233
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.795169731598406e-05,
+      "loss": 0.0011,
+      "step": 9234
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7949296586162194e-05,
+      "loss": 0.0008,
+      "step": 9235
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7946895856340326e-05,
+      "loss": 0.0008,
+      "step": 9236
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.794449512651846e-05,
+      "loss": 0.0008,
+      "step": 9237
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.79420943966966e-05,
+      "loss": 0.001,
+      "step": 9238
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7939693666874732e-05,
+      "loss": 0.001,
+      "step": 9239
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7937292937052867e-05,
+      "loss": 0.001,
+      "step": 9240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7934892207231e-05,
+      "loss": 0.0008,
+      "step": 9241
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7932491477409134e-05,
+      "loss": 0.001,
+      "step": 9242
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.793009074758727e-05,
+      "loss": 0.0007,
+      "step": 9243
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.79276900177654e-05,
+      "loss": 0.0008,
+      "step": 9244
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7925289287943536e-05,
+      "loss": 0.0007,
+      "step": 9245
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7922888558121668e-05,
+      "loss": 0.0007,
+      "step": 9246
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7920487828299807e-05,
+      "loss": 0.0006,
+      "step": 9247
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7918087098477942e-05,
+      "loss": 0.0006,
+      "step": 9248
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7915686368656074e-05,
+      "loss": 0.0006,
+      "step": 9249
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.791328563883421e-05,
+      "loss": 0.0006,
+      "step": 9250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.791088490901234e-05,
+      "loss": 0.0009,
+      "step": 9251
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7908484179190476e-05,
+      "loss": 0.0005,
+      "step": 9252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7906083449368607e-05,
+      "loss": 0.0005,
+      "step": 9253
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7903682719546743e-05,
+      "loss": 0.0007,
+      "step": 9254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7901281989724874e-05,
+      "loss": 0.0006,
+      "step": 9255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.789888125990301e-05,
+      "loss": 0.0007,
+      "step": 9256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7896480530081148e-05,
+      "loss": 0.0007,
+      "step": 9257
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7894079800259283e-05,
+      "loss": 0.255,
+      "step": 9258
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7891679070437415e-05,
+      "loss": 0.0007,
+      "step": 9259
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.788927834061555e-05,
+      "loss": 0.0005,
+      "step": 9260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7886877610793682e-05,
+      "loss": 0.0005,
+      "step": 9261
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7884476880971817e-05,
+      "loss": 0.0006,
+      "step": 9262
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.788207615114995e-05,
+      "loss": 0.0007,
+      "step": 9263
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7879675421328084e-05,
+      "loss": 0.0005,
+      "step": 9264
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7877274691506216e-05,
+      "loss": 0.0007,
+      "step": 9265
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7874873961684355e-05,
+      "loss": 0.0006,
+      "step": 9266
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.787247323186249e-05,
+      "loss": 0.0005,
+      "step": 9267
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7870072502040622e-05,
+      "loss": 0.0005,
+      "step": 9268
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7867671772218757e-05,
+      "loss": 0.0006,
+      "step": 9269
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786527104239689e-05,
+      "loss": 0.0004,
+      "step": 9270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7862870312575024e-05,
+      "loss": 0.0006,
+      "step": 9271
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7860469582753156e-05,
+      "loss": 0.0005,
+      "step": 9272
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.785806885293129e-05,
+      "loss": 0.0003,
+      "step": 9273
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7855668123109423e-05,
+      "loss": 0.0004,
+      "step": 9274
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7853267393287565e-05,
+      "loss": 0.0005,
+      "step": 9275
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7850866663465697e-05,
+      "loss": 0.0003,
+      "step": 9276
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.784846593364383e-05,
+      "loss": 0.0005,
+      "step": 9277
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7846065203821963e-05,
+      "loss": 0.9102,
+      "step": 9278
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.78436644740001e-05,
+      "loss": 0.0004,
+      "step": 9279
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.784126374417823e-05,
+      "loss": 0.0004,
+      "step": 9280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7838863014356366e-05,
+      "loss": 0.0006,
+      "step": 9281
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7836462284534497e-05,
+      "loss": 0.0005,
+      "step": 9282
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7834061554712633e-05,
+      "loss": 0.8321,
+      "step": 9283
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.783166082489077e-05,
+      "loss": 0.0004,
+      "step": 9284
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7829260095068903e-05,
+      "loss": 0.9385,
+      "step": 9285
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7826859365247038e-05,
+      "loss": 0.0004,
+      "step": 9286
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.782445863542517e-05,
+      "loss": 0.0006,
+      "step": 9287
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7822057905603305e-05,
+      "loss": 0.0006,
+      "step": 9288
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7819657175781437e-05,
+      "loss": 1.6176,
+      "step": 9289
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7817256445959572e-05,
+      "loss": 0.0006,
+      "step": 9290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7814855716137704e-05,
+      "loss": 0.0006,
+      "step": 9291
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.781245498631584e-05,
+      "loss": 0.0008,
+      "step": 9292
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7810054256493978e-05,
+      "loss": 0.0005,
+      "step": 9293
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7807653526672113e-05,
+      "loss": 0.0007,
+      "step": 9294
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7805252796850245e-05,
+      "loss": 0.0006,
+      "step": 9295
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.780285206702838e-05,
+      "loss": 0.001,
+      "step": 9296
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7800451337206512e-05,
+      "loss": 0.001,
+      "step": 9297
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7798050607384647e-05,
+      "loss": 0.0008,
+      "step": 9298
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.779564987756278e-05,
+      "loss": 0.0009,
+      "step": 9299
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7793249147740914e-05,
+      "loss": 0.0007,
+      "step": 9300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7790848417919046e-05,
+      "loss": 0.0009,
+      "step": 9301
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7788447688097184e-05,
+      "loss": 0.0008,
+      "step": 9302
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.778604695827532e-05,
+      "loss": 0.0007,
+      "step": 9303
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.778364622845345e-05,
+      "loss": 0.0007,
+      "step": 9304
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7781245498631586e-05,
+      "loss": 0.0007,
+      "step": 9305
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7778844768809718e-05,
+      "loss": 0.0007,
+      "step": 9306
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7776444038987853e-05,
+      "loss": 0.0007,
+      "step": 9307
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7774043309165985e-05,
+      "loss": 0.0053,
+      "step": 9308
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.777164257934412e-05,
+      "loss": 0.0009,
+      "step": 9309
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7769241849522252e-05,
+      "loss": 0.0007,
+      "step": 9310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7766841119700387e-05,
+      "loss": 0.0007,
+      "step": 9311
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7764440389878526e-05,
+      "loss": 0.0139,
+      "step": 9312
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.776203966005666e-05,
+      "loss": 0.0006,
+      "step": 9313
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7759638930234793e-05,
+      "loss": 0.0006,
+      "step": 9314
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7757238200412928e-05,
+      "loss": 0.0008,
+      "step": 9315
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.775483747059106e-05,
+      "loss": 0.0006,
+      "step": 9316
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7752436740769195e-05,
+      "loss": 0.0007,
+      "step": 9317
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7750036010947327e-05,
+      "loss": 0.0009,
+      "step": 9318
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7747635281125462e-05,
+      "loss": 0.0016,
+      "step": 9319
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7745234551303594e-05,
+      "loss": 0.0008,
+      "step": 9320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7742833821481733e-05,
+      "loss": 0.0006,
+      "step": 9321
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7740433091659868e-05,
+      "loss": 0.0005,
+      "step": 9322
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7738032361838e-05,
+      "loss": 0.0006,
+      "step": 9323
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7735631632016135e-05,
+      "loss": 0.0006,
+      "step": 9324
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.773323090219427e-05,
+      "loss": 0.7997,
+      "step": 9325
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7730830172372402e-05,
+      "loss": 0.0005,
+      "step": 9326
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7728429442550537e-05,
+      "loss": 0.001,
+      "step": 9327
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.772602871272867e-05,
+      "loss": 0.0014,
+      "step": 9328
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7723627982906804e-05,
+      "loss": 0.0007,
+      "step": 9329
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7721227253084942e-05,
+      "loss": 0.0007,
+      "step": 9330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7718826523263074e-05,
+      "loss": 0.7805,
+      "step": 9331
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.771642579344121e-05,
+      "loss": 0.0007,
+      "step": 9332
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.771402506361934e-05,
+      "loss": 0.0006,
+      "step": 9333
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7711624333797476e-05,
+      "loss": 0.0007,
+      "step": 9334
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7709223603975608e-05,
+      "loss": 0.0009,
+      "step": 9335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7706822874153743e-05,
+      "loss": 0.002,
+      "step": 9336
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7704422144331875e-05,
+      "loss": 0.0352,
+      "step": 9337
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.770202141451001e-05,
+      "loss": 0.0019,
+      "step": 9338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.769962068468815e-05,
+      "loss": 0.0011,
+      "step": 9339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7697219954866284e-05,
+      "loss": 0.0008,
+      "step": 9340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7694819225044416e-05,
+      "loss": 0.0006,
+      "step": 9341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.769241849522255e-05,
+      "loss": 0.0011,
+      "step": 9342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7690017765400683e-05,
+      "loss": 0.002,
+      "step": 9343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7687617035578818e-05,
+      "loss": 0.0017,
+      "step": 9344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.768521630575695e-05,
+      "loss": 0.0012,
+      "step": 9345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7682815575935085e-05,
+      "loss": 0.0007,
+      "step": 9346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7680414846113217e-05,
+      "loss": 0.0006,
+      "step": 9347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7678014116291356e-05,
+      "loss": 0.0012,
+      "step": 9348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.767561338646949e-05,
+      "loss": 0.0012,
+      "step": 9349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7673212656647623e-05,
+      "loss": 0.0016,
+      "step": 9350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7670811926825758e-05,
+      "loss": 0.0007,
+      "step": 9351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.766841119700389e-05,
+      "loss": 0.0004,
+      "step": 9352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7666010467182025e-05,
+      "loss": 0.0009,
+      "step": 9353
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7663609737360156e-05,
+      "loss": 0.0004,
+      "step": 9354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.766120900753829e-05,
+      "loss": 0.0005,
+      "step": 9355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7658808277716423e-05,
+      "loss": 0.0007,
+      "step": 9356
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7656407547894565e-05,
+      "loss": 0.0006,
+      "step": 9357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7654006818072697e-05,
+      "loss": 0.0006,
+      "step": 9358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7651606088250832e-05,
+      "loss": 0.0006,
+      "step": 9359
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7649205358428964e-05,
+      "loss": 0.0007,
+      "step": 9360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.76468046286071e-05,
+      "loss": 0.0011,
+      "step": 9361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.764440389878523e-05,
+      "loss": 0.0004,
+      "step": 9362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7642003168963366e-05,
+      "loss": 0.0004,
+      "step": 9363
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7639602439141498e-05,
+      "loss": 0.0003,
+      "step": 9364
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7637201709319633e-05,
+      "loss": 0.0006,
+      "step": 9365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7634800979497772e-05,
+      "loss": 0.0002,
+      "step": 9366
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7632400249675904e-05,
+      "loss": 0.0007,
+      "step": 9367
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.762999951985404e-05,
+      "loss": 0.001,
+      "step": 9368
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.762759879003217e-05,
+      "loss": 0.0004,
+      "step": 9369
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7625198060210306e-05,
+      "loss": 0.0006,
+      "step": 9370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7622797330388438e-05,
+      "loss": 0.0008,
+      "step": 9371
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7620396600566573e-05,
+      "loss": 0.0009,
+      "step": 9372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7617995870744705e-05,
+      "loss": 0.0006,
+      "step": 9373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.761559514092284e-05,
+      "loss": 0.0007,
+      "step": 9374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7613194411100972e-05,
+      "loss": 0.0004,
+      "step": 9375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7610793681279114e-05,
+      "loss": 0.0006,
+      "step": 9376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7608392951457246e-05,
+      "loss": 0.0003,
+      "step": 9377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.760599222163538e-05,
+      "loss": 0.0008,
+      "step": 9378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7603591491813512e-05,
+      "loss": 0.0005,
+      "step": 9379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7601190761991648e-05,
+      "loss": 0.0006,
+      "step": 9380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.759879003216978e-05,
+      "loss": 0.0003,
+      "step": 9381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7596389302347915e-05,
+      "loss": 0.0003,
+      "step": 9382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7593988572526046e-05,
+      "loss": 0.0023,
+      "step": 9383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.759158784270418e-05,
+      "loss": 0.0005,
+      "step": 9384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.758918711288232e-05,
+      "loss": 0.0005,
+      "step": 9385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7586786383060452e-05,
+      "loss": 0.0006,
+      "step": 9386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7584385653238587e-05,
+      "loss": 1.1479,
+      "step": 9387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.758198492341672e-05,
+      "loss": 0.0007,
+      "step": 9388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7579584193594854e-05,
+      "loss": 0.0006,
+      "step": 9389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7577183463772986e-05,
+      "loss": 0.0007,
+      "step": 9390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.757478273395112e-05,
+      "loss": 0.0009,
+      "step": 9391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7572382004129253e-05,
+      "loss": 0.0009,
+      "step": 9392
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7569981274307388e-05,
+      "loss": 0.0009,
+      "step": 9393
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7567580544485527e-05,
+      "loss": 0.0009,
+      "step": 9394
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7565179814663662e-05,
+      "loss": 0.001,
+      "step": 9395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7562779084841794e-05,
+      "loss": 0.0008,
+      "step": 9396
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.756037835501993e-05,
+      "loss": 0.0009,
+      "step": 9397
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.755797762519806e-05,
+      "loss": 0.0007,
+      "step": 9398
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7555576895376196e-05,
+      "loss": 0.0007,
+      "step": 9399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7553176165554328e-05,
+      "loss": 0.0008,
+      "step": 9400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7550775435732463e-05,
+      "loss": 0.0008,
+      "step": 9401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7548374705910595e-05,
+      "loss": 0.0006,
+      "step": 9402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7545973976088733e-05,
+      "loss": 0.0006,
+      "step": 9403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.754357324626687e-05,
+      "loss": 0.0007,
+      "step": 9404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7541172516445e-05,
+      "loss": 0.0007,
+      "step": 9405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7538771786623135e-05,
+      "loss": 0.0007,
+      "step": 9406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7536371056801267e-05,
+      "loss": 0.0007,
+      "step": 9407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7533970326979402e-05,
+      "loss": 0.0008,
+      "step": 9408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7531569597157538e-05,
+      "loss": 0.0009,
+      "step": 9409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.752916886733567e-05,
+      "loss": 0.0008,
+      "step": 9410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7526768137513805e-05,
+      "loss": 0.0006,
+      "step": 9411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7524367407691943e-05,
+      "loss": 0.0007,
+      "step": 9412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7521966677870075e-05,
+      "loss": 0.0007,
+      "step": 9413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.751956594804821e-05,
+      "loss": 0.0007,
+      "step": 9414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7517165218226342e-05,
+      "loss": 0.0006,
+      "step": 9415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7514764488404477e-05,
+      "loss": 0.0007,
+      "step": 9416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.751236375858261e-05,
+      "loss": 0.0005,
+      "step": 9417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7509963028760744e-05,
+      "loss": 0.8693,
+      "step": 9418
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7507562298938876e-05,
+      "loss": 0.0008,
+      "step": 9419
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.750516156911701e-05,
+      "loss": 0.0007,
+      "step": 9420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.750276083929515e-05,
+      "loss": 0.0007,
+      "step": 9421
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.750036010947328e-05,
+      "loss": 0.0006,
+      "step": 9422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7497959379651417e-05,
+      "loss": 0.0006,
+      "step": 9423
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7495558649829552e-05,
+      "loss": 0.0008,
+      "step": 9424
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7493157920007684e-05,
+      "loss": 0.0008,
+      "step": 9425
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.749075719018582e-05,
+      "loss": 0.0012,
+      "step": 9426
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.748835646036395e-05,
+      "loss": 0.0008,
+      "step": 9427
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7485955730542086e-05,
+      "loss": 0.0006,
+      "step": 9428
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7483555000720218e-05,
+      "loss": 0.0011,
+      "step": 9429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7481154270898356e-05,
+      "loss": 0.0006,
+      "step": 9430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.747875354107649e-05,
+      "loss": 0.0007,
+      "step": 9431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7476352811254623e-05,
+      "loss": 0.0009,
+      "step": 9432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.747395208143276e-05,
+      "loss": 0.001,
+      "step": 9433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.747155135161089e-05,
+      "loss": 0.0008,
+      "step": 9434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7469150621789025e-05,
+      "loss": 0.0008,
+      "step": 9435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7466749891967157e-05,
+      "loss": 0.0008,
+      "step": 9436
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7464349162145292e-05,
+      "loss": 0.0007,
+      "step": 9437
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7461948432323424e-05,
+      "loss": 0.0008,
+      "step": 9438
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.745954770250156e-05,
+      "loss": 0.001,
+      "step": 9439
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7457146972679698e-05,
+      "loss": 0.001,
+      "step": 9440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7454746242857833e-05,
+      "loss": 0.0006,
+      "step": 9441
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7452345513035965e-05,
+      "loss": 0.0008,
+      "step": 9442
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.74499447832141e-05,
+      "loss": 0.001,
+      "step": 9443
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7447544053392232e-05,
+      "loss": 0.0007,
+      "step": 9444
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7445143323570367e-05,
+      "loss": 0.5318,
+      "step": 9445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.74427425937485e-05,
+      "loss": 0.0009,
+      "step": 9446
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7440341863926634e-05,
+      "loss": 0.0007,
+      "step": 9447
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7437941134104766e-05,
+      "loss": 0.0007,
+      "step": 9448
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7435540404282905e-05,
+      "loss": 0.0006,
+      "step": 9449
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.743313967446104e-05,
+      "loss": 0.0008,
+      "step": 9450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.743073894463917e-05,
+      "loss": 0.0007,
+      "step": 9451
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7428338214817307e-05,
+      "loss": 0.0007,
+      "step": 9452
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.742593748499544e-05,
+      "loss": 0.96,
+      "step": 9453
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7423536755173574e-05,
+      "loss": 0.0007,
+      "step": 9454
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7421136025351705e-05,
+      "loss": 0.0008,
+      "step": 9455
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.741873529552984e-05,
+      "loss": 0.0009,
+      "step": 9456
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7416334565707972e-05,
+      "loss": 0.0012,
+      "step": 9457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7413933835886114e-05,
+      "loss": 0.0009,
+      "step": 9458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7411533106064246e-05,
+      "loss": 0.0007,
+      "step": 9459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.740913237624238e-05,
+      "loss": 0.0008,
+      "step": 9460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7406731646420513e-05,
+      "loss": 0.001,
+      "step": 9461
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.740433091659865e-05,
+      "loss": 0.0007,
+      "step": 9462
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.740193018677678e-05,
+      "loss": 0.0008,
+      "step": 9463
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7399529456954915e-05,
+      "loss": 0.0008,
+      "step": 9464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7397128727133047e-05,
+      "loss": 0.0008,
+      "step": 9465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7394727997311182e-05,
+      "loss": 0.0008,
+      "step": 9466
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.739232726748932e-05,
+      "loss": 0.0007,
+      "step": 9467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7389926537667453e-05,
+      "loss": 0.0007,
+      "step": 9468
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7387525807845588e-05,
+      "loss": 0.001,
+      "step": 9469
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.738512507802372e-05,
+      "loss": 0.0008,
+      "step": 9470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7382724348201855e-05,
+      "loss": 0.0073,
+      "step": 9471
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7380323618379987e-05,
+      "loss": 0.0008,
+      "step": 9472
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7377922888558122e-05,
+      "loss": 0.0007,
+      "step": 9473
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7375522158736254e-05,
+      "loss": 0.0007,
+      "step": 9474
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.737312142891439e-05,
+      "loss": 0.0008,
+      "step": 9475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7370720699092528e-05,
+      "loss": 0.0007,
+      "step": 9476
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7368319969270663e-05,
+      "loss": 0.0007,
+      "step": 9477
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7365919239448795e-05,
+      "loss": 0.0007,
+      "step": 9478
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.736351850962693e-05,
+      "loss": 0.0007,
+      "step": 9479
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.736111777980506e-05,
+      "loss": 0.0008,
+      "step": 9480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7358717049983197e-05,
+      "loss": 0.0009,
+      "step": 9481
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.735631632016133e-05,
+      "loss": 0.0006,
+      "step": 9482
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7353915590339464e-05,
+      "loss": 0.0007,
+      "step": 9483
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7351514860517595e-05,
+      "loss": 0.0007,
+      "step": 9484
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7349114130695734e-05,
+      "loss": 0.0007,
+      "step": 9485
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.734671340087387e-05,
+      "loss": 0.0009,
+      "step": 9486
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7344312671052e-05,
+      "loss": 0.0008,
+      "step": 9487
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7341911941230136e-05,
+      "loss": 0.0008,
+      "step": 9488
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7339511211408268e-05,
+      "loss": 0.0006,
+      "step": 9489
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7337110481586403e-05,
+      "loss": 0.0007,
+      "step": 9490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.733470975176454e-05,
+      "loss": 0.0008,
+      "step": 9491
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.733230902194267e-05,
+      "loss": 0.0008,
+      "step": 9492
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7329908292120805e-05,
+      "loss": 0.0007,
+      "step": 9493
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7327507562298944e-05,
+      "loss": 0.0008,
+      "step": 9494
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7325106832477076e-05,
+      "loss": 0.0006,
+      "step": 9495
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.732270610265521e-05,
+      "loss": 0.0007,
+      "step": 9496
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7320305372833343e-05,
+      "loss": 0.0006,
+      "step": 9497
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7317904643011478e-05,
+      "loss": 0.0008,
+      "step": 9498
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.731550391318961e-05,
+      "loss": 0.0007,
+      "step": 9499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7313103183367745e-05,
+      "loss": 0.0006,
+      "step": 9500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7310702453545877e-05,
+      "loss": 0.0006,
+      "step": 9501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7308301723724012e-05,
+      "loss": 0.0007,
+      "step": 9502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7305900993902144e-05,
+      "loss": 0.0006,
+      "step": 9503
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7303500264080282e-05,
+      "loss": 0.0006,
+      "step": 9504
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7301099534258417e-05,
+      "loss": 0.0006,
+      "step": 9505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7298698804436553e-05,
+      "loss": 0.0007,
+      "step": 9506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7296298074614684e-05,
+      "loss": 0.0007,
+      "step": 9507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.729389734479282e-05,
+      "loss": 0.0006,
+      "step": 9508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.729149661497095e-05,
+      "loss": 0.0006,
+      "step": 9509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7289095885149087e-05,
+      "loss": 0.0006,
+      "step": 9510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.728669515532722e-05,
+      "loss": 0.0007,
+      "step": 9511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7284294425505354e-05,
+      "loss": 0.0008,
+      "step": 9512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7281893695683492e-05,
+      "loss": 0.0006,
+      "step": 9513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7279492965861624e-05,
+      "loss": 0.0007,
+      "step": 9514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.727709223603976e-05,
+      "loss": 0.0006,
+      "step": 9515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.727469150621789e-05,
+      "loss": 0.0008,
+      "step": 9516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7272290776396026e-05,
+      "loss": 0.0006,
+      "step": 9517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7269890046574158e-05,
+      "loss": 0.0006,
+      "step": 9518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7267489316752293e-05,
+      "loss": 0.9255,
+      "step": 9519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7265088586930425e-05,
+      "loss": 0.0007,
+      "step": 9520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.726268785710856e-05,
+      "loss": 0.0008,
+      "step": 9521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.72602871272867e-05,
+      "loss": 0.0006,
+      "step": 9522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7257886397464834e-05,
+      "loss": 0.0007,
+      "step": 9523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7255485667642966e-05,
+      "loss": 0.0006,
+      "step": 9524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.72530849378211e-05,
+      "loss": 0.0006,
+      "step": 9525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7250684207999233e-05,
+      "loss": 0.0009,
+      "step": 9526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7248283478177368e-05,
+      "loss": 0.0006,
+      "step": 9527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.72458827483555e-05,
+      "loss": 0.0007,
+      "step": 9528
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7243482018533635e-05,
+      "loss": 0.0007,
+      "step": 9529
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7241081288711767e-05,
+      "loss": 0.0006,
+      "step": 9530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7238680558889905e-05,
+      "loss": 0.0008,
+      "step": 9531
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.723627982906804e-05,
+      "loss": 0.7806,
+      "step": 9532
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7233879099246172e-05,
+      "loss": 0.0007,
+      "step": 9533
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7231478369424307e-05,
+      "loss": 0.0014,
+      "step": 9534
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.722907763960244e-05,
+      "loss": 0.0006,
+      "step": 9535
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7226676909780574e-05,
+      "loss": 0.0006,
+      "step": 9536
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7224276179958706e-05,
+      "loss": 0.0007,
+      "step": 9537
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.722187545013684e-05,
+      "loss": 0.0007,
+      "step": 9538
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7219474720314973e-05,
+      "loss": 0.0006,
+      "step": 9539
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7217073990493115e-05,
+      "loss": 0.0006,
+      "step": 9540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7214673260671247e-05,
+      "loss": 0.0006,
+      "step": 9541
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7212272530849382e-05,
+      "loss": 0.0006,
+      "step": 9542
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7209871801027514e-05,
+      "loss": 0.0007,
+      "step": 9543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.720747107120565e-05,
+      "loss": 0.0007,
+      "step": 9544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.720507034138378e-05,
+      "loss": 0.0006,
+      "step": 9545
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7202669611561916e-05,
+      "loss": 0.0007,
+      "step": 9546
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7200268881740048e-05,
+      "loss": 0.0006,
+      "step": 9547
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7197868151918183e-05,
+      "loss": 0.0006,
+      "step": 9548
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7195467422096322e-05,
+      "loss": 0.0006,
+      "step": 9549
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7193066692274454e-05,
+      "loss": 0.0006,
+      "step": 9550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.719066596245259e-05,
+      "loss": 0.0007,
+      "step": 9551
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.718826523263072e-05,
+      "loss": 0.0006,
+      "step": 9552
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7185864502808856e-05,
+      "loss": 0.0006,
+      "step": 9553
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7183463772986988e-05,
+      "loss": 0.0006,
+      "step": 9554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7181063043165123e-05,
+      "loss": 0.0007,
+      "step": 9555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7178662313343254e-05,
+      "loss": 0.0006,
+      "step": 9556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.717626158352139e-05,
+      "loss": 0.0008,
+      "step": 9557
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7173860853699528e-05,
+      "loss": 0.0006,
+      "step": 9558
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7171460123877663e-05,
+      "loss": 0.0006,
+      "step": 9559
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7169059394055795e-05,
+      "loss": 0.0005,
+      "step": 9560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.716665866423393e-05,
+      "loss": 0.0006,
+      "step": 9561
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7164257934412062e-05,
+      "loss": 0.0006,
+      "step": 9562
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7161857204590197e-05,
+      "loss": 0.0005,
+      "step": 9563
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.715945647476833e-05,
+      "loss": 0.0006,
+      "step": 9564
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7157055744946464e-05,
+      "loss": 0.0006,
+      "step": 9565
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7154655015124596e-05,
+      "loss": 0.0006,
+      "step": 9566
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.715225428530273e-05,
+      "loss": 0.0006,
+      "step": 9567
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.714985355548087e-05,
+      "loss": 0.0005,
+      "step": 9568
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7147452825659002e-05,
+      "loss": 0.0007,
+      "step": 9569
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7145052095837137e-05,
+      "loss": 0.0005,
+      "step": 9570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.714265136601527e-05,
+      "loss": 0.0006,
+      "step": 9571
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7140250636193404e-05,
+      "loss": 0.0006,
+      "step": 9572
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.713784990637154e-05,
+      "loss": 0.0006,
+      "step": 9573
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.713544917654967e-05,
+      "loss": 0.0005,
+      "step": 9574
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7133048446727806e-05,
+      "loss": 0.0006,
+      "step": 9575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7130647716905938e-05,
+      "loss": 0.0005,
+      "step": 9576
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7128246987084077e-05,
+      "loss": 0.0005,
+      "step": 9577
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7125846257262212e-05,
+      "loss": 0.0004,
+      "step": 9578
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7123445527440344e-05,
+      "loss": 0.0005,
+      "step": 9579
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.712104479761848e-05,
+      "loss": 0.0005,
+      "step": 9580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.711864406779661e-05,
+      "loss": 0.0006,
+      "step": 9581
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7116243337974746e-05,
+      "loss": 0.0005,
+      "step": 9582
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7113842608152877e-05,
+      "loss": 0.0005,
+      "step": 9583
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7111441878331013e-05,
+      "loss": 0.0006,
+      "step": 9584
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7109041148509144e-05,
+      "loss": 0.0005,
+      "step": 9585
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7106640418687283e-05,
+      "loss": 0.0005,
+      "step": 9586
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7104239688865418e-05,
+      "loss": 0.0005,
+      "step": 9587
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7101838959043553e-05,
+      "loss": 0.0005,
+      "step": 9588
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7099438229221685e-05,
+      "loss": 0.0005,
+      "step": 9589
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.709703749939982e-05,
+      "loss": 0.0005,
+      "step": 9590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7094636769577952e-05,
+      "loss": 0.0005,
+      "step": 9591
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7092236039756087e-05,
+      "loss": 0.0006,
+      "step": 9592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.708983530993422e-05,
+      "loss": 0.0004,
+      "step": 9593
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7087434580112354e-05,
+      "loss": 0.0005,
+      "step": 9594
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7085033850290493e-05,
+      "loss": 0.0005,
+      "step": 9595
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7082633120468625e-05,
+      "loss": 0.0005,
+      "step": 9596
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.708023239064676e-05,
+      "loss": 0.0005,
+      "step": 9597
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7077831660824892e-05,
+      "loss": 0.0005,
+      "step": 9598
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7075430931003027e-05,
+      "loss": 0.0006,
+      "step": 9599
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.707303020118116e-05,
+      "loss": 0.0005,
+      "step": 9600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7070629471359294e-05,
+      "loss": 0.0005,
+      "step": 9601
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7068228741537426e-05,
+      "loss": 0.0005,
+      "step": 9602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.706582801171556e-05,
+      "loss": 0.0006,
+      "step": 9603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.70634272818937e-05,
+      "loss": 0.0004,
+      "step": 9604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7061026552071835e-05,
+      "loss": 0.0006,
+      "step": 9605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7058625822249966e-05,
+      "loss": 0.0004,
+      "step": 9606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.70562250924281e-05,
+      "loss": 0.0006,
+      "step": 9607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7053824362606233e-05,
+      "loss": 0.0006,
+      "step": 9608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.705142363278437e-05,
+      "loss": 0.0005,
+      "step": 9609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.70490229029625e-05,
+      "loss": 0.0004,
+      "step": 9610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7046622173140636e-05,
+      "loss": 0.0004,
+      "step": 9611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7044221443318767e-05,
+      "loss": 0.0005,
+      "step": 9612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7041820713496906e-05,
+      "loss": 0.0004,
+      "step": 9613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.703941998367504e-05,
+      "loss": 0.0004,
+      "step": 9614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7037019253853173e-05,
+      "loss": 0.0004,
+      "step": 9615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7034618524031308e-05,
+      "loss": 0.0005,
+      "step": 9616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.703221779420944e-05,
+      "loss": 0.0005,
+      "step": 9617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7029817064387575e-05,
+      "loss": 0.0004,
+      "step": 9618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7027416334565707e-05,
+      "loss": 0.0005,
+      "step": 9619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7025015604743842e-05,
+      "loss": 0.0004,
+      "step": 9620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7022614874921974e-05,
+      "loss": 0.0005,
+      "step": 9621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.702021414510011e-05,
+      "loss": 0.0005,
+      "step": 9622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7017813415278248e-05,
+      "loss": 0.0005,
+      "step": 9623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7015412685456383e-05,
+      "loss": 0.0005,
+      "step": 9624
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7013011955634515e-05,
+      "loss": 0.0005,
+      "step": 9625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.701061122581265e-05,
+      "loss": 0.0004,
+      "step": 9626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7008210495990782e-05,
+      "loss": 0.0004,
+      "step": 9627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7005809766168917e-05,
+      "loss": 0.0004,
+      "step": 9628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.700340903634705e-05,
+      "loss": 0.0005,
+      "step": 9629
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7001008306525184e-05,
+      "loss": 0.0004,
+      "step": 9630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6998607576703316e-05,
+      "loss": 0.0004,
+      "step": 9631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6996206846881454e-05,
+      "loss": 0.0005,
+      "step": 9632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.699380611705959e-05,
+      "loss": 0.0005,
+      "step": 9633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.699140538723772e-05,
+      "loss": 0.0004,
+      "step": 9634
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6989004657415856e-05,
+      "loss": 1.0118,
+      "step": 9635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6986603927593988e-05,
+      "loss": 0.0004,
+      "step": 9636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6984203197772123e-05,
+      "loss": 0.0004,
+      "step": 9637
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6981802467950255e-05,
+      "loss": 0.0004,
+      "step": 9638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.697940173812839e-05,
+      "loss": 0.0005,
+      "step": 9639
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6977001008306522e-05,
+      "loss": 0.0917,
+      "step": 9640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6974600278484664e-05,
+      "loss": 0.0004,
+      "step": 9641
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6972199548662796e-05,
+      "loss": 0.0004,
+      "step": 9642
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.696979881884093e-05,
+      "loss": 0.0004,
+      "step": 9643
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6967398089019063e-05,
+      "loss": 0.0006,
+      "step": 9644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6964997359197198e-05,
+      "loss": 0.9624,
+      "step": 9645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.696259662937533e-05,
+      "loss": 0.0004,
+      "step": 9646
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6960195899553465e-05,
+      "loss": 0.0005,
+      "step": 9647
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6957795169731597e-05,
+      "loss": 0.0074,
+      "step": 9648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6955394439909732e-05,
+      "loss": 0.0006,
+      "step": 9649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.695299371008787e-05,
+      "loss": 0.0006,
+      "step": 9650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6950592980266003e-05,
+      "loss": 0.9876,
+      "step": 9651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6948192250444138e-05,
+      "loss": 0.0006,
+      "step": 9652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.694579152062227e-05,
+      "loss": 0.0007,
+      "step": 9653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6943390790800405e-05,
+      "loss": 0.0008,
+      "step": 9654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6940990060978537e-05,
+      "loss": 0.0008,
+      "step": 9655
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6938589331156672e-05,
+      "loss": 0.0009,
+      "step": 9656
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6936188601334807e-05,
+      "loss": 0.0008,
+      "step": 9657
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.693378787151294e-05,
+      "loss": 0.0007,
+      "step": 9658
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6931387141691077e-05,
+      "loss": 0.0009,
+      "step": 9659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6928986411869212e-05,
+      "loss": 0.0009,
+      "step": 9660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6926585682047344e-05,
+      "loss": 0.001,
+      "step": 9661
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.692418495222548e-05,
+      "loss": 0.0007,
+      "step": 9662
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.692178422240361e-05,
+      "loss": 0.0061,
+      "step": 9663
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6919383492581746e-05,
+      "loss": 0.0007,
+      "step": 9664
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6916982762759878e-05,
+      "loss": 0.0011,
+      "step": 9665
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6914582032938013e-05,
+      "loss": 0.0006,
+      "step": 9666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6912181303116145e-05,
+      "loss": 0.001,
+      "step": 9667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6909780573294284e-05,
+      "loss": 0.0007,
+      "step": 9668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.690737984347242e-05,
+      "loss": 0.0007,
+      "step": 9669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.690497911365055e-05,
+      "loss": 0.0008,
+      "step": 9670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6902578383828686e-05,
+      "loss": 0.0006,
+      "step": 9671
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.690017765400682e-05,
+      "loss": 0.0008,
+      "step": 9672
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6897776924184953e-05,
+      "loss": 0.0008,
+      "step": 9673
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6895376194363088e-05,
+      "loss": 0.0008,
+      "step": 9674
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.689297546454122e-05,
+      "loss": 0.0008,
+      "step": 9675
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6890574734719355e-05,
+      "loss": 0.0009,
+      "step": 9676
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6888174004897494e-05,
+      "loss": 0.0008,
+      "step": 9677
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6885773275075626e-05,
+      "loss": 0.0009,
+      "step": 9678
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.688337254525376e-05,
+      "loss": 0.0008,
+      "step": 9679
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6880971815431893e-05,
+      "loss": 0.0006,
+      "step": 9680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6878571085610028e-05,
+      "loss": 0.1093,
+      "step": 9681
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.687617035578816e-05,
+      "loss": 0.0006,
+      "step": 9682
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6873769625966295e-05,
+      "loss": 0.0007,
+      "step": 9683
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6871368896144426e-05,
+      "loss": 0.0007,
+      "step": 9684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.686896816632256e-05,
+      "loss": 0.0006,
+      "step": 9685
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6866567436500693e-05,
+      "loss": 0.0007,
+      "step": 9686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6864166706678835e-05,
+      "loss": 0.0007,
+      "step": 9687
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6861765976856967e-05,
+      "loss": 0.0007,
+      "step": 9688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6859365247035102e-05,
+      "loss": 0.0006,
+      "step": 9689
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6856964517213234e-05,
+      "loss": 0.0007,
+      "step": 9690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.685456378739137e-05,
+      "loss": 0.0006,
+      "step": 9691
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.68521630575695e-05,
+      "loss": 0.0007,
+      "step": 9692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6849762327747636e-05,
+      "loss": 0.0008,
+      "step": 9693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6847361597925768e-05,
+      "loss": 0.0007,
+      "step": 9694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6844960868103903e-05,
+      "loss": 0.0007,
+      "step": 9695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6842560138282042e-05,
+      "loss": 0.0005,
+      "step": 9696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6840159408460174e-05,
+      "loss": 0.0006,
+      "step": 9697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.683775867863831e-05,
+      "loss": 0.0006,
+      "step": 9698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.683535794881644e-05,
+      "loss": 0.0005,
+      "step": 9699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6832957218994576e-05,
+      "loss": 0.0006,
+      "step": 9700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6830556489172708e-05,
+      "loss": 0.0006,
+      "step": 9701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6828155759350843e-05,
+      "loss": 0.0006,
+      "step": 9702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6825755029528975e-05,
+      "loss": 0.0006,
+      "step": 9703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.682335429970711e-05,
+      "loss": 0.0006,
+      "step": 9704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.682095356988525e-05,
+      "loss": 0.0007,
+      "step": 9705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6818552840063384e-05,
+      "loss": 0.0005,
+      "step": 9706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6816152110241515e-05,
+      "loss": 0.5273,
+      "step": 9707
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.681375138041965e-05,
+      "loss": 0.0006,
+      "step": 9708
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6811350650597782e-05,
+      "loss": 0.0006,
+      "step": 9709
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6808949920775918e-05,
+      "loss": 0.0006,
+      "step": 9710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.680654919095405e-05,
+      "loss": 0.0007,
+      "step": 9711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6804148461132185e-05,
+      "loss": 0.0006,
+      "step": 9712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6801747731310316e-05,
+      "loss": 0.0005,
+      "step": 9713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6799347001488455e-05,
+      "loss": 0.0006,
+      "step": 9714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.679694627166659e-05,
+      "loss": 0.0007,
+      "step": 9715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6794545541844722e-05,
+      "loss": 0.0006,
+      "step": 9716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6792144812022857e-05,
+      "loss": 0.0005,
+      "step": 9717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.678974408220099e-05,
+      "loss": 0.0006,
+      "step": 9718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6787343352379124e-05,
+      "loss": 0.0006,
+      "step": 9719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6784942622557256e-05,
+      "loss": 0.0005,
+      "step": 9720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.678254189273539e-05,
+      "loss": 0.0006,
+      "step": 9721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6780141162913523e-05,
+      "loss": 0.8226,
+      "step": 9722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6777740433091665e-05,
+      "loss": 0.9413,
+      "step": 9723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6775339703269797e-05,
+      "loss": 0.0005,
+      "step": 9724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6772938973447932e-05,
+      "loss": 0.0006,
+      "step": 9725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6770538243626064e-05,
+      "loss": 0.0005,
+      "step": 9726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.67681375138042e-05,
+      "loss": 0.0005,
+      "step": 9727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.676573678398233e-05,
+      "loss": 0.0006,
+      "step": 9728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6763336054160466e-05,
+      "loss": 0.0006,
+      "step": 9729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6760935324338598e-05,
+      "loss": 0.0005,
+      "step": 9730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6758534594516733e-05,
+      "loss": 0.0005,
+      "step": 9731
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.675613386469487e-05,
+      "loss": 0.0006,
+      "step": 9732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6753733134873003e-05,
+      "loss": 0.0004,
+      "step": 9733
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.675133240505114e-05,
+      "loss": 0.0006,
+      "step": 9734
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.674893167522927e-05,
+      "loss": 0.8702,
+      "step": 9735
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6746530945407405e-05,
+      "loss": 0.0006,
+      "step": 9736
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6744130215585537e-05,
+      "loss": 1.0217,
+      "step": 9737
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6741729485763672e-05,
+      "loss": 0.0006,
+      "step": 9738
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6739328755941808e-05,
+      "loss": 0.0006,
+      "step": 9739
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.673692802611994e-05,
+      "loss": 0.0005,
+      "step": 9740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6734527296298078e-05,
+      "loss": 0.0006,
+      "step": 9741
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6732126566476213e-05,
+      "loss": 0.0006,
+      "step": 9742
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6729725836654345e-05,
+      "loss": 0.0006,
+      "step": 9743
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.672732510683248e-05,
+      "loss": 0.0007,
+      "step": 9744
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6724924377010612e-05,
+      "loss": 0.0007,
+      "step": 9745
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6722523647188747e-05,
+      "loss": 0.0006,
+      "step": 9746
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.672012291736688e-05,
+      "loss": 0.0006,
+      "step": 9747
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6717722187545014e-05,
+      "loss": 0.0007,
+      "step": 9748
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6715321457723146e-05,
+      "loss": 0.0006,
+      "step": 9749
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.671292072790128e-05,
+      "loss": 0.0006,
+      "step": 9750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.671051999807942e-05,
+      "loss": 0.0007,
+      "step": 9751
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.670811926825755e-05,
+      "loss": 0.0007,
+      "step": 9752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6705718538435687e-05,
+      "loss": 0.0014,
+      "step": 9753
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6703317808613822e-05,
+      "loss": 0.7181,
+      "step": 9754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6700917078791954e-05,
+      "loss": 0.0007,
+      "step": 9755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.669851634897009e-05,
+      "loss": 0.0008,
+      "step": 9756
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.669611561914822e-05,
+      "loss": 0.0008,
+      "step": 9757
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6693714889326356e-05,
+      "loss": 0.2008,
+      "step": 9758
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6691314159504488e-05,
+      "loss": 0.0006,
+      "step": 9759
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6688913429682626e-05,
+      "loss": 0.0011,
+      "step": 9760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.668651269986076e-05,
+      "loss": 0.0012,
+      "step": 9761
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6684111970038893e-05,
+      "loss": 0.001,
+      "step": 9762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.668171124021703e-05,
+      "loss": 0.0008,
+      "step": 9763
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.667931051039516e-05,
+      "loss": 0.0006,
+      "step": 9764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6676909780573295e-05,
+      "loss": 0.001,
+      "step": 9765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6674509050751427e-05,
+      "loss": 0.0008,
+      "step": 9766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6672108320929562e-05,
+      "loss": 0.0006,
+      "step": 9767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6669707591107694e-05,
+      "loss": 0.0006,
+      "step": 9768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6667306861285836e-05,
+      "loss": 0.0011,
+      "step": 9769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6664906131463968e-05,
+      "loss": 0.0007,
+      "step": 9770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6662505401642103e-05,
+      "loss": 0.0006,
+      "step": 9771
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6660104671820235e-05,
+      "loss": 0.0006,
+      "step": 9772
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.665770394199837e-05,
+      "loss": 0.0007,
+      "step": 9773
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6655303212176502e-05,
+      "loss": 0.0005,
+      "step": 9774
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6652902482354637e-05,
+      "loss": 0.0006,
+      "step": 9775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.665050175253277e-05,
+      "loss": 0.0007,
+      "step": 9776
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6648101022710904e-05,
+      "loss": 0.0006,
+      "step": 9777
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6645700292889043e-05,
+      "loss": 0.0007,
+      "step": 9778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6643299563067175e-05,
+      "loss": 0.0063,
+      "step": 9779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.664089883324531e-05,
+      "loss": 0.0006,
+      "step": 9780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.663849810342344e-05,
+      "loss": 0.0006,
+      "step": 9781
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6636097373601577e-05,
+      "loss": 0.0007,
+      "step": 9782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.663369664377971e-05,
+      "loss": 0.0005,
+      "step": 9783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6631295913957844e-05,
+      "loss": 0.0006,
+      "step": 9784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6628895184135975e-05,
+      "loss": 0.0006,
+      "step": 9785
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.662649445431411e-05,
+      "loss": 0.0006,
+      "step": 9786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.662409372449225e-05,
+      "loss": 0.0005,
+      "step": 9787
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6621692994670384e-05,
+      "loss": 0.0005,
+      "step": 9788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6619292264848516e-05,
+      "loss": 0.0004,
+      "step": 9789
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.661689153502665e-05,
+      "loss": 0.0006,
+      "step": 9790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6614490805204783e-05,
+      "loss": 0.0006,
+      "step": 9791
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.661209007538292e-05,
+      "loss": 0.0005,
+      "step": 9792
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.660968934556105e-05,
+      "loss": 0.0006,
+      "step": 9793
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6607288615739185e-05,
+      "loss": 0.0005,
+      "step": 9794
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6604887885917317e-05,
+      "loss": 0.0006,
+      "step": 9795
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6602487156095456e-05,
+      "loss": 0.0006,
+      "step": 9796
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.660008642627359e-05,
+      "loss": 0.0006,
+      "step": 9797
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6597685696451723e-05,
+      "loss": 0.0006,
+      "step": 9798
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6595284966629858e-05,
+      "loss": 0.0005,
+      "step": 9799
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.659288423680799e-05,
+      "loss": 0.0005,
+      "step": 9800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6590483506986125e-05,
+      "loss": 0.0006,
+      "step": 9801
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6588082777164257e-05,
+      "loss": 0.0006,
+      "step": 9802
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6585682047342392e-05,
+      "loss": 0.0005,
+      "step": 9803
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6583281317520524e-05,
+      "loss": 0.0005,
+      "step": 9804
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6580880587698666e-05,
+      "loss": 0.0005,
+      "step": 9805
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6578479857876798e-05,
+      "loss": 0.0005,
+      "step": 9806
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6576079128054933e-05,
+      "loss": 0.0005,
+      "step": 9807
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6573678398233064e-05,
+      "loss": 0.0005,
+      "step": 9808
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.65712776684112e-05,
+      "loss": 0.0005,
+      "step": 9809
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.656887693858933e-05,
+      "loss": 0.0005,
+      "step": 9810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6566476208767467e-05,
+      "loss": 0.0005,
+      "step": 9811
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.65640754789456e-05,
+      "loss": 0.0005,
+      "step": 9812
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6561674749123734e-05,
+      "loss": 0.0005,
+      "step": 9813
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6559274019301865e-05,
+      "loss": 0.0005,
+      "step": 9814
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6556873289480004e-05,
+      "loss": 0.0005,
+      "step": 9815
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.655447255965814e-05,
+      "loss": 0.0024,
+      "step": 9816
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.655207182983627e-05,
+      "loss": 0.0008,
+      "step": 9817
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6549671100014406e-05,
+      "loss": 0.0005,
+      "step": 9818
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6547270370192538e-05,
+      "loss": 0.0005,
+      "step": 9819
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6544869640370673e-05,
+      "loss": 0.0009,
+      "step": 9820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6542468910548805e-05,
+      "loss": 0.0005,
+      "step": 9821
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.654006818072694e-05,
+      "loss": 0.0006,
+      "step": 9822
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6537667450905075e-05,
+      "loss": 0.0004,
+      "step": 9823
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6535266721083214e-05,
+      "loss": 0.0005,
+      "step": 9824
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6532865991261346e-05,
+      "loss": 0.9941,
+      "step": 9825
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.653046526143948e-05,
+      "loss": 0.0005,
+      "step": 9826
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6528064531617613e-05,
+      "loss": 0.0005,
+      "step": 9827
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6525663801795748e-05,
+      "loss": 0.0005,
+      "step": 9828
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.652326307197388e-05,
+      "loss": 0.0005,
+      "step": 9829
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6520862342152015e-05,
+      "loss": 0.0006,
+      "step": 9830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6518461612330147e-05,
+      "loss": 0.0005,
+      "step": 9831
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6516060882508282e-05,
+      "loss": 0.0005,
+      "step": 9832
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.651366015268642e-05,
+      "loss": 0.0005,
+      "step": 9833
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6511259422864552e-05,
+      "loss": 0.0005,
+      "step": 9834
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6508858693042687e-05,
+      "loss": 0.0005,
+      "step": 9835
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6506457963220823e-05,
+      "loss": 0.0004,
+      "step": 9836
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6504057233398954e-05,
+      "loss": 0.0005,
+      "step": 9837
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.650165650357709e-05,
+      "loss": 0.0006,
+      "step": 9838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.649925577375522e-05,
+      "loss": 0.0006,
+      "step": 9839
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6496855043933357e-05,
+      "loss": 0.0004,
+      "step": 9840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.649445431411149e-05,
+      "loss": 0.0005,
+      "step": 9841
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6492053584289627e-05,
+      "loss": 0.0005,
+      "step": 9842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6489652854467762e-05,
+      "loss": 0.0004,
+      "step": 9843
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6487252124645894e-05,
+      "loss": 0.0005,
+      "step": 9844
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.648485139482403e-05,
+      "loss": 0.0005,
+      "step": 9845
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.648245066500216e-05,
+      "loss": 0.0005,
+      "step": 9846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6480049935180296e-05,
+      "loss": 0.0005,
+      "step": 9847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6477649205358428e-05,
+      "loss": 0.0005,
+      "step": 9848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6475248475536563e-05,
+      "loss": 0.0005,
+      "step": 9849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6472847745714695e-05,
+      "loss": 0.0005,
+      "step": 9850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6470447015892837e-05,
+      "loss": 0.0004,
+      "step": 9851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.646804628607097e-05,
+      "loss": 0.0005,
+      "step": 9852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6465645556249104e-05,
+      "loss": 0.0005,
+      "step": 9853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6463244826427236e-05,
+      "loss": 0.259,
+      "step": 9854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.646084409660537e-05,
+      "loss": 0.0005,
+      "step": 9855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6458443366783503e-05,
+      "loss": 0.0005,
+      "step": 9856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6456042636961638e-05,
+      "loss": 0.0005,
+      "step": 9857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.645364190713977e-05,
+      "loss": 0.0006,
+      "step": 9858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6451241177317905e-05,
+      "loss": 0.0005,
+      "step": 9859
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6448840447496043e-05,
+      "loss": 0.0005,
+      "step": 9860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6446439717674175e-05,
+      "loss": 0.0005,
+      "step": 9861
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.644403898785231e-05,
+      "loss": 0.0004,
+      "step": 9862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6441638258030442e-05,
+      "loss": 0.0005,
+      "step": 9863
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6439237528208577e-05,
+      "loss": 0.0005,
+      "step": 9864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.643683679838671e-05,
+      "loss": 0.0004,
+      "step": 9865
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6434436068564844e-05,
+      "loss": 0.0004,
+      "step": 9866
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6432035338742976e-05,
+      "loss": 0.0004,
+      "step": 9867
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.642963460892111e-05,
+      "loss": 0.0004,
+      "step": 9868
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.642723387909925e-05,
+      "loss": 0.0005,
+      "step": 9869
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6424833149277385e-05,
+      "loss": 0.0005,
+      "step": 9870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6422432419455517e-05,
+      "loss": 0.0005,
+      "step": 9871
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6420031689633652e-05,
+      "loss": 0.0005,
+      "step": 9872
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6417630959811784e-05,
+      "loss": 0.0005,
+      "step": 9873
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.641523022998992e-05,
+      "loss": 0.0005,
+      "step": 9874
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.641282950016805e-05,
+      "loss": 0.0006,
+      "step": 9875
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6410428770346186e-05,
+      "loss": 0.0005,
+      "step": 9876
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6408028040524318e-05,
+      "loss": 0.0004,
+      "step": 9877
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6405627310702453e-05,
+      "loss": 0.0005,
+      "step": 9878
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6403226580880592e-05,
+      "loss": 0.0005,
+      "step": 9879
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6400825851058724e-05,
+      "loss": 0.0005,
+      "step": 9880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.639842512123686e-05,
+      "loss": 0.0004,
+      "step": 9881
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.639602439141499e-05,
+      "loss": 0.0004,
+      "step": 9882
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6393623661593126e-05,
+      "loss": 0.0006,
+      "step": 9883
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6391222931771257e-05,
+      "loss": 0.0004,
+      "step": 9884
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6388822201949393e-05,
+      "loss": 0.0005,
+      "step": 9885
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6386421472127524e-05,
+      "loss": 0.0005,
+      "step": 9886
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.638402074230566e-05,
+      "loss": 0.0004,
+      "step": 9887
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6381620012483798e-05,
+      "loss": 0.0005,
+      "step": 9888
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6379219282661933e-05,
+      "loss": 0.0004,
+      "step": 9889
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6376818552840065e-05,
+      "loss": 0.0004,
+      "step": 9890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.63744178230182e-05,
+      "loss": 0.0005,
+      "step": 9891
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6372017093196332e-05,
+      "loss": 0.0004,
+      "step": 9892
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6369616363374467e-05,
+      "loss": 0.0006,
+      "step": 9893
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.63672156335526e-05,
+      "loss": 0.0005,
+      "step": 9894
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6364814903730734e-05,
+      "loss": 0.0004,
+      "step": 9895
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6362414173908866e-05,
+      "loss": 0.0005,
+      "step": 9896
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6360013444087005e-05,
+      "loss": 0.0004,
+      "step": 9897
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635761271426514e-05,
+      "loss": 0.0004,
+      "step": 9898
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6355211984443272e-05,
+      "loss": 0.0003,
+      "step": 9899
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6352811254621407e-05,
+      "loss": 0.0005,
+      "step": 9900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635041052479954e-05,
+      "loss": 0.0004,
+      "step": 9901
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6348009794977674e-05,
+      "loss": 0.0004,
+      "step": 9902
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6345609065155806e-05,
+      "loss": 0.0004,
+      "step": 9903
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.634320833533394e-05,
+      "loss": 0.0004,
+      "step": 9904
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6340807605512076e-05,
+      "loss": 0.0004,
+      "step": 9905
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6338406875690215e-05,
+      "loss": 0.0004,
+      "step": 9906
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6336006145868347e-05,
+      "loss": 0.0004,
+      "step": 9907
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.633360541604648e-05,
+      "loss": 0.0004,
+      "step": 9908
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6331204686224613e-05,
+      "loss": 0.0004,
+      "step": 9909
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632880395640275e-05,
+      "loss": 0.0004,
+      "step": 9910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632640322658088e-05,
+      "loss": 0.6864,
+      "step": 9911
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6324002496759016e-05,
+      "loss": 0.0005,
+      "step": 9912
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6321601766937147e-05,
+      "loss": 0.0005,
+      "step": 9913
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6319201037115283e-05,
+      "loss": 0.0003,
+      "step": 9914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.631680030729342e-05,
+      "loss": 0.0003,
+      "step": 9915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6314399577471553e-05,
+      "loss": 0.0004,
+      "step": 9916
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6311998847649688e-05,
+      "loss": 0.0004,
+      "step": 9917
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630959811782782e-05,
+      "loss": 0.0004,
+      "step": 9918
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6307197388005955e-05,
+      "loss": 0.0005,
+      "step": 9919
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630479665818409e-05,
+      "loss": 0.0004,
+      "step": 9920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6302395928362222e-05,
+      "loss": 0.0004,
+      "step": 9921
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6299995198540357e-05,
+      "loss": 0.0003,
+      "step": 9922
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629759446871849e-05,
+      "loss": 0.0004,
+      "step": 9923
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6295193738896628e-05,
+      "loss": 0.0004,
+      "step": 9924
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6292793009074763e-05,
+      "loss": 0.0004,
+      "step": 9925
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6290392279252895e-05,
+      "loss": 0.0004,
+      "step": 9926
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628799154943103e-05,
+      "loss": 0.0004,
+      "step": 9927
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6285590819609162e-05,
+      "loss": 0.0004,
+      "step": 9928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6283190089787297e-05,
+      "loss": 0.0004,
+      "step": 9929
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628078935996543e-05,
+      "loss": 0.0004,
+      "step": 9930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6278388630143564e-05,
+      "loss": 0.0005,
+      "step": 9931
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6275987900321696e-05,
+      "loss": 0.0004,
+      "step": 9932
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6273587170499834e-05,
+      "loss": 0.0004,
+      "step": 9933
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627118644067797e-05,
+      "loss": 0.0004,
+      "step": 9934
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6268785710856105e-05,
+      "loss": 0.0003,
+      "step": 9935
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6266384981034236e-05,
+      "loss": 0.0004,
+      "step": 9936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.626398425121237e-05,
+      "loss": 0.0003,
+      "step": 9937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6261583521390503e-05,
+      "loss": 0.0004,
+      "step": 9938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.625918279156864e-05,
+      "loss": 0.0004,
+      "step": 9939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.625678206174677e-05,
+      "loss": 0.0004,
+      "step": 9940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6254381331924906e-05,
+      "loss": 0.0003,
+      "step": 9941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6251980602103037e-05,
+      "loss": 0.0004,
+      "step": 9942
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6249579872281176e-05,
+      "loss": 0.0003,
+      "step": 9943
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.624717914245931e-05,
+      "loss": 0.0004,
+      "step": 9944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6244778412637443e-05,
+      "loss": 0.0048,
+      "step": 9945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6242377682815578e-05,
+      "loss": 0.0004,
+      "step": 9946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.623997695299371e-05,
+      "loss": 0.0004,
+      "step": 9947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6237576223171845e-05,
+      "loss": 0.0004,
+      "step": 9948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6235175493349977e-05,
+      "loss": 0.0004,
+      "step": 9949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6232774763528112e-05,
+      "loss": 0.0003,
+      "step": 9950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6230374033706244e-05,
+      "loss": 0.0003,
+      "step": 9951
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6227973303884386e-05,
+      "loss": 0.0003,
+      "step": 9952
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6225572574062518e-05,
+      "loss": 0.0004,
+      "step": 9953
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6223171844240653e-05,
+      "loss": 0.0004,
+      "step": 9954
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6220771114418785e-05,
+      "loss": 0.0004,
+      "step": 9955
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.621837038459692e-05,
+      "loss": 0.0006,
+      "step": 9956
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6215969654775052e-05,
+      "loss": 0.0004,
+      "step": 9957
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6213568924953187e-05,
+      "loss": 0.0003,
+      "step": 9958
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.621116819513132e-05,
+      "loss": 0.0003,
+      "step": 9959
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6208767465309454e-05,
+      "loss": 0.0003,
+      "step": 9960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6206366735487592e-05,
+      "loss": 0.0003,
+      "step": 9961
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6203966005665724e-05,
+      "loss": 0.0004,
+      "step": 9962
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.620156527584386e-05,
+      "loss": 0.0005,
+      "step": 9963
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.619916454602199e-05,
+      "loss": 0.0004,
+      "step": 9964
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6196763816200126e-05,
+      "loss": 0.0003,
+      "step": 9965
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6194363086378258e-05,
+      "loss": 1.0581,
+      "step": 9966
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6191962356556393e-05,
+      "loss": 0.0004,
+      "step": 9967
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6189561626734525e-05,
+      "loss": 0.0003,
+      "step": 9968
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.618716089691266e-05,
+      "loss": 0.0004,
+      "step": 9969
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.61847601670908e-05,
+      "loss": 0.0004,
+      "step": 9970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6182359437268934e-05,
+      "loss": 0.0004,
+      "step": 9971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6179958707447066e-05,
+      "loss": 0.0003,
+      "step": 9972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.61775579776252e-05,
+      "loss": 0.0003,
+      "step": 9973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6175157247803333e-05,
+      "loss": 0.0004,
+      "step": 9974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6172756517981468e-05,
+      "loss": 0.0004,
+      "step": 9975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.61703557881596e-05,
+      "loss": 0.0004,
+      "step": 9976
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6167955058337735e-05,
+      "loss": 0.0003,
+      "step": 9977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6165554328515867e-05,
+      "loss": 0.0004,
+      "step": 9978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6163153598694006e-05,
+      "loss": 0.8894,
+      "step": 9979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.616075286887214e-05,
+      "loss": 0.0005,
+      "step": 9980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6158352139050273e-05,
+      "loss": 0.0005,
+      "step": 9981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6155951409228408e-05,
+      "loss": 0.0004,
+      "step": 9982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.615355067940654e-05,
+      "loss": 0.0004,
+      "step": 9983
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6151149949584675e-05,
+      "loss": 0.001,
+      "step": 9984
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6148749219762806e-05,
+      "loss": 0.0004,
+      "step": 9985
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.614634848994094e-05,
+      "loss": 0.0004,
+      "step": 9986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6143947760119077e-05,
+      "loss": 0.0004,
+      "step": 9987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6141547030297215e-05,
+      "loss": 0.0003,
+      "step": 9988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6139146300475347e-05,
+      "loss": 0.0004,
+      "step": 9989
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6136745570653482e-05,
+      "loss": 0.0005,
+      "step": 9990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6134344840831614e-05,
+      "loss": 0.0004,
+      "step": 9991
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.613194411100975e-05,
+      "loss": 0.0014,
+      "step": 9992
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.612954338118788e-05,
+      "loss": 0.525,
+      "step": 9993
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6127142651366016e-05,
+      "loss": 0.0004,
+      "step": 9994
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6124741921544148e-05,
+      "loss": 0.0005,
+      "step": 9995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6122341191722283e-05,
+      "loss": 0.4147,
+      "step": 9996
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6119940461900415e-05,
+      "loss": 0.0004,
+      "step": 9997
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6117539732078554e-05,
+      "loss": 0.0004,
+      "step": 9998
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.611513900225669e-05,
+      "loss": 0.0006,
+      "step": 9999
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.611273827243482e-05,
+      "loss": 0.4405,
+      "step": 10000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6110337542612956e-05,
+      "loss": 0.0122,
+      "step": 10001
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.610793681279109e-05,
+      "loss": 0.106,
+      "step": 10002
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6105536082969223e-05,
+      "loss": 0.0005,
+      "step": 10003
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6103135353147358e-05,
+      "loss": 0.0004,
+      "step": 10004
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.610073462332549e-05,
+      "loss": 0.0004,
+      "step": 10005
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6098333893503625e-05,
+      "loss": 0.0004,
+      "step": 10006
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6095933163681764e-05,
+      "loss": 0.0005,
+      "step": 10007
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6093532433859896e-05,
+      "loss": 0.0004,
+      "step": 10008
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.609113170403803e-05,
+      "loss": 0.0004,
+      "step": 10009
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6088730974216162e-05,
+      "loss": 0.0004,
+      "step": 10010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6086330244394298e-05,
+      "loss": 0.0004,
+      "step": 10011
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.608392951457243e-05,
+      "loss": 0.0005,
+      "step": 10012
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6081528784750565e-05,
+      "loss": 0.0024,
+      "step": 10013
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6079128054928696e-05,
+      "loss": 0.0004,
+      "step": 10014
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.607672732510683e-05,
+      "loss": 0.0004,
+      "step": 10015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.607432659528497e-05,
+      "loss": 0.0004,
+      "step": 10016
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6071925865463105e-05,
+      "loss": 0.0007,
+      "step": 10017
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6069525135641237e-05,
+      "loss": 0.0004,
+      "step": 10018
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6067124405819372e-05,
+      "loss": 0.0004,
+      "step": 10019
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6064723675997504e-05,
+      "loss": 0.0004,
+      "step": 10020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.606232294617564e-05,
+      "loss": 0.0125,
+      "step": 10021
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.605992221635377e-05,
+      "loss": 0.0004,
+      "step": 10022
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6057521486531906e-05,
+      "loss": 0.0004,
+      "step": 10023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6055120756710038e-05,
+      "loss": 0.0004,
+      "step": 10024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6052720026888177e-05,
+      "loss": 0.0004,
+      "step": 10025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6050319297066312e-05,
+      "loss": 0.0004,
+      "step": 10026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6047918567244444e-05,
+      "loss": 0.0004,
+      "step": 10027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.604551783742258e-05,
+      "loss": 0.7096,
+      "step": 10028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.604311710760071e-05,
+      "loss": 0.0004,
+      "step": 10029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6040716377778846e-05,
+      "loss": 0.0004,
+      "step": 10030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6038315647956978e-05,
+      "loss": 0.6639,
+      "step": 10031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6035914918135113e-05,
+      "loss": 0.0004,
+      "step": 10032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6033514188313245e-05,
+      "loss": 0.6726,
+      "step": 10033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6031113458491387e-05,
+      "loss": 0.0003,
+      "step": 10034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.602871272866952e-05,
+      "loss": 0.4093,
+      "step": 10035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6026311998847654e-05,
+      "loss": 0.0004,
+      "step": 10036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6023911269025785e-05,
+      "loss": 0.0004,
+      "step": 10037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.602151053920392e-05,
+      "loss": 0.0004,
+      "step": 10038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6019109809382052e-05,
+      "loss": 0.5408,
+      "step": 10039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6016709079560188e-05,
+      "loss": 0.0004,
+      "step": 10040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.601430834973832e-05,
+      "loss": 0.0004,
+      "step": 10041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6011907619916455e-05,
+      "loss": 0.0004,
+      "step": 10042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6009506890094593e-05,
+      "loss": 0.0004,
+      "step": 10043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6007106160272725e-05,
+      "loss": 0.0004,
+      "step": 10044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.600470543045086e-05,
+      "loss": 0.0004,
+      "step": 10045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6002304700628992e-05,
+      "loss": 0.0004,
+      "step": 10046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5999903970807127e-05,
+      "loss": 0.0004,
+      "step": 10047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.599750324098526e-05,
+      "loss": 0.0004,
+      "step": 10048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5995102511163394e-05,
+      "loss": 0.0205,
+      "step": 10049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5992701781341526e-05,
+      "loss": 0.0004,
+      "step": 10050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.599030105151966e-05,
+      "loss": 0.0004,
+      "step": 10051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.59879003216978e-05,
+      "loss": 0.0004,
+      "step": 10052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5985499591875935e-05,
+      "loss": 0.0004,
+      "step": 10053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5983098862054067e-05,
+      "loss": 0.0004,
+      "step": 10054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5980698132232202e-05,
+      "loss": 0.0004,
+      "step": 10055
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5978297402410334e-05,
+      "loss": 0.0004,
+      "step": 10056
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.597589667258847e-05,
+      "loss": 0.0004,
+      "step": 10057
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.59734959427666e-05,
+      "loss": 0.0004,
+      "step": 10058
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5971095212944736e-05,
+      "loss": 0.9483,
+      "step": 10059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5968694483122868e-05,
+      "loss": 0.0004,
+      "step": 10060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5966293753301003e-05,
+      "loss": 0.0004,
+      "step": 10061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.596389302347914e-05,
+      "loss": 0.0004,
+      "step": 10062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5961492293657273e-05,
+      "loss": 0.0005,
+      "step": 10063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.595909156383541e-05,
+      "loss": 0.0004,
+      "step": 10064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.595669083401354e-05,
+      "loss": 0.0004,
+      "step": 10065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5954290104191675e-05,
+      "loss": 0.0004,
+      "step": 10066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5951889374369807e-05,
+      "loss": 0.9452,
+      "step": 10067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5949488644547942e-05,
+      "loss": 0.0004,
+      "step": 10068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5947087914726074e-05,
+      "loss": 0.0005,
+      "step": 10069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.594468718490421e-05,
+      "loss": 0.0005,
+      "step": 10070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5942286455082348e-05,
+      "loss": 0.0005,
+      "step": 10071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5939885725260483e-05,
+      "loss": 0.0004,
+      "step": 10072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5937484995438615e-05,
+      "loss": 0.0004,
+      "step": 10073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.593508426561675e-05,
+      "loss": 0.0007,
+      "step": 10074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5932683535794882e-05,
+      "loss": 0.0005,
+      "step": 10075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5930282805973017e-05,
+      "loss": 0.0005,
+      "step": 10076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.592788207615115e-05,
+      "loss": 0.9278,
+      "step": 10077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5925481346329284e-05,
+      "loss": 0.0005,
+      "step": 10078
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5923080616507416e-05,
+      "loss": 0.0005,
+      "step": 10079
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5920679886685555e-05,
+      "loss": 0.0005,
+      "step": 10080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.591827915686369e-05,
+      "loss": 0.1077,
+      "step": 10081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.591587842704182e-05,
+      "loss": 0.0005,
+      "step": 10082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5913477697219957e-05,
+      "loss": 0.0009,
+      "step": 10083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5911076967398092e-05,
+      "loss": 0.4078,
+      "step": 10084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5908676237576224e-05,
+      "loss": 0.0005,
+      "step": 10085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.590627550775436e-05,
+      "loss": 0.0005,
+      "step": 10086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.590387477793249e-05,
+      "loss": 0.0007,
+      "step": 10087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5901474048110626e-05,
+      "loss": 0.0006,
+      "step": 10088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5899073318288764e-05,
+      "loss": 0.0007,
+      "step": 10089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5896672588466896e-05,
+      "loss": 0.0006,
+      "step": 10090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.589427185864503e-05,
+      "loss": 0.0006,
+      "step": 10091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5891871128823163e-05,
+      "loss": 0.001,
+      "step": 10092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.58894703990013e-05,
+      "loss": 0.0007,
+      "step": 10093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.588706966917943e-05,
+      "loss": 0.0007,
+      "step": 10094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5884668939357565e-05,
+      "loss": 0.0006,
+      "step": 10095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5882268209535697e-05,
+      "loss": 0.0007,
+      "step": 10096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5879867479713832e-05,
+      "loss": 0.0009,
+      "step": 10097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.587746674989197e-05,
+      "loss": 0.0006,
+      "step": 10098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5875066020070106e-05,
+      "loss": 0.0009,
+      "step": 10099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5872665290248238e-05,
+      "loss": 0.7625,
+      "step": 10100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5870264560426373e-05,
+      "loss": 0.0005,
+      "step": 10101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5867863830604505e-05,
+      "loss": 0.0005,
+      "step": 10102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.586546310078264e-05,
+      "loss": 0.0006,
+      "step": 10103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5863062370960772e-05,
+      "loss": 0.0007,
+      "step": 10104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5860661641138907e-05,
+      "loss": 0.0008,
+      "step": 10105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.585826091131704e-05,
+      "loss": 0.0006,
+      "step": 10106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5855860181495178e-05,
+      "loss": 0.0098,
+      "step": 10107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5853459451673313e-05,
+      "loss": 0.0007,
+      "step": 10108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5851058721851445e-05,
+      "loss": 0.0007,
+      "step": 10109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.584865799202958e-05,
+      "loss": 0.0008,
+      "step": 10110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.584625726220771e-05,
+      "loss": 0.0007,
+      "step": 10111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5843856532385847e-05,
+      "loss": 0.0006,
+      "step": 10112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.584145580256398e-05,
+      "loss": 0.0007,
+      "step": 10113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5839055072742114e-05,
+      "loss": 0.3625,
+      "step": 10114
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5836654342920245e-05,
+      "loss": 0.0007,
+      "step": 10115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5834253613098387e-05,
+      "loss": 0.0006,
+      "step": 10116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.583185288327652e-05,
+      "loss": 0.0005,
+      "step": 10117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5829452153454654e-05,
+      "loss": 0.0006,
+      "step": 10118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5827051423632786e-05,
+      "loss": 0.0008,
+      "step": 10119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.582465069381092e-05,
+      "loss": 0.0008,
+      "step": 10120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5822249963989053e-05,
+      "loss": 0.0008,
+      "step": 10121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.581984923416719e-05,
+      "loss": 0.0007,
+      "step": 10122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.581744850434532e-05,
+      "loss": 0.0006,
+      "step": 10123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5815047774523455e-05,
+      "loss": 0.0007,
+      "step": 10124
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5812647044701587e-05,
+      "loss": 0.0008,
+      "step": 10125
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5810246314879726e-05,
+      "loss": 0.0007,
+      "step": 10126
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.580784558505786e-05,
+      "loss": 0.0006,
+      "step": 10127
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5805444855235993e-05,
+      "loss": 0.0007,
+      "step": 10128
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5803044125414128e-05,
+      "loss": 0.0008,
+      "step": 10129
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.580064339559226e-05,
+      "loss": 0.0007,
+      "step": 10130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5798242665770395e-05,
+      "loss": 0.0005,
+      "step": 10131
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5795841935948527e-05,
+      "loss": 0.0006,
+      "step": 10132
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5793441206126662e-05,
+      "loss": 0.0008,
+      "step": 10133
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5791040476304794e-05,
+      "loss": 0.4507,
+      "step": 10134
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5788639746482936e-05,
+      "loss": 0.0006,
+      "step": 10135
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5786239016661067e-05,
+      "loss": 0.0006,
+      "step": 10136
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5783838286839203e-05,
+      "loss": 0.0005,
+      "step": 10137
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5781437557017334e-05,
+      "loss": 0.0005,
+      "step": 10138
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.577903682719547e-05,
+      "loss": 0.0008,
+      "step": 10139
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.57766360973736e-05,
+      "loss": 0.0007,
+      "step": 10140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5774235367551737e-05,
+      "loss": 0.0006,
+      "step": 10141
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.577183463772987e-05,
+      "loss": 0.0004,
+      "step": 10142
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5769433907908004e-05,
+      "loss": 0.0006,
+      "step": 10143
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5767033178086142e-05,
+      "loss": 0.0005,
+      "step": 10144
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5764632448264274e-05,
+      "loss": 0.0006,
+      "step": 10145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.576223171844241e-05,
+      "loss": 0.0005,
+      "step": 10146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.575983098862054e-05,
+      "loss": 0.0006,
+      "step": 10147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5757430258798676e-05,
+      "loss": 0.0005,
+      "step": 10148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5755029528976808e-05,
+      "loss": 0.0006,
+      "step": 10149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5752628799154943e-05,
+      "loss": 0.0006,
+      "step": 10150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5750228069333075e-05,
+      "loss": 0.0005,
+      "step": 10151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.574782733951121e-05,
+      "loss": 0.0004,
+      "step": 10152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.574542660968935e-05,
+      "loss": 0.0007,
+      "step": 10153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5743025879867484e-05,
+      "loss": 0.0005,
+      "step": 10154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5740625150045616e-05,
+      "loss": 0.0004,
+      "step": 10155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.573822442022375e-05,
+      "loss": 0.0005,
+      "step": 10156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5735823690401883e-05,
+      "loss": 0.0006,
+      "step": 10157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5733422960580018e-05,
+      "loss": 0.0005,
+      "step": 10158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.573102223075815e-05,
+      "loss": 0.0625,
+      "step": 10159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5728621500936285e-05,
+      "loss": 0.0006,
+      "step": 10160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5726220771114417e-05,
+      "loss": 0.0007,
+      "step": 10161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5723820041292555e-05,
+      "loss": 0.0006,
+      "step": 10162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.572141931147069e-05,
+      "loss": 0.9002,
+      "step": 10163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5719018581648822e-05,
+      "loss": 0.0006,
+      "step": 10164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5716617851826957e-05,
+      "loss": 0.0006,
+      "step": 10165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.571421712200509e-05,
+      "loss": 0.0006,
+      "step": 10166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5711816392183224e-05,
+      "loss": 0.0005,
+      "step": 10167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.570941566236136e-05,
+      "loss": 0.0006,
+      "step": 10168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.570701493253949e-05,
+      "loss": 0.0006,
+      "step": 10169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5704614202717627e-05,
+      "loss": 0.0006,
+      "step": 10170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5702213472895765e-05,
+      "loss": 0.0005,
+      "step": 10171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5699812743073897e-05,
+      "loss": 0.0006,
+      "step": 10172
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5697412013252032e-05,
+      "loss": 0.0007,
+      "step": 10173
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5695011283430164e-05,
+      "loss": 0.0006,
+      "step": 10174
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.56926105536083e-05,
+      "loss": 0.0005,
+      "step": 10175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.569020982378643e-05,
+      "loss": 0.0007,
+      "step": 10176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5687809093964566e-05,
+      "loss": 0.0008,
+      "step": 10177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5685408364142698e-05,
+      "loss": 0.0005,
+      "step": 10178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5683007634320833e-05,
+      "loss": 0.0007,
+      "step": 10179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5680606904498972e-05,
+      "loss": 0.0007,
+      "step": 10180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5678206174677104e-05,
+      "loss": 0.0006,
+      "step": 10181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.567580544485524e-05,
+      "loss": 0.8727,
+      "step": 10182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5673404715033374e-05,
+      "loss": 0.0006,
+      "step": 10183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5671003985211506e-05,
+      "loss": 0.0004,
+      "step": 10184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.566860325538964e-05,
+      "loss": 0.0007,
+      "step": 10185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5666202525567773e-05,
+      "loss": 0.0004,
+      "step": 10186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5663801795745908e-05,
+      "loss": 0.0007,
+      "step": 10187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.566140106592404e-05,
+      "loss": 0.0009,
+      "step": 10188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5659000336102175e-05,
+      "loss": 0.0007,
+      "step": 10189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5656599606280313e-05,
+      "loss": 0.0006,
+      "step": 10190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5654198876458445e-05,
+      "loss": 0.001,
+      "step": 10191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.565179814663658e-05,
+      "loss": 0.0007,
+      "step": 10192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5649397416814712e-05,
+      "loss": 0.0006,
+      "step": 10193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5646996686992847e-05,
+      "loss": 0.0094,
+      "step": 10194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.564459595717098e-05,
+      "loss": 0.0009,
+      "step": 10195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5642195227349114e-05,
+      "loss": 0.0007,
+      "step": 10196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5639794497527246e-05,
+      "loss": 0.0006,
+      "step": 10197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.563739376770538e-05,
+      "loss": 0.0008,
+      "step": 10198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.563499303788352e-05,
+      "loss": 0.0009,
+      "step": 10199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5632592308061655e-05,
+      "loss": 0.0009,
+      "step": 10200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5630191578239787e-05,
+      "loss": 0.0007,
+      "step": 10201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5627790848417922e-05,
+      "loss": 0.0007,
+      "step": 10202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5625390118596054e-05,
+      "loss": 0.0008,
+      "step": 10203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.562298938877419e-05,
+      "loss": 0.0008,
+      "step": 10204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.562058865895232e-05,
+      "loss": 0.0008,
+      "step": 10205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5618187929130456e-05,
+      "loss": 0.0019,
+      "step": 10206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5615787199308588e-05,
+      "loss": 0.0008,
+      "step": 10207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5613386469486727e-05,
+      "loss": 0.0008,
+      "step": 10208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5610985739664862e-05,
+      "loss": 0.0006,
+      "step": 10209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5608585009842994e-05,
+      "loss": 0.0007,
+      "step": 10210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.560618428002113e-05,
+      "loss": 0.0023,
+      "step": 10211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.560378355019926e-05,
+      "loss": 0.0006,
+      "step": 10212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5601382820377396e-05,
+      "loss": 0.0006,
+      "step": 10213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5598982090555527e-05,
+      "loss": 0.0008,
+      "step": 10214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5596581360733663e-05,
+      "loss": 0.0008,
+      "step": 10215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5594180630911794e-05,
+      "loss": 0.0006,
+      "step": 10216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5591779901089936e-05,
+      "loss": 0.0006,
+      "step": 10217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5589379171268068e-05,
+      "loss": 0.0007,
+      "step": 10218
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5586978441446203e-05,
+      "loss": 0.0007,
+      "step": 10219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5584577711624335e-05,
+      "loss": 0.4278,
+      "step": 10220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.558217698180247e-05,
+      "loss": 0.0007,
+      "step": 10221
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5579776251980602e-05,
+      "loss": 0.0007,
+      "step": 10222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5577375522158737e-05,
+      "loss": 1.0085,
+      "step": 10223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.557497479233687e-05,
+      "loss": 0.0004,
+      "step": 10224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5572574062515004e-05,
+      "loss": 0.0007,
+      "step": 10225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5570173332693143e-05,
+      "loss": 0.0008,
+      "step": 10226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5567772602871275e-05,
+      "loss": 0.9942,
+      "step": 10227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.556537187304941e-05,
+      "loss": 0.0007,
+      "step": 10228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5562971143227542e-05,
+      "loss": 0.0008,
+      "step": 10229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5560570413405677e-05,
+      "loss": 0.0008,
+      "step": 10230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.555816968358381e-05,
+      "loss": 0.0008,
+      "step": 10231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5555768953761944e-05,
+      "loss": 0.0007,
+      "step": 10232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5553368223940076e-05,
+      "loss": 0.0007,
+      "step": 10233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.555096749411821e-05,
+      "loss": 0.0007,
+      "step": 10234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.554856676429635e-05,
+      "loss": 0.0007,
+      "step": 10235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5546166034474485e-05,
+      "loss": 0.0008,
+      "step": 10236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5543765304652616e-05,
+      "loss": 0.0009,
+      "step": 10237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.554136457483075e-05,
+      "loss": 0.0009,
+      "step": 10238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5538963845008883e-05,
+      "loss": 0.0007,
+      "step": 10239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.553656311518702e-05,
+      "loss": 0.0008,
+      "step": 10240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.553416238536515e-05,
+      "loss": 0.0008,
+      "step": 10241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5531761655543286e-05,
+      "loss": 0.0007,
+      "step": 10242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5529360925721417e-05,
+      "loss": 0.0007,
+      "step": 10243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5526960195899556e-05,
+      "loss": 0.0007,
+      "step": 10244
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.552455946607769e-05,
+      "loss": 0.8747,
+      "step": 10245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5522158736255823e-05,
+      "loss": 0.0008,
+      "step": 10246
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5519758006433958e-05,
+      "loss": 0.0007,
+      "step": 10247
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.551735727661209e-05,
+      "loss": 0.0008,
+      "step": 10248
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5514956546790225e-05,
+      "loss": 0.001,
+      "step": 10249
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.551255581696836e-05,
+      "loss": 0.0008,
+      "step": 10250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5510155087146492e-05,
+      "loss": 0.001,
+      "step": 10251
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5507754357324627e-05,
+      "loss": 0.0009,
+      "step": 10252
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.550535362750276e-05,
+      "loss": 0.001,
+      "step": 10253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5502952897680898e-05,
+      "loss": 0.0009,
+      "step": 10254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5500552167859033e-05,
+      "loss": 0.0009,
+      "step": 10255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5498151438037165e-05,
+      "loss": 0.9288,
+      "step": 10256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.54957507082153e-05,
+      "loss": 0.0009,
+      "step": 10257
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5493349978393432e-05,
+      "loss": 0.001,
+      "step": 10258
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5490949248571567e-05,
+      "loss": 0.0009,
+      "step": 10259
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.54885485187497e-05,
+      "loss": 0.001,
+      "step": 10260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5486147788927834e-05,
+      "loss": 0.8864,
+      "step": 10261
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5483747059105966e-05,
+      "loss": 0.0012,
+      "step": 10262
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5481346329284104e-05,
+      "loss": 0.0024,
+      "step": 10263
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.547894559946224e-05,
+      "loss": 0.001,
+      "step": 10264
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5476544869640375e-05,
+      "loss": 0.0011,
+      "step": 10265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5474144139818506e-05,
+      "loss": 0.0031,
+      "step": 10266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.547174340999664e-05,
+      "loss": 0.0012,
+      "step": 10267
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5469342680174773e-05,
+      "loss": 0.804,
+      "step": 10268
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.546694195035291e-05,
+      "loss": 0.0012,
+      "step": 10269
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.546454122053104e-05,
+      "loss": 0.0013,
+      "step": 10270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5462140490709176e-05,
+      "loss": 0.0012,
+      "step": 10271
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5459739760887314e-05,
+      "loss": 0.0015,
+      "step": 10272
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5457339031065446e-05,
+      "loss": 0.0016,
+      "step": 10273
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.545493830124358e-05,
+      "loss": 0.0015,
+      "step": 10274
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5452537571421713e-05,
+      "loss": 0.0014,
+      "step": 10275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5450136841599848e-05,
+      "loss": 0.0013,
+      "step": 10276
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.544773611177798e-05,
+      "loss": 0.8123,
+      "step": 10277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5445335381956115e-05,
+      "loss": 0.0014,
+      "step": 10278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5442934652134247e-05,
+      "loss": 0.0014,
+      "step": 10279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5440533922312382e-05,
+      "loss": 0.0017,
+      "step": 10280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.543813319249052e-05,
+      "loss": 0.0015,
+      "step": 10281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5435732462668656e-05,
+      "loss": 0.0017,
+      "step": 10282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5433331732846788e-05,
+      "loss": 0.0016,
+      "step": 10283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5430931003024923e-05,
+      "loss": 0.0017,
+      "step": 10284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5428530273203055e-05,
+      "loss": 0.0017,
+      "step": 10285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.542612954338119e-05,
+      "loss": 0.0015,
+      "step": 10286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5423728813559322e-05,
+      "loss": 0.0014,
+      "step": 10287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5421328083737457e-05,
+      "loss": 0.0016,
+      "step": 10288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.541892735391559e-05,
+      "loss": 0.002,
+      "step": 10289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5416526624093727e-05,
+      "loss": 0.0019,
+      "step": 10290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5414125894271862e-05,
+      "loss": 0.0021,
+      "step": 10291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5411725164449994e-05,
+      "loss": 0.0019,
+      "step": 10292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.540932443462813e-05,
+      "loss": 0.0017,
+      "step": 10293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.540692370480626e-05,
+      "loss": 0.0015,
+      "step": 10294
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5404522974984396e-05,
+      "loss": 0.0017,
+      "step": 10295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5402122245162528e-05,
+      "loss": 0.0016,
+      "step": 10296
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5399721515340663e-05,
+      "loss": 0.0015,
+      "step": 10297
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5397320785518795e-05,
+      "loss": 0.0012,
+      "step": 10298
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5394920055696937e-05,
+      "loss": 0.0022,
+      "step": 10299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.539251932587507e-05,
+      "loss": 0.0014,
+      "step": 10300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5390118596053204e-05,
+      "loss": 0.0015,
+      "step": 10301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5387717866231336e-05,
+      "loss": 0.0016,
+      "step": 10302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.538531713640947e-05,
+      "loss": 0.0012,
+      "step": 10303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5382916406587603e-05,
+      "loss": 0.0014,
+      "step": 10304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5380515676765738e-05,
+      "loss": 0.0015,
+      "step": 10305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.537811494694387e-05,
+      "loss": 0.0015,
+      "step": 10306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5375714217122005e-05,
+      "loss": 0.0014,
+      "step": 10307
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5373313487300137e-05,
+      "loss": 0.0015,
+      "step": 10308
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5370912757478276e-05,
+      "loss": 0.0017,
+      "step": 10309
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.536851202765641e-05,
+      "loss": 0.0016,
+      "step": 10310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5366111297834543e-05,
+      "loss": 0.8119,
+      "step": 10311
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5363710568012678e-05,
+      "loss": 0.0012,
+      "step": 10312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.536130983819081e-05,
+      "loss": 0.0017,
+      "step": 10313
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5358909108368945e-05,
+      "loss": 0.0015,
+      "step": 10314
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5356508378547076e-05,
+      "loss": 0.0012,
+      "step": 10315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.535410764872521e-05,
+      "loss": 0.0011,
+      "step": 10316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5351706918903343e-05,
+      "loss": 0.8552,
+      "step": 10317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5349306189081485e-05,
+      "loss": 0.0012,
+      "step": 10318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5346905459259617e-05,
+      "loss": 0.8386,
+      "step": 10319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5344504729437752e-05,
+      "loss": 0.0012,
+      "step": 10320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5342103999615884e-05,
+      "loss": 0.0012,
+      "step": 10321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.533970326979402e-05,
+      "loss": 0.0015,
+      "step": 10322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.533730253997215e-05,
+      "loss": 0.0014,
+      "step": 10323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5334901810150286e-05,
+      "loss": 0.0016,
+      "step": 10324
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5332501080328418e-05,
+      "loss": 0.0014,
+      "step": 10325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5330100350506553e-05,
+      "loss": 0.0016,
+      "step": 10326
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5327699620684692e-05,
+      "loss": 0.0014,
+      "step": 10327
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5325298890862824e-05,
+      "loss": 0.0016,
+      "step": 10328
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.532289816104096e-05,
+      "loss": 0.0016,
+      "step": 10329
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.532049743121909e-05,
+      "loss": 0.0014,
+      "step": 10330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5318096701397226e-05,
+      "loss": 0.0013,
+      "step": 10331
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5315695971575358e-05,
+      "loss": 0.0016,
+      "step": 10332
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5313295241753493e-05,
+      "loss": 0.0018,
+      "step": 10333
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5310894511931628e-05,
+      "loss": 0.0014,
+      "step": 10334
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.530849378210976e-05,
+      "loss": 0.0014,
+      "step": 10335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.53060930522879e-05,
+      "loss": 0.0014,
+      "step": 10336
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5303692322466034e-05,
+      "loss": 0.002,
+      "step": 10337
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5301291592644165e-05,
+      "loss": 0.8797,
+      "step": 10338
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.52988908628223e-05,
+      "loss": 0.0015,
+      "step": 10339
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5296490133000432e-05,
+      "loss": 0.0014,
+      "step": 10340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5294089403178568e-05,
+      "loss": 0.0015,
+      "step": 10341
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.52916886733567e-05,
+      "loss": 0.0014,
+      "step": 10342
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5289287943534835e-05,
+      "loss": 0.0016,
+      "step": 10343
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5286887213712966e-05,
+      "loss": 0.0013,
+      "step": 10344
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5284486483891105e-05,
+      "loss": 0.0014,
+      "step": 10345
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.528208575406924e-05,
+      "loss": 0.0014,
+      "step": 10346
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5279685024247375e-05,
+      "loss": 0.0019,
+      "step": 10347
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5277284294425507e-05,
+      "loss": 0.0019,
+      "step": 10348
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5274883564603642e-05,
+      "loss": 0.0014,
+      "step": 10349
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5272482834781774e-05,
+      "loss": 0.0019,
+      "step": 10350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.527008210495991e-05,
+      "loss": 0.0015,
+      "step": 10351
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.526768137513804e-05,
+      "loss": 0.0016,
+      "step": 10352
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5265280645316176e-05,
+      "loss": 0.0013,
+      "step": 10353
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5262879915494315e-05,
+      "loss": 0.0022,
+      "step": 10354
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5260479185672447e-05,
+      "loss": 0.0015,
+      "step": 10355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5258078455850582e-05,
+      "loss": 0.0016,
+      "step": 10356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5255677726028714e-05,
+      "loss": 0.0014,
+      "step": 10357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.525327699620685e-05,
+      "loss": 0.0013,
+      "step": 10358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.525087626638498e-05,
+      "loss": 0.0014,
+      "step": 10359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5248475536563116e-05,
+      "loss": 0.0013,
+      "step": 10360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5246074806741248e-05,
+      "loss": 0.0013,
+      "step": 10361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5243674076919383e-05,
+      "loss": 0.0013,
+      "step": 10362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.524127334709752e-05,
+      "loss": 0.0015,
+      "step": 10363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5238872617275657e-05,
+      "loss": 0.0012,
+      "step": 10364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.523647188745379e-05,
+      "loss": 0.8085,
+      "step": 10365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5234071157631924e-05,
+      "loss": 0.0014,
+      "step": 10366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5231670427810055e-05,
+      "loss": 0.0013,
+      "step": 10367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.522926969798819e-05,
+      "loss": 0.0013,
+      "step": 10368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5226868968166322e-05,
+      "loss": 0.0014,
+      "step": 10369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5224468238344458e-05,
+      "loss": 0.0013,
+      "step": 10370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.522206750852259e-05,
+      "loss": 0.0011,
+      "step": 10371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5219666778700725e-05,
+      "loss": 0.0012,
+      "step": 10372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5217266048878863e-05,
+      "loss": 0.0014,
+      "step": 10373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5214865319056995e-05,
+      "loss": 0.0014,
+      "step": 10374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.521246458923513e-05,
+      "loss": 0.0013,
+      "step": 10375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5210063859413262e-05,
+      "loss": 0.0012,
+      "step": 10376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5207663129591397e-05,
+      "loss": 0.0015,
+      "step": 10377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.520526239976953e-05,
+      "loss": 0.0015,
+      "step": 10378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5202861669947664e-05,
+      "loss": 0.0013,
+      "step": 10379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5200460940125796e-05,
+      "loss": 0.0013,
+      "step": 10380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.519806021030393e-05,
+      "loss": 0.0014,
+      "step": 10381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.519565948048207e-05,
+      "loss": 0.0015,
+      "step": 10382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5193258750660205e-05,
+      "loss": 0.0014,
+      "step": 10383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5190858020838337e-05,
+      "loss": 0.0015,
+      "step": 10384
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5188457291016472e-05,
+      "loss": 0.0012,
+      "step": 10385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5186056561194604e-05,
+      "loss": 0.0011,
+      "step": 10386
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.518365583137274e-05,
+      "loss": 0.851,
+      "step": 10387
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.518125510155087e-05,
+      "loss": 0.0013,
+      "step": 10388
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5178854371729006e-05,
+      "loss": 0.0014,
+      "step": 10389
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5176453641907138e-05,
+      "loss": 0.0013,
+      "step": 10390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5174052912085276e-05,
+      "loss": 0.0016,
+      "step": 10391
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.517165218226341e-05,
+      "loss": 0.0013,
+      "step": 10392
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5169251452441543e-05,
+      "loss": 0.0013,
+      "step": 10393
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.516685072261968e-05,
+      "loss": 0.0014,
+      "step": 10394
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.516444999279781e-05,
+      "loss": 0.0013,
+      "step": 10395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5162049262975945e-05,
+      "loss": 0.0017,
+      "step": 10396
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5159648533154077e-05,
+      "loss": 0.0015,
+      "step": 10397
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5157247803332212e-05,
+      "loss": 0.0013,
+      "step": 10398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5154847073510344e-05,
+      "loss": 0.0014,
+      "step": 10399
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5152446343688486e-05,
+      "loss": 0.0012,
+      "step": 10400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5150045613866618e-05,
+      "loss": 0.0015,
+      "step": 10401
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5147644884044753e-05,
+      "loss": 0.0014,
+      "step": 10402
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5145244154222885e-05,
+      "loss": 0.0011,
+      "step": 10403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.514284342440102e-05,
+      "loss": 0.0011,
+      "step": 10404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5140442694579152e-05,
+      "loss": 0.0012,
+      "step": 10405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5138041964757287e-05,
+      "loss": 0.0014,
+      "step": 10406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.513564123493542e-05,
+      "loss": 0.0014,
+      "step": 10407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5133240505113554e-05,
+      "loss": 0.0014,
+      "step": 10408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5130839775291693e-05,
+      "loss": 0.0012,
+      "step": 10409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5128439045469825e-05,
+      "loss": 0.0013,
+      "step": 10410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.512603831564796e-05,
+      "loss": 0.0014,
+      "step": 10411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.512363758582609e-05,
+      "loss": 0.8025,
+      "step": 10412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5121236856004227e-05,
+      "loss": 0.0013,
+      "step": 10413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.511883612618236e-05,
+      "loss": 0.0012,
+      "step": 10414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5116435396360494e-05,
+      "loss": 0.0014,
+      "step": 10415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.511403466653863e-05,
+      "loss": 0.001,
+      "step": 10416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.511163393671676e-05,
+      "loss": 0.0014,
+      "step": 10417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.51092332068949e-05,
+      "loss": 0.0014,
+      "step": 10418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5106832477073034e-05,
+      "loss": 0.001,
+      "step": 10419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5104431747251166e-05,
+      "loss": 0.0013,
+      "step": 10420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.51020310174293e-05,
+      "loss": 0.0016,
+      "step": 10421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5099630287607433e-05,
+      "loss": 0.0013,
+      "step": 10422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.509722955778557e-05,
+      "loss": 0.0011,
+      "step": 10423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.50948288279637e-05,
+      "loss": 0.0014,
+      "step": 10424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5092428098141835e-05,
+      "loss": 0.756,
+      "step": 10425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5090027368319967e-05,
+      "loss": 0.0014,
+      "step": 10426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5087626638498106e-05,
+      "loss": 0.0014,
+      "step": 10427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.508522590867624e-05,
+      "loss": 0.0014,
+      "step": 10428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5082825178854373e-05,
+      "loss": 0.0015,
+      "step": 10429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5080424449032508e-05,
+      "loss": 0.0014,
+      "step": 10430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5078023719210643e-05,
+      "loss": 0.0015,
+      "step": 10431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5075622989388775e-05,
+      "loss": 0.0015,
+      "step": 10432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.507322225956691e-05,
+      "loss": 0.0016,
+      "step": 10433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5070821529745042e-05,
+      "loss": 0.0016,
+      "step": 10434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5068420799923177e-05,
+      "loss": 0.0016,
+      "step": 10435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.506602007010131e-05,
+      "loss": 0.0016,
+      "step": 10436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5063619340279448e-05,
+      "loss": 0.0016,
+      "step": 10437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5061218610457583e-05,
+      "loss": 0.0016,
+      "step": 10438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5058817880635714e-05,
+      "loss": 0.0013,
+      "step": 10439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.505641715081385e-05,
+      "loss": 0.0014,
+      "step": 10440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.505401642099198e-05,
+      "loss": 0.0013,
+      "step": 10441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5051615691170117e-05,
+      "loss": 0.0015,
+      "step": 10442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.504921496134825e-05,
+      "loss": 0.0014,
+      "step": 10443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5046814231526384e-05,
+      "loss": 0.0012,
+      "step": 10444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5044413501704515e-05,
+      "loss": 0.0014,
+      "step": 10445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5042012771882657e-05,
+      "loss": 0.0013,
+      "step": 10446
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.503961204206079e-05,
+      "loss": 0.0011,
+      "step": 10447
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5037211312238924e-05,
+      "loss": 0.0011,
+      "step": 10448
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5034810582417056e-05,
+      "loss": 0.0015,
+      "step": 10449
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.503240985259519e-05,
+      "loss": 0.0013,
+      "step": 10450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5030009122773323e-05,
+      "loss": 0.8704,
+      "step": 10451
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.502760839295146e-05,
+      "loss": 0.0016,
+      "step": 10452
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.502520766312959e-05,
+      "loss": 0.0013,
+      "step": 10453
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5022806933307725e-05,
+      "loss": 0.0013,
+      "step": 10454
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5020406203485864e-05,
+      "loss": 0.0013,
+      "step": 10455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5018005473663996e-05,
+      "loss": 0.0013,
+      "step": 10456
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.501560474384213e-05,
+      "loss": 0.0014,
+      "step": 10457
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5013204014020263e-05,
+      "loss": 0.0012,
+      "step": 10458
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5010803284198398e-05,
+      "loss": 0.001,
+      "step": 10459
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.500840255437653e-05,
+      "loss": 0.0013,
+      "step": 10460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5006001824554665e-05,
+      "loss": 0.0015,
+      "step": 10461
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5003601094732797e-05,
+      "loss": 0.0013,
+      "step": 10462
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5001200364910932e-05,
+      "loss": 0.0015,
+      "step": 10463
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4998799635089067e-05,
+      "loss": 0.0012,
+      "step": 10464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4996398905267202e-05,
+      "loss": 0.0013,
+      "step": 10465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4993998175445337e-05,
+      "loss": 0.001,
+      "step": 10466
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4991597445623473e-05,
+      "loss": 0.0011,
+      "step": 10467
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4989196715801604e-05,
+      "loss": 0.0009,
+      "step": 10468
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.498679598597974e-05,
+      "loss": 0.0012,
+      "step": 10469
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.498439525615787e-05,
+      "loss": 0.0009,
+      "step": 10470
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.498199452633601e-05,
+      "loss": 0.0011,
+      "step": 10471
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4979593796514142e-05,
+      "loss": 0.0011,
+      "step": 10472
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4977193066692277e-05,
+      "loss": 0.0012,
+      "step": 10473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.497479233687041e-05,
+      "loss": 0.001,
+      "step": 10474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4972391607048544e-05,
+      "loss": 0.0011,
+      "step": 10475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.496999087722668e-05,
+      "loss": 0.001,
+      "step": 10476
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.496759014740481e-05,
+      "loss": 0.0011,
+      "step": 10477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4965189417582946e-05,
+      "loss": 0.0011,
+      "step": 10478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4962788687761078e-05,
+      "loss": 0.001,
+      "step": 10479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4960387957939217e-05,
+      "loss": 0.001,
+      "step": 10480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.495798722811735e-05,
+      "loss": 0.001,
+      "step": 10481
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4955586498295484e-05,
+      "loss": 0.8593,
+      "step": 10482
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4953185768473615e-05,
+      "loss": 0.0011,
+      "step": 10483
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.495078503865175e-05,
+      "loss": 0.0011,
+      "step": 10484
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4948384308829886e-05,
+      "loss": 0.0009,
+      "step": 10485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.494598357900802e-05,
+      "loss": 0.0011,
+      "step": 10486
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4943582849186153e-05,
+      "loss": 0.0009,
+      "step": 10487
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4941182119364288e-05,
+      "loss": 0.0008,
+      "step": 10488
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4938781389542423e-05,
+      "loss": 0.0013,
+      "step": 10489
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4936380659720558e-05,
+      "loss": 0.001,
+      "step": 10490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.493397992989869e-05,
+      "loss": 0.0012,
+      "step": 10491
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4931579200076825e-05,
+      "loss": 0.0012,
+      "step": 10492
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4929178470254957e-05,
+      "loss": 0.001,
+      "step": 10493
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4926777740433092e-05,
+      "loss": 0.0011,
+      "step": 10494
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4924377010611227e-05,
+      "loss": 0.0009,
+      "step": 10495
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.492197628078936e-05,
+      "loss": 0.8103,
+      "step": 10496
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4919575550967494e-05,
+      "loss": 0.0012,
+      "step": 10497
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.491717482114563e-05,
+      "loss": 0.0012,
+      "step": 10498
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4914774091323765e-05,
+      "loss": 0.001,
+      "step": 10499
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4912373361501897e-05,
+      "loss": 1.0763,
+      "step": 10500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4909972631680032e-05,
+      "loss": 0.0012,
+      "step": 10501
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4907571901858164e-05,
+      "loss": 0.0018,
+      "step": 10502
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4905171172036302e-05,
+      "loss": 0.0013,
+      "step": 10503
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4902770442214434e-05,
+      "loss": 0.0015,
+      "step": 10504
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.490036971239257e-05,
+      "loss": 0.8035,
+      "step": 10505
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.48979689825707e-05,
+      "loss": 0.0015,
+      "step": 10506
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.489556825274884e-05,
+      "loss": 0.0013,
+      "step": 10507
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.489316752292697e-05,
+      "loss": 0.0019,
+      "step": 10508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4890766793105107e-05,
+      "loss": 0.7396,
+      "step": 10509
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488836606328324e-05,
+      "loss": 0.0018,
+      "step": 10510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4885965333461374e-05,
+      "loss": 0.0018,
+      "step": 10511
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488356460363951e-05,
+      "loss": 0.0022,
+      "step": 10512
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4881163873817644e-05,
+      "loss": 0.0026,
+      "step": 10513
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4878763143995776e-05,
+      "loss": 0.7054,
+      "step": 10514
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.487636241417391e-05,
+      "loss": 0.003,
+      "step": 10515
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4873961684352043e-05,
+      "loss": 0.0028,
+      "step": 10516
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4871560954530178e-05,
+      "loss": 0.0025,
+      "step": 10517
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4869160224708313e-05,
+      "loss": 0.0029,
+      "step": 10518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4866759494886445e-05,
+      "loss": 0.2384,
+      "step": 10519
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.486435876506458e-05,
+      "loss": 0.0063,
+      "step": 10520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4861958035242715e-05,
+      "loss": 0.0062,
+      "step": 10521
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.485955730542085e-05,
+      "loss": 0.0061,
+      "step": 10522
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4857156575598982e-05,
+      "loss": 0.0068,
+      "step": 10523
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4854755845777117e-05,
+      "loss": 0.0042,
+      "step": 10524
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.485235511595525e-05,
+      "loss": 0.003,
+      "step": 10525
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4849954386133388e-05,
+      "loss": 0.0036,
+      "step": 10526
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.484755365631152e-05,
+      "loss": 0.0031,
+      "step": 10527
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4845152926489655e-05,
+      "loss": 0.002,
+      "step": 10528
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4842752196667787e-05,
+      "loss": 0.7282,
+      "step": 10529
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4840351466845925e-05,
+      "loss": 0.0024,
+      "step": 10530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4837950737024057e-05,
+      "loss": 0.0023,
+      "step": 10531
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4835550007202192e-05,
+      "loss": 0.0022,
+      "step": 10532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4833149277380324e-05,
+      "loss": 0.0019,
+      "step": 10533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.483074854755846e-05,
+      "loss": 0.0025,
+      "step": 10534
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4828347817736594e-05,
+      "loss": 0.0033,
+      "step": 10535
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4825947087914726e-05,
+      "loss": 0.0015,
+      "step": 10536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.482354635809286e-05,
+      "loss": 0.0024,
+      "step": 10537
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4821145628270993e-05,
+      "loss": 0.0026,
+      "step": 10538
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.481874489844913e-05,
+      "loss": 0.0031,
+      "step": 10539
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4816344168627263e-05,
+      "loss": 0.0025,
+      "step": 10540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.48139434388054e-05,
+      "loss": 0.0028,
+      "step": 10541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.481154270898353e-05,
+      "loss": 0.0016,
+      "step": 10542
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4809141979161666e-05,
+      "loss": 0.0016,
+      "step": 10543
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.48067412493398e-05,
+      "loss": 0.0023,
+      "step": 10544
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4804340519517936e-05,
+      "loss": 0.0025,
+      "step": 10545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4801939789696068e-05,
+      "loss": 0.0023,
+      "step": 10546
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4799539059874203e-05,
+      "loss": 0.0024,
+      "step": 10547
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4797138330052335e-05,
+      "loss": 0.0014,
+      "step": 10548
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4794737600230473e-05,
+      "loss": 0.0016,
+      "step": 10549
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4792336870408605e-05,
+      "loss": 0.0018,
+      "step": 10550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.478993614058674e-05,
+      "loss": 0.0015,
+      "step": 10551
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4787535410764872e-05,
+      "loss": 0.0012,
+      "step": 10552
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4785134680943007e-05,
+      "loss": 0.002,
+      "step": 10553
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4782733951121143e-05,
+      "loss": 0.002,
+      "step": 10554
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4780333221299278e-05,
+      "loss": 0.0017,
+      "step": 10555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.477793249147741e-05,
+      "loss": 0.0016,
+      "step": 10556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4775531761655545e-05,
+      "loss": 0.0019,
+      "step": 10557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.477313103183368e-05,
+      "loss": 0.8322,
+      "step": 10558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4770730302011812e-05,
+      "loss": 0.0427,
+      "step": 10559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4768329572189947e-05,
+      "loss": 0.0011,
+      "step": 10560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.476592884236808e-05,
+      "loss": 0.0014,
+      "step": 10561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4763528112546217e-05,
+      "loss": 0.0022,
+      "step": 10562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.476112738272435e-05,
+      "loss": 0.0012,
+      "step": 10563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4758726652902484e-05,
+      "loss": 0.0012,
+      "step": 10564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4756325923080616e-05,
+      "loss": 0.0013,
+      "step": 10565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.475392519325875e-05,
+      "loss": 0.0017,
+      "step": 10566
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4751524463436886e-05,
+      "loss": 0.0012,
+      "step": 10567
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.474912373361502e-05,
+      "loss": 0.0012,
+      "step": 10568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4746723003793153e-05,
+      "loss": 0.0019,
+      "step": 10569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.474432227397129e-05,
+      "loss": 0.0014,
+      "step": 10570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4741921544149424e-05,
+      "loss": 0.0012,
+      "step": 10571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.473952081432756e-05,
+      "loss": 0.0014,
+      "step": 10572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.473712008450569e-05,
+      "loss": 0.0014,
+      "step": 10573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4734719354683826e-05,
+      "loss": 0.0016,
+      "step": 10574
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4732318624861958e-05,
+      "loss": 0.0015,
+      "step": 10575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4729917895040093e-05,
+      "loss": 0.0014,
+      "step": 10576
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4727517165218228e-05,
+      "loss": 0.0009,
+      "step": 10577
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.472511643539636e-05,
+      "loss": 0.0015,
+      "step": 10578
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4722715705574495e-05,
+      "loss": 0.0015,
+      "step": 10579
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4720314975752627e-05,
+      "loss": 0.0009,
+      "step": 10580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4717914245930766e-05,
+      "loss": 0.0011,
+      "step": 10581
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4715513516108897e-05,
+      "loss": 0.7939,
+      "step": 10582
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4713112786287033e-05,
+      "loss": 0.0011,
+      "step": 10583
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4710712056465164e-05,
+      "loss": 0.0016,
+      "step": 10584
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4708311326643303e-05,
+      "loss": 0.0017,
+      "step": 10585
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4705910596821435e-05,
+      "loss": 0.0018,
+      "step": 10586
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.470350986699957e-05,
+      "loss": 0.0015,
+      "step": 10587
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4701109137177702e-05,
+      "loss": 0.805,
+      "step": 10588
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4698708407355837e-05,
+      "loss": 0.001,
+      "step": 10589
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4696307677533972e-05,
+      "loss": 0.0015,
+      "step": 10590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4693906947712107e-05,
+      "loss": 0.001,
+      "step": 10591
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.469150621789024e-05,
+      "loss": 0.0016,
+      "step": 10592
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4689105488068374e-05,
+      "loss": 0.0014,
+      "step": 10593
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.468670475824651e-05,
+      "loss": 0.0015,
+      "step": 10594
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4684304028424645e-05,
+      "loss": 0.0015,
+      "step": 10595
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4681903298602776e-05,
+      "loss": 0.0013,
+      "step": 10596
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.467950256878091e-05,
+      "loss": 0.7513,
+      "step": 10597
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4677101838959043e-05,
+      "loss": 0.0016,
+      "step": 10598
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.467470110913718e-05,
+      "loss": 0.002,
+      "step": 10599
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4672300379315314e-05,
+      "loss": 0.002,
+      "step": 10600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4669899649493446e-05,
+      "loss": 0.7602,
+      "step": 10601
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.466749891967158e-05,
+      "loss": 0.0018,
+      "step": 10602
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4665098189849716e-05,
+      "loss": 0.0019,
+      "step": 10603
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.466269746002785e-05,
+      "loss": 0.0021,
+      "step": 10604
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4660296730205983e-05,
+      "loss": 0.0026,
+      "step": 10605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4657896000384118e-05,
+      "loss": 0.7333,
+      "step": 10606
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.465549527056225e-05,
+      "loss": 0.7058,
+      "step": 10607
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.465309454074039e-05,
+      "loss": 0.001,
+      "step": 10608
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.465069381091852e-05,
+      "loss": 0.0013,
+      "step": 10609
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4648293081096656e-05,
+      "loss": 0.0021,
+      "step": 10610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4645892351274787e-05,
+      "loss": 0.0021,
+      "step": 10611
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4643491621452923e-05,
+      "loss": 0.0036,
+      "step": 10612
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4641090891631058e-05,
+      "loss": 0.0024,
+      "step": 10613
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4638690161809193e-05,
+      "loss": 0.0032,
+      "step": 10614
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4636289431987325e-05,
+      "loss": 0.0021,
+      "step": 10615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.463388870216546e-05,
+      "loss": 0.0029,
+      "step": 10616
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4631487972343595e-05,
+      "loss": 0.0031,
+      "step": 10617
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4629087242521727e-05,
+      "loss": 0.0026,
+      "step": 10618
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4626686512699862e-05,
+      "loss": 0.0039,
+      "step": 10619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4624285782877994e-05,
+      "loss": 0.0038,
+      "step": 10620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.462188505305613e-05,
+      "loss": 0.0037,
+      "step": 10621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4619484323234264e-05,
+      "loss": 0.0032,
+      "step": 10622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.46170835934124e-05,
+      "loss": 0.0029,
+      "step": 10623
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.461468286359053e-05,
+      "loss": 0.6718,
+      "step": 10624
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4612282133768666e-05,
+      "loss": 0.0043,
+      "step": 10625
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.46098814039468e-05,
+      "loss": 0.6538,
+      "step": 10626
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4607480674124937e-05,
+      "loss": 0.0027,
+      "step": 10627
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.460507994430307e-05,
+      "loss": 0.0031,
+      "step": 10628
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4602679214481204e-05,
+      "loss": 0.0049,
+      "step": 10629
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4600278484659336e-05,
+      "loss": 0.0044,
+      "step": 10630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4597877754837474e-05,
+      "loss": 0.0039,
+      "step": 10631
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4595477025015606e-05,
+      "loss": 0.0044,
+      "step": 10632
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.459307629519374e-05,
+      "loss": 0.0039,
+      "step": 10633
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4590675565371873e-05,
+      "loss": 0.1806,
+      "step": 10634
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4588274835550008e-05,
+      "loss": 0.0054,
+      "step": 10635
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4585874105728143e-05,
+      "loss": 0.0052,
+      "step": 10636
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.458347337590628e-05,
+      "loss": 0.0022,
+      "step": 10637
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.458107264608441e-05,
+      "loss": 0.0039,
+      "step": 10638
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4578671916262546e-05,
+      "loss": 0.6322,
+      "step": 10639
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.457627118644068e-05,
+      "loss": 0.0032,
+      "step": 10640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4573870456618812e-05,
+      "loss": 0.0033,
+      "step": 10641
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4571469726796948e-05,
+      "loss": 0.0048,
+      "step": 10642
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.456906899697508e-05,
+      "loss": 0.0021,
+      "step": 10643
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4566668267153215e-05,
+      "loss": 0.0033,
+      "step": 10644
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.456426753733135e-05,
+      "loss": 0.0031,
+      "step": 10645
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4561866807509485e-05,
+      "loss": 0.0038,
+      "step": 10646
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4559466077687617e-05,
+      "loss": 0.0031,
+      "step": 10647
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4557065347865752e-05,
+      "loss": 0.6684,
+      "step": 10648
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4554664618043887e-05,
+      "loss": 0.0041,
+      "step": 10649
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4552263888222022e-05,
+      "loss": 0.0037,
+      "step": 10650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4549863158400154e-05,
+      "loss": 0.0029,
+      "step": 10651
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.454746242857829e-05,
+      "loss": 0.0042,
+      "step": 10652
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.454506169875642e-05,
+      "loss": 0.0028,
+      "step": 10653
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.454266096893456e-05,
+      "loss": 0.0028,
+      "step": 10654
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.454026023911269e-05,
+      "loss": 0.005,
+      "step": 10655
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4537859509290827e-05,
+      "loss": 0.0013,
+      "step": 10656
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.453545877946896e-05,
+      "loss": 0.0044,
+      "step": 10657
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4533058049647094e-05,
+      "loss": 0.0029,
+      "step": 10658
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.453065731982523e-05,
+      "loss": 0.0031,
+      "step": 10659
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.452825659000336e-05,
+      "loss": 0.0045,
+      "step": 10660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4525855860181496e-05,
+      "loss": 0.0031,
+      "step": 10661
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4523455130359628e-05,
+      "loss": 0.0032,
+      "step": 10662
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4521054400537766e-05,
+      "loss": 0.0023,
+      "step": 10663
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4518653670715898e-05,
+      "loss": 0.0029,
+      "step": 10664
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4516252940894033e-05,
+      "loss": 0.0036,
+      "step": 10665
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4513852211072165e-05,
+      "loss": 0.0038,
+      "step": 10666
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.45114514812503e-05,
+      "loss": 0.6626,
+      "step": 10667
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4509050751428435e-05,
+      "loss": 0.0024,
+      "step": 10668
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.450665002160657e-05,
+      "loss": 0.0031,
+      "step": 10669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4504249291784702e-05,
+      "loss": 0.0024,
+      "step": 10670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4501848561962838e-05,
+      "loss": 0.0027,
+      "step": 10671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4499447832140973e-05,
+      "loss": 0.0026,
+      "step": 10672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4497047102319108e-05,
+      "loss": 0.0041,
+      "step": 10673
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.449464637249724e-05,
+      "loss": 0.0015,
+      "step": 10674
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4492245642675375e-05,
+      "loss": 0.0017,
+      "step": 10675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4489844912853507e-05,
+      "loss": 0.002,
+      "step": 10676
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4487444183031642e-05,
+      "loss": 0.0019,
+      "step": 10677
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4485043453209777e-05,
+      "loss": 0.001,
+      "step": 10678
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4482642723387912e-05,
+      "loss": 0.0021,
+      "step": 10679
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4480241993566044e-05,
+      "loss": 0.0019,
+      "step": 10680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.447784126374418e-05,
+      "loss": 0.0018,
+      "step": 10681
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4475440533922315e-05,
+      "loss": 0.0016,
+      "step": 10682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4473039804100446e-05,
+      "loss": 0.0028,
+      "step": 10683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.447063907427858e-05,
+      "loss": 0.0023,
+      "step": 10684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4468238344456713e-05,
+      "loss": 0.0036,
+      "step": 10685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4465837614634852e-05,
+      "loss": 0.0026,
+      "step": 10686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4463436884812984e-05,
+      "loss": 0.0026,
+      "step": 10687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.446103615499112e-05,
+      "loss": 0.9019,
+      "step": 10688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.445863542516925e-05,
+      "loss": 0.0017,
+      "step": 10689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.445623469534739e-05,
+      "loss": 0.0018,
+      "step": 10690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.445383396552552e-05,
+      "loss": 0.002,
+      "step": 10691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4451433235703656e-05,
+      "loss": 0.0018,
+      "step": 10692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4449032505881788e-05,
+      "loss": 1.4786,
+      "step": 10693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4446631776059923e-05,
+      "loss": 0.0034,
+      "step": 10694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.444423104623806e-05,
+      "loss": 0.0028,
+      "step": 10695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4441830316416194e-05,
+      "loss": 0.0022,
+      "step": 10696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4439429586594325e-05,
+      "loss": 0.0015,
+      "step": 10697
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.443702885677246e-05,
+      "loss": 0.0014,
+      "step": 10698
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4434628126950592e-05,
+      "loss": 0.003,
+      "step": 10699
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4432227397128728e-05,
+      "loss": 0.002,
+      "step": 10700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4429826667306863e-05,
+      "loss": 0.0015,
+      "step": 10701
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4427425937484995e-05,
+      "loss": 0.0012,
+      "step": 10702
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.442502520766313e-05,
+      "loss": 0.0021,
+      "step": 10703
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4422624477841265e-05,
+      "loss": 0.0018,
+      "step": 10704
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.44202237480194e-05,
+      "loss": 0.0021,
+      "step": 10705
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4417823018197532e-05,
+      "loss": 0.0025,
+      "step": 10706
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4415422288375667e-05,
+      "loss": 0.0025,
+      "step": 10707
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.44130215585538e-05,
+      "loss": 0.002,
+      "step": 10708
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4410620828731938e-05,
+      "loss": 0.0015,
+      "step": 10709
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.440822009891007e-05,
+      "loss": 0.0019,
+      "step": 10710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4405819369088205e-05,
+      "loss": 0.0012,
+      "step": 10711
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4403418639266336e-05,
+      "loss": 0.0017,
+      "step": 10712
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4401017909444475e-05,
+      "loss": 0.0015,
+      "step": 10713
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4398617179622607e-05,
+      "loss": 0.0014,
+      "step": 10714
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4396216449800742e-05,
+      "loss": 0.0011,
+      "step": 10715
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4393815719978874e-05,
+      "loss": 0.0019,
+      "step": 10716
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.439141499015701e-05,
+      "loss": 0.0013,
+      "step": 10717
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4389014260335144e-05,
+      "loss": 0.0018,
+      "step": 10718
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.438661353051328e-05,
+      "loss": 0.0021,
+      "step": 10719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.438421280069141e-05,
+      "loss": 0.0017,
+      "step": 10720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4381812070869546e-05,
+      "loss": 0.0017,
+      "step": 10721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.437941134104768e-05,
+      "loss": 0.0209,
+      "step": 10722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4377010611225813e-05,
+      "loss": 0.0014,
+      "step": 10723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.437460988140395e-05,
+      "loss": 0.002,
+      "step": 10724
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.437220915158208e-05,
+      "loss": 0.0009,
+      "step": 10725
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4369808421760215e-05,
+      "loss": 0.0015,
+      "step": 10726
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.436740769193835e-05,
+      "loss": 0.0015,
+      "step": 10727
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4365006962116486e-05,
+      "loss": 0.0017,
+      "step": 10728
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4362606232294618e-05,
+      "loss": 0.0013,
+      "step": 10729
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4360205502472753e-05,
+      "loss": 0.0012,
+      "step": 10730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4357804772650885e-05,
+      "loss": 0.0011,
+      "step": 10731
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4355404042829023e-05,
+      "loss": 0.0013,
+      "step": 10732
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4353003313007155e-05,
+      "loss": 0.0011,
+      "step": 10733
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.435060258318529e-05,
+      "loss": 0.0014,
+      "step": 10734
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4348201853363422e-05,
+      "loss": 0.0013,
+      "step": 10735
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.434580112354156e-05,
+      "loss": 0.0008,
+      "step": 10736
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4343400393719692e-05,
+      "loss": 0.0012,
+      "step": 10737
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4340999663897828e-05,
+      "loss": 0.0007,
+      "step": 10738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.433859893407596e-05,
+      "loss": 0.001,
+      "step": 10739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4336198204254095e-05,
+      "loss": 0.001,
+      "step": 10740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.433379747443223e-05,
+      "loss": 0.0012,
+      "step": 10741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.433139674461036e-05,
+      "loss": 0.0006,
+      "step": 10742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4328996014788497e-05,
+      "loss": 0.0009,
+      "step": 10743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.432659528496663e-05,
+      "loss": 0.0013,
+      "step": 10744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4324194555144767e-05,
+      "loss": 0.0011,
+      "step": 10745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.43217938253229e-05,
+      "loss": 0.0013,
+      "step": 10746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4319393095501034e-05,
+      "loss": 0.0007,
+      "step": 10747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4316992365679166e-05,
+      "loss": 0.0008,
+      "step": 10748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.43145916358573e-05,
+      "loss": 0.9952,
+      "step": 10749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4312190906035436e-05,
+      "loss": 0.0013,
+      "step": 10750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.430979017621357e-05,
+      "loss": 0.0009,
+      "step": 10751
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4307389446391703e-05,
+      "loss": 0.0018,
+      "step": 10752
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.430498871656984e-05,
+      "loss": 0.0012,
+      "step": 10753
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4302587986747974e-05,
+      "loss": 0.0014,
+      "step": 10754
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.430018725692611e-05,
+      "loss": 0.0016,
+      "step": 10755
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.429778652710424e-05,
+      "loss": 0.0014,
+      "step": 10756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4295385797282376e-05,
+      "loss": 0.0017,
+      "step": 10757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4292985067460508e-05,
+      "loss": 0.0014,
+      "step": 10758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4290584337638643e-05,
+      "loss": 0.0016,
+      "step": 10759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4288183607816778e-05,
+      "loss": 0.5008,
+      "step": 10760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4285782877994913e-05,
+      "loss": 0.0013,
+      "step": 10761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4283382148173045e-05,
+      "loss": 0.0015,
+      "step": 10762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.428098141835118e-05,
+      "loss": 0.0015,
+      "step": 10763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4278580688529315e-05,
+      "loss": 0.0015,
+      "step": 10764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4276179958707447e-05,
+      "loss": 0.0015,
+      "step": 10765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4273779228885582e-05,
+      "loss": 0.0017,
+      "step": 10766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4271378499063714e-05,
+      "loss": 0.0016,
+      "step": 10767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4268977769241853e-05,
+      "loss": 0.0015,
+      "step": 10768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4266577039419984e-05,
+      "loss": 0.0014,
+      "step": 10769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.426417630959812e-05,
+      "loss": 0.0014,
+      "step": 10770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.426177557977625e-05,
+      "loss": 0.0011,
+      "step": 10771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4259374849954387e-05,
+      "loss": 0.0015,
+      "step": 10772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4256974120132522e-05,
+      "loss": 0.0014,
+      "step": 10773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4254573390310657e-05,
+      "loss": 0.0012,
+      "step": 10774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.425217266048879e-05,
+      "loss": 0.0011,
+      "step": 10775
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4249771930666924e-05,
+      "loss": 0.0013,
+      "step": 10776
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424737120084506e-05,
+      "loss": 0.0015,
+      "step": 10777
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4244970471023194e-05,
+      "loss": 0.0013,
+      "step": 10778
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4242569741201326e-05,
+      "loss": 0.0404,
+      "step": 10779
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424016901137946e-05,
+      "loss": 0.0013,
+      "step": 10780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4237768281557593e-05,
+      "loss": 0.0013,
+      "step": 10781
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.423536755173573e-05,
+      "loss": 0.001,
+      "step": 10782
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4232966821913864e-05,
+      "loss": 0.0011,
+      "step": 10783
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4230566092091995e-05,
+      "loss": 0.001,
+      "step": 10784
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.422816536227013e-05,
+      "loss": 0.001,
+      "step": 10785
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4225764632448266e-05,
+      "loss": 0.0011,
+      "step": 10786
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.42233639026264e-05,
+      "loss": 0.0014,
+      "step": 10787
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4220963172804533e-05,
+      "loss": 0.0013,
+      "step": 10788
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4218562442982668e-05,
+      "loss": 0.0011,
+      "step": 10789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.42161617131608e-05,
+      "loss": 0.001,
+      "step": 10790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4213760983338938e-05,
+      "loss": 0.0008,
+      "step": 10791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.421136025351707e-05,
+      "loss": 0.0009,
+      "step": 10792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4208959523695205e-05,
+      "loss": 0.8434,
+      "step": 10793
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4206558793873337e-05,
+      "loss": 0.001,
+      "step": 10794
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4204158064051472e-05,
+      "loss": 0.0009,
+      "step": 10795
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4201757334229607e-05,
+      "loss": 0.0011,
+      "step": 10796
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4199356604407743e-05,
+      "loss": 0.0011,
+      "step": 10797
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4196955874585874e-05,
+      "loss": 0.001,
+      "step": 10798
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.419455514476401e-05,
+      "loss": 0.0009,
+      "step": 10799
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4192154414942145e-05,
+      "loss": 0.0011,
+      "step": 10800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4189753685120277e-05,
+      "loss": 0.0011,
+      "step": 10801
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4187352955298412e-05,
+      "loss": 0.001,
+      "step": 10802
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4184952225476547e-05,
+      "loss": 0.001,
+      "step": 10803
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.418255149565468e-05,
+      "loss": 0.001,
+      "step": 10804
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4180150765832814e-05,
+      "loss": 0.001,
+      "step": 10805
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.417775003601095e-05,
+      "loss": 0.0011,
+      "step": 10806
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.417534930618908e-05,
+      "loss": 0.001,
+      "step": 10807
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4172948576367216e-05,
+      "loss": 0.0009,
+      "step": 10808
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.417054784654535e-05,
+      "loss": 0.0012,
+      "step": 10809
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4168147116723487e-05,
+      "loss": 0.0008,
+      "step": 10810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.416574638690162e-05,
+      "loss": 0.0009,
+      "step": 10811
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4163345657079754e-05,
+      "loss": 0.0012,
+      "step": 10812
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4160944927257885e-05,
+      "loss": 0.0009,
+      "step": 10813
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4158544197436024e-05,
+      "loss": 0.0009,
+      "step": 10814
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4156143467614156e-05,
+      "loss": 0.0009,
+      "step": 10815
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.415374273779229e-05,
+      "loss": 0.0008,
+      "step": 10816
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4151342007970423e-05,
+      "loss": 0.0007,
+      "step": 10817
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.414894127814856e-05,
+      "loss": 0.0008,
+      "step": 10818
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4146540548326693e-05,
+      "loss": 0.0009,
+      "step": 10819
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4144139818504828e-05,
+      "loss": 0.0009,
+      "step": 10820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.414173908868296e-05,
+      "loss": 0.9652,
+      "step": 10821
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4139338358861095e-05,
+      "loss": 0.0009,
+      "step": 10822
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.413693762903923e-05,
+      "loss": 0.001,
+      "step": 10823
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4134536899217362e-05,
+      "loss": 0.001,
+      "step": 10824
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4132136169395497e-05,
+      "loss": 0.0013,
+      "step": 10825
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.412973543957363e-05,
+      "loss": 0.0007,
+      "step": 10826
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4127334709751764e-05,
+      "loss": 0.9154,
+      "step": 10827
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.41249339799299e-05,
+      "loss": 0.001,
+      "step": 10828
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4122533250108035e-05,
+      "loss": 0.0009,
+      "step": 10829
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4120132520286167e-05,
+      "loss": 0.0013,
+      "step": 10830
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4117731790464302e-05,
+      "loss": 0.001,
+      "step": 10831
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4115331060642437e-05,
+      "loss": 0.8531,
+      "step": 10832
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4112930330820572e-05,
+      "loss": 0.0014,
+      "step": 10833
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4110529600998704e-05,
+      "loss": 0.0018,
+      "step": 10834
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410812887117684e-05,
+      "loss": 0.0012,
+      "step": 10835
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410572814135497e-05,
+      "loss": 0.0013,
+      "step": 10836
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410332741153311e-05,
+      "loss": 0.0018,
+      "step": 10837
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.410092668171124e-05,
+      "loss": 0.0016,
+      "step": 10838
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4098525951889377e-05,
+      "loss": 0.0017,
+      "step": 10839
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.409612522206751e-05,
+      "loss": 0.0016,
+      "step": 10840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4093724492245644e-05,
+      "loss": 0.0017,
+      "step": 10841
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.409132376242378e-05,
+      "loss": 0.0016,
+      "step": 10842
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.408892303260191e-05,
+      "loss": 0.0019,
+      "step": 10843
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4086522302780046e-05,
+      "loss": 0.0014,
+      "step": 10844
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.408412157295818e-05,
+      "loss": 0.0021,
+      "step": 10845
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4081720843136316e-05,
+      "loss": 0.0014,
+      "step": 10846
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4079320113314448e-05,
+      "loss": 0.002,
+      "step": 10847
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4076919383492583e-05,
+      "loss": 0.0015,
+      "step": 10848
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4074518653670715e-05,
+      "loss": 0.0016,
+      "step": 10849
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4072117923848853e-05,
+      "loss": 0.0014,
+      "step": 10850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4069717194026985e-05,
+      "loss": 0.0014,
+      "step": 10851
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.406731646420512e-05,
+      "loss": 0.0014,
+      "step": 10852
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4064915734383252e-05,
+      "loss": 0.0014,
+      "step": 10853
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4062515004561387e-05,
+      "loss": 0.0013,
+      "step": 10854
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4060114274739523e-05,
+      "loss": 0.0013,
+      "step": 10855
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4057713544917658e-05,
+      "loss": 0.0013,
+      "step": 10856
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.405531281509579e-05,
+      "loss": 0.0014,
+      "step": 10857
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4052912085273925e-05,
+      "loss": 0.0012,
+      "step": 10858
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4050511355452057e-05,
+      "loss": 0.0012,
+      "step": 10859
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4048110625630195e-05,
+      "loss": 0.0012,
+      "step": 10860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4045709895808327e-05,
+      "loss": 0.8528,
+      "step": 10861
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4043309165986462e-05,
+      "loss": 0.0012,
+      "step": 10862
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4040908436164594e-05,
+      "loss": 0.0012,
+      "step": 10863
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.403850770634273e-05,
+      "loss": 0.0014,
+      "step": 10864
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4036106976520864e-05,
+      "loss": 0.0016,
+      "step": 10865
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4033706246698996e-05,
+      "loss": 0.0015,
+      "step": 10866
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.403130551687713e-05,
+      "loss": 0.0015,
+      "step": 10867
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4028904787055263e-05,
+      "loss": 0.0016,
+      "step": 10868
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.40265040572334e-05,
+      "loss": 0.0013,
+      "step": 10869
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4024103327411533e-05,
+      "loss": 0.0012,
+      "step": 10870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.402170259758967e-05,
+      "loss": 0.0016,
+      "step": 10871
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.40193018677678e-05,
+      "loss": 0.0015,
+      "step": 10872
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.401690113794594e-05,
+      "loss": 0.8347,
+      "step": 10873
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.401450040812407e-05,
+      "loss": 0.0016,
+      "step": 10874
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4012099678302206e-05,
+      "loss": 0.0012,
+      "step": 10875
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4009698948480338e-05,
+      "loss": 0.0015,
+      "step": 10876
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4007298218658473e-05,
+      "loss": 0.0017,
+      "step": 10877
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4004897488836608e-05,
+      "loss": 0.0015,
+      "step": 10878
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4002496759014743e-05,
+      "loss": 0.0018,
+      "step": 10879
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4000096029192875e-05,
+      "loss": 0.0016,
+      "step": 10880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.399769529937101e-05,
+      "loss": 0.0017,
+      "step": 10881
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3995294569549146e-05,
+      "loss": 0.0011,
+      "step": 10882
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3992893839727277e-05,
+      "loss": 0.0014,
+      "step": 10883
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3990493109905413e-05,
+      "loss": 0.0013,
+      "step": 10884
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3988092380083548e-05,
+      "loss": 0.0015,
+      "step": 10885
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.398569165026168e-05,
+      "loss": 0.0013,
+      "step": 10886
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3983290920439815e-05,
+      "loss": 0.0014,
+      "step": 10887
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.398089019061795e-05,
+      "loss": 0.0017,
+      "step": 10888
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3978489460796082e-05,
+      "loss": 0.0018,
+      "step": 10889
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3976088730974217e-05,
+      "loss": 0.0015,
+      "step": 10890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.397368800115235e-05,
+      "loss": 0.0014,
+      "step": 10891
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3971287271330487e-05,
+      "loss": 0.0013,
+      "step": 10892
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.396888654150862e-05,
+      "loss": 0.7338,
+      "step": 10893
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3966485811686754e-05,
+      "loss": 0.0015,
+      "step": 10894
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3964085081864886e-05,
+      "loss": 0.0013,
+      "step": 10895
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3961684352043025e-05,
+      "loss": 0.0015,
+      "step": 10896
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3959283622221156e-05,
+      "loss": 0.0015,
+      "step": 10897
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.395688289239929e-05,
+      "loss": 0.0015,
+      "step": 10898
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3954482162577423e-05,
+      "loss": 0.0016,
+      "step": 10899
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.395208143275556e-05,
+      "loss": 0.0016,
+      "step": 10900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3949680702933694e-05,
+      "loss": 0.4476,
+      "step": 10901
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.394727997311183e-05,
+      "loss": 0.536,
+      "step": 10902
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.394487924328996e-05,
+      "loss": 0.0015,
+      "step": 10903
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3942478513468096e-05,
+      "loss": 0.0015,
+      "step": 10904
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.394007778364623e-05,
+      "loss": 0.0013,
+      "step": 10905
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3937677053824363e-05,
+      "loss": 0.0013,
+      "step": 10906
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3935276324002498e-05,
+      "loss": 0.0014,
+      "step": 10907
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.393287559418063e-05,
+      "loss": 0.0016,
+      "step": 10908
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3930474864358765e-05,
+      "loss": 0.8289,
+      "step": 10909
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.39280741345369e-05,
+      "loss": 0.0016,
+      "step": 10910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3925673404715036e-05,
+      "loss": 0.002,
+      "step": 10911
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3923272674893167e-05,
+      "loss": 0.0016,
+      "step": 10912
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3920871945071303e-05,
+      "loss": 0.0016,
+      "step": 10913
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3918471215249438e-05,
+      "loss": 0.0013,
+      "step": 10914
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3916070485427573e-05,
+      "loss": 0.0016,
+      "step": 10915
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3913669755605705e-05,
+      "loss": 0.0028,
+      "step": 10916
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.391126902578384e-05,
+      "loss": 0.0019,
+      "step": 10917
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3908868295961972e-05,
+      "loss": 0.0014,
+      "step": 10918
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.390646756614011e-05,
+      "loss": 0.0012,
+      "step": 10919
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3904066836318242e-05,
+      "loss": 0.0013,
+      "step": 10920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3901666106496377e-05,
+      "loss": 0.0068,
+      "step": 10921
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.389926537667451e-05,
+      "loss": 0.0015,
+      "step": 10922
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3896864646852644e-05,
+      "loss": 0.0013,
+      "step": 10923
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.389446391703078e-05,
+      "loss": 0.0012,
+      "step": 10924
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.389206318720891e-05,
+      "loss": 0.0015,
+      "step": 10925
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3889662457387046e-05,
+      "loss": 0.0012,
+      "step": 10926
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.388726172756518e-05,
+      "loss": 0.0011,
+      "step": 10927
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3884860997743317e-05,
+      "loss": 0.0012,
+      "step": 10928
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.388246026792145e-05,
+      "loss": 0.0013,
+      "step": 10929
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3880059538099584e-05,
+      "loss": 0.0011,
+      "step": 10930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3877658808277716e-05,
+      "loss": 0.0009,
+      "step": 10931
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.387525807845585e-05,
+      "loss": 0.0011,
+      "step": 10932
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3872857348633986e-05,
+      "loss": 0.0009,
+      "step": 10933
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.387045661881212e-05,
+      "loss": 0.001,
+      "step": 10934
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3868055888990253e-05,
+      "loss": 0.0011,
+      "step": 10935
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3865655159168388e-05,
+      "loss": 0.001,
+      "step": 10936
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3863254429346523e-05,
+      "loss": 0.0011,
+      "step": 10937
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.386085369952466e-05,
+      "loss": 0.0008,
+      "step": 10938
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.385845296970279e-05,
+      "loss": 0.0009,
+      "step": 10939
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3856052239880926e-05,
+      "loss": 0.0011,
+      "step": 10940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3853651510059057e-05,
+      "loss": 0.001,
+      "step": 10941
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3851250780237196e-05,
+      "loss": 0.0005,
+      "step": 10942
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3848850050415328e-05,
+      "loss": 0.8105,
+      "step": 10943
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3846449320593463e-05,
+      "loss": 0.0008,
+      "step": 10944
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3844048590771595e-05,
+      "loss": 0.0007,
+      "step": 10945
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.384164786094973e-05,
+      "loss": 0.001,
+      "step": 10946
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3839247131127865e-05,
+      "loss": 0.0009,
+      "step": 10947
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3836846401305997e-05,
+      "loss": 0.0009,
+      "step": 10948
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3834445671484132e-05,
+      "loss": 0.001,
+      "step": 10949
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3832044941662264e-05,
+      "loss": 0.0014,
+      "step": 10950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3829644211840402e-05,
+      "loss": 0.0007,
+      "step": 10951
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3827243482018534e-05,
+      "loss": 0.0008,
+      "step": 10952
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.382484275219667e-05,
+      "loss": 0.0012,
+      "step": 10953
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.38224420223748e-05,
+      "loss": 0.2201,
+      "step": 10954
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3820041292552936e-05,
+      "loss": 0.0007,
+      "step": 10955
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.381764056273107e-05,
+      "loss": 0.0008,
+      "step": 10956
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3815239832909207e-05,
+      "loss": 0.8371,
+      "step": 10957
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.381283910308734e-05,
+      "loss": 0.0007,
+      "step": 10958
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3810438373265474e-05,
+      "loss": 0.0009,
+      "step": 10959
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.380803764344361e-05,
+      "loss": 0.8579,
+      "step": 10960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3805636913621744e-05,
+      "loss": 0.0008,
+      "step": 10961
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3803236183799876e-05,
+      "loss": 0.0009,
+      "step": 10962
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.380083545397801e-05,
+      "loss": 0.0007,
+      "step": 10963
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3798434724156143e-05,
+      "loss": 0.0011,
+      "step": 10964
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3796033994334278e-05,
+      "loss": 0.0011,
+      "step": 10965
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3793633264512413e-05,
+      "loss": 0.0011,
+      "step": 10966
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3791232534690545e-05,
+      "loss": 0.8037,
+      "step": 10967
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.378883180486868e-05,
+      "loss": 0.0009,
+      "step": 10968
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3786431075046815e-05,
+      "loss": 0.0011,
+      "step": 10969
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.378403034522495e-05,
+      "loss": 0.0012,
+      "step": 10970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3781629615403082e-05,
+      "loss": 0.0013,
+      "step": 10971
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3779228885581218e-05,
+      "loss": 0.0016,
+      "step": 10972
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.377682815575935e-05,
+      "loss": 0.1332,
+      "step": 10973
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3774427425937488e-05,
+      "loss": 0.0008,
+      "step": 10974
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.377202669611562e-05,
+      "loss": 0.0017,
+      "step": 10975
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3769625966293755e-05,
+      "loss": 0.001,
+      "step": 10976
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3767225236471887e-05,
+      "loss": 0.0012,
+      "step": 10977
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3764824506650022e-05,
+      "loss": 0.0008,
+      "step": 10978
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3762423776828157e-05,
+      "loss": 0.8484,
+      "step": 10979
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3760023047006292e-05,
+      "loss": 0.0008,
+      "step": 10980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3757622317184424e-05,
+      "loss": 0.0008,
+      "step": 10981
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.375522158736256e-05,
+      "loss": 0.0013,
+      "step": 10982
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3752820857540695e-05,
+      "loss": 0.0015,
+      "step": 10983
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.375042012771883e-05,
+      "loss": 0.0009,
+      "step": 10984
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.374801939789696e-05,
+      "loss": 0.0011,
+      "step": 10985
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3745618668075097e-05,
+      "loss": 0.0011,
+      "step": 10986
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.374321793825323e-05,
+      "loss": 0.0016,
+      "step": 10987
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3740817208431364e-05,
+      "loss": 0.0017,
+      "step": 10988
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.37384164786095e-05,
+      "loss": 0.0015,
+      "step": 10989
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.373601574878763e-05,
+      "loss": 0.0016,
+      "step": 10990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3733615018965766e-05,
+      "loss": 0.0013,
+      "step": 10991
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.37312142891439e-05,
+      "loss": 0.0015,
+      "step": 10992
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3728813559322036e-05,
+      "loss": 0.1181,
+      "step": 10993
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3726412829500168e-05,
+      "loss": 0.0016,
+      "step": 10994
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3724012099678303e-05,
+      "loss": 0.0012,
+      "step": 10995
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3721611369856435e-05,
+      "loss": 0.0013,
+      "step": 10996
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3719210640034574e-05,
+      "loss": 0.0012,
+      "step": 10997
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3716809910212705e-05,
+      "loss": 0.0013,
+      "step": 10998
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.371440918039084e-05,
+      "loss": 0.0015,
+      "step": 10999
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3712008450568972e-05,
+      "loss": 0.0011,
+      "step": 11000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.370960772074711e-05,
+      "loss": 0.0011,
+      "step": 11001
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3707206990925243e-05,
+      "loss": 0.0008,
+      "step": 11002
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3704806261103378e-05,
+      "loss": 0.0011,
+      "step": 11003
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.370240553128151e-05,
+      "loss": 0.0014,
+      "step": 11004
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3700004801459645e-05,
+      "loss": 0.0009,
+      "step": 11005
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.369760407163778e-05,
+      "loss": 0.0015,
+      "step": 11006
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3695203341815912e-05,
+      "loss": 0.0008,
+      "step": 11007
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3692802611994047e-05,
+      "loss": 0.0007,
+      "step": 11008
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3690401882172182e-05,
+      "loss": 0.0013,
+      "step": 11009
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3688001152350314e-05,
+      "loss": 0.0014,
+      "step": 11010
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.368560042252845e-05,
+      "loss": 0.0012,
+      "step": 11011
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3683199692706585e-05,
+      "loss": 0.001,
+      "step": 11012
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3680798962884716e-05,
+      "loss": 0.001,
+      "step": 11013
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.367839823306285e-05,
+      "loss": 0.001,
+      "step": 11014
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3675997503240987e-05,
+      "loss": 0.0012,
+      "step": 11015
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3673596773419122e-05,
+      "loss": 0.0009,
+      "step": 11016
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3671196043597254e-05,
+      "loss": 0.0011,
+      "step": 11017
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.366879531377539e-05,
+      "loss": 0.0011,
+      "step": 11018
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.366639458395352e-05,
+      "loss": 0.0009,
+      "step": 11019
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.366399385413166e-05,
+      "loss": 0.0009,
+      "step": 11020
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.366159312430979e-05,
+      "loss": 0.0009,
+      "step": 11021
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3659192394487926e-05,
+      "loss": 0.0011,
+      "step": 11022
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3656791664666058e-05,
+      "loss": 0.0007,
+      "step": 11023
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3654390934844197e-05,
+      "loss": 0.0006,
+      "step": 11024
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.365199020502233e-05,
+      "loss": 0.001,
+      "step": 11025
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3649589475200464e-05,
+      "loss": 0.0008,
+      "step": 11026
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3647188745378595e-05,
+      "loss": 0.0009,
+      "step": 11027
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.364478801555673e-05,
+      "loss": 0.001,
+      "step": 11028
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3642387285734866e-05,
+      "loss": 0.0008,
+      "step": 11029
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3639986555912998e-05,
+      "loss": 0.0006,
+      "step": 11030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3637585826091133e-05,
+      "loss": 0.0007,
+      "step": 11031
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3635185096269265e-05,
+      "loss": 0.001,
+      "step": 11032
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3632784366447403e-05,
+      "loss": 0.001,
+      "step": 11033
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3630383636625535e-05,
+      "loss": 0.0007,
+      "step": 11034
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.362798290680367e-05,
+      "loss": 0.8944,
+      "step": 11035
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3625582176981802e-05,
+      "loss": 0.0012,
+      "step": 11036
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3623181447159937e-05,
+      "loss": 0.001,
+      "step": 11037
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3620780717338072e-05,
+      "loss": 0.0008,
+      "step": 11038
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3618379987516208e-05,
+      "loss": 0.0009,
+      "step": 11039
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.361597925769434e-05,
+      "loss": 0.001,
+      "step": 11040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3613578527872475e-05,
+      "loss": 0.0011,
+      "step": 11041
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3611177798050606e-05,
+      "loss": 0.0008,
+      "step": 11042
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3608777068228745e-05,
+      "loss": 0.0011,
+      "step": 11043
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3606376338406877e-05,
+      "loss": 0.0013,
+      "step": 11044
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3603975608585012e-05,
+      "loss": 0.0012,
+      "step": 11045
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3601574878763144e-05,
+      "loss": 0.0016,
+      "step": 11046
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.359917414894128e-05,
+      "loss": 0.0011,
+      "step": 11047
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3596773419119414e-05,
+      "loss": 0.0012,
+      "step": 11048
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3594372689297546e-05,
+      "loss": 0.0009,
+      "step": 11049
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.359197195947568e-05,
+      "loss": 0.001,
+      "step": 11050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3589571229653816e-05,
+      "loss": 0.001,
+      "step": 11051
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.358717049983195e-05,
+      "loss": 0.001,
+      "step": 11052
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3584769770010083e-05,
+      "loss": 0.001,
+      "step": 11053
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.358236904018822e-05,
+      "loss": 0.0012,
+      "step": 11054
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.357996831036635e-05,
+      "loss": 0.001,
+      "step": 11055
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.357756758054449e-05,
+      "loss": 0.0007,
+      "step": 11056
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.357516685072262e-05,
+      "loss": 0.0008,
+      "step": 11057
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3572766120900756e-05,
+      "loss": 0.001,
+      "step": 11058
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3570365391078888e-05,
+      "loss": 0.8844,
+      "step": 11059
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3567964661257023e-05,
+      "loss": 0.0009,
+      "step": 11060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3565563931435158e-05,
+      "loss": 0.0013,
+      "step": 11061
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3563163201613293e-05,
+      "loss": 0.0008,
+      "step": 11062
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3560762471791425e-05,
+      "loss": 0.001,
+      "step": 11063
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.355836174196956e-05,
+      "loss": 0.0011,
+      "step": 11064
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3555961012147695e-05,
+      "loss": 0.0008,
+      "step": 11065
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.355356028232583e-05,
+      "loss": 0.001,
+      "step": 11066
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3551159552503962e-05,
+      "loss": 0.0008,
+      "step": 11067
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3548758822682098e-05,
+      "loss": 0.0007,
+      "step": 11068
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.354635809286023e-05,
+      "loss": 0.001,
+      "step": 11069
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3543957363038364e-05,
+      "loss": 0.0007,
+      "step": 11070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.35415566332165e-05,
+      "loss": 0.0011,
+      "step": 11071
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.353915590339463e-05,
+      "loss": 0.8678,
+      "step": 11072
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3536755173572767e-05,
+      "loss": 0.0009,
+      "step": 11073
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.35343544437509e-05,
+      "loss": 0.0008,
+      "step": 11074
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3531953713929037e-05,
+      "loss": 0.0175,
+      "step": 11075
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.352955298410717e-05,
+      "loss": 0.001,
+      "step": 11076
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3527152254285304e-05,
+      "loss": 0.0008,
+      "step": 11077
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3524751524463436e-05,
+      "loss": 0.0013,
+      "step": 11078
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3522350794641574e-05,
+      "loss": 0.0008,
+      "step": 11079
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3519950064819706e-05,
+      "loss": 0.001,
+      "step": 11080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.351754933499784e-05,
+      "loss": 0.001,
+      "step": 11081
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3515148605175973e-05,
+      "loss": 0.0011,
+      "step": 11082
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.351274787535411e-05,
+      "loss": 0.001,
+      "step": 11083
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3510347145532244e-05,
+      "loss": 0.866,
+      "step": 11084
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.350794641571038e-05,
+      "loss": 0.0017,
+      "step": 11085
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.350554568588851e-05,
+      "loss": 0.0014,
+      "step": 11086
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3503144956066646e-05,
+      "loss": 0.0009,
+      "step": 11087
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.350074422624478e-05,
+      "loss": 0.0007,
+      "step": 11088
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3498343496422913e-05,
+      "loss": 0.0011,
+      "step": 11089
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3495942766601048e-05,
+      "loss": 0.0007,
+      "step": 11090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.349354203677918e-05,
+      "loss": 0.001,
+      "step": 11091
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3491141306957315e-05,
+      "loss": 0.0012,
+      "step": 11092
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.348874057713545e-05,
+      "loss": 0.0006,
+      "step": 11093
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3486339847313585e-05,
+      "loss": 0.0009,
+      "step": 11094
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3483939117491717e-05,
+      "loss": 0.0008,
+      "step": 11095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3481538387669852e-05,
+      "loss": 0.0011,
+      "step": 11096
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3479137657847987e-05,
+      "loss": 0.0011,
+      "step": 11097
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3476736928026123e-05,
+      "loss": 0.001,
+      "step": 11098
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3474336198204254e-05,
+      "loss": 0.0008,
+      "step": 11099
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.347193546838239e-05,
+      "loss": 0.0008,
+      "step": 11100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.346953473856052e-05,
+      "loss": 0.0007,
+      "step": 11101
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.346713400873866e-05,
+      "loss": 0.001,
+      "step": 11102
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3464733278916792e-05,
+      "loss": 0.0012,
+      "step": 11103
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3462332549094927e-05,
+      "loss": 0.001,
+      "step": 11104
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.345993181927306e-05,
+      "loss": 0.0007,
+      "step": 11105
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3457531089451194e-05,
+      "loss": 0.0016,
+      "step": 11106
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.345513035962933e-05,
+      "loss": 0.0009,
+      "step": 11107
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3452729629807464e-05,
+      "loss": 0.0009,
+      "step": 11108
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3450328899985596e-05,
+      "loss": 0.2904,
+      "step": 11109
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344792817016373e-05,
+      "loss": 0.001,
+      "step": 11110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3445527440341867e-05,
+      "loss": 0.0006,
+      "step": 11111
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.344312671052e-05,
+      "loss": 0.0013,
+      "step": 11112
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3440725980698134e-05,
+      "loss": 0.0007,
+      "step": 11113
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3438325250876265e-05,
+      "loss": 0.0007,
+      "step": 11114
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.34359245210544e-05,
+      "loss": 0.0011,
+      "step": 11115
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3433523791232536e-05,
+      "loss": 0.001,
+      "step": 11116
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.343112306141067e-05,
+      "loss": 0.014,
+      "step": 11117
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3428722331588803e-05,
+      "loss": 0.0011,
+      "step": 11118
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3426321601766938e-05,
+      "loss": 0.77,
+      "step": 11119
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3423920871945073e-05,
+      "loss": 0.0007,
+      "step": 11120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3421520142123208e-05,
+      "loss": 0.0008,
+      "step": 11121
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.341911941230134e-05,
+      "loss": 0.0009,
+      "step": 11122
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3416718682479475e-05,
+      "loss": 0.001,
+      "step": 11123
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3414317952657607e-05,
+      "loss": 0.0007,
+      "step": 11124
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3411917222835746e-05,
+      "loss": 0.0012,
+      "step": 11125
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3409516493013877e-05,
+      "loss": 0.0007,
+      "step": 11126
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3407115763192013e-05,
+      "loss": 0.6812,
+      "step": 11127
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3404715033370144e-05,
+      "loss": 0.0007,
+      "step": 11128
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.340231430354828e-05,
+      "loss": 0.0006,
+      "step": 11129
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3399913573726415e-05,
+      "loss": 0.0008,
+      "step": 11130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3397512843904547e-05,
+      "loss": 0.0011,
+      "step": 11131
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3395112114082682e-05,
+      "loss": 0.0012,
+      "step": 11132
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3392711384260817e-05,
+      "loss": 0.0008,
+      "step": 11133
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3390310654438952e-05,
+      "loss": 0.0007,
+      "step": 11134
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3387909924617084e-05,
+      "loss": 0.0055,
+      "step": 11135
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.338550919479522e-05,
+      "loss": 0.0009,
+      "step": 11136
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.338310846497335e-05,
+      "loss": 0.0009,
+      "step": 11137
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3380707735151486e-05,
+      "loss": 0.0023,
+      "step": 11138
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.337830700532962e-05,
+      "loss": 0.0008,
+      "step": 11139
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3375906275507757e-05,
+      "loss": 0.0118,
+      "step": 11140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.337350554568589e-05,
+      "loss": 0.0017,
+      "step": 11141
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3371104815864024e-05,
+      "loss": 0.0012,
+      "step": 11142
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.336870408604216e-05,
+      "loss": 0.939,
+      "step": 11143
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3366303356220294e-05,
+      "loss": 0.0011,
+      "step": 11144
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3363902626398426e-05,
+      "loss": 0.001,
+      "step": 11145
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.336150189657656e-05,
+      "loss": 0.0021,
+      "step": 11146
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3359101166754693e-05,
+      "loss": 0.002,
+      "step": 11147
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.335670043693283e-05,
+      "loss": 0.0023,
+      "step": 11148
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3354299707110963e-05,
+      "loss": 0.0046,
+      "step": 11149
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3351898977289098e-05,
+      "loss": 0.0025,
+      "step": 11150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.334949824746723e-05,
+      "loss": 0.0019,
+      "step": 11151
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3347097517645365e-05,
+      "loss": 0.0035,
+      "step": 11152
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.33446967878235e-05,
+      "loss": 0.0042,
+      "step": 11153
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3342296058001632e-05,
+      "loss": 0.0027,
+      "step": 11154
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3339895328179767e-05,
+      "loss": 0.0041,
+      "step": 11155
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.33374945983579e-05,
+      "loss": 0.001,
+      "step": 11156
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3335093868536038e-05,
+      "loss": 0.0009,
+      "step": 11157
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.333269313871417e-05,
+      "loss": 0.0012,
+      "step": 11158
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3330292408892305e-05,
+      "loss": 0.001,
+      "step": 11159
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3327891679070437e-05,
+      "loss": 0.5396,
+      "step": 11160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3325490949248575e-05,
+      "loss": 0.001,
+      "step": 11161
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3323090219426707e-05,
+      "loss": 0.0008,
+      "step": 11162
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3320689489604842e-05,
+      "loss": 0.0007,
+      "step": 11163
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3318288759782974e-05,
+      "loss": 0.0009,
+      "step": 11164
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.331588802996111e-05,
+      "loss": 0.0008,
+      "step": 11165
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3313487300139244e-05,
+      "loss": 0.0007,
+      "step": 11166
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.331108657031738e-05,
+      "loss": 0.0007,
+      "step": 11167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.330868584049551e-05,
+      "loss": 0.0011,
+      "step": 11168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3306285110673647e-05,
+      "loss": 0.0011,
+      "step": 11169
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.330388438085178e-05,
+      "loss": 0.0009,
+      "step": 11170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3301483651029913e-05,
+      "loss": 0.0007,
+      "step": 11171
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.329908292120805e-05,
+      "loss": 0.0026,
+      "step": 11172
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.329668219138618e-05,
+      "loss": 0.0007,
+      "step": 11173
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3294281461564316e-05,
+      "loss": 0.0007,
+      "step": 11174
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.329188073174245e-05,
+      "loss": 0.0022,
+      "step": 11175
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3289480001920586e-05,
+      "loss": 0.8874,
+      "step": 11176
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3287079272098718e-05,
+      "loss": 0.0008,
+      "step": 11177
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3284678542276853e-05,
+      "loss": 0.001,
+      "step": 11178
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3282277812454985e-05,
+      "loss": 0.0006,
+      "step": 11179
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3279877082633123e-05,
+      "loss": 0.0007,
+      "step": 11180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3277476352811255e-05,
+      "loss": 0.0018,
+      "step": 11181
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.327507562298939e-05,
+      "loss": 0.0008,
+      "step": 11182
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3272674893167522e-05,
+      "loss": 0.0006,
+      "step": 11183
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.327027416334566e-05,
+      "loss": 0.0027,
+      "step": 11184
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3267873433523793e-05,
+      "loss": 0.0007,
+      "step": 11185
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3265472703701928e-05,
+      "loss": 0.0007,
+      "step": 11186
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.326307197388006e-05,
+      "loss": 0.0009,
+      "step": 11187
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3260671244058195e-05,
+      "loss": 0.001,
+      "step": 11188
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.325827051423633e-05,
+      "loss": 0.0011,
+      "step": 11189
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3255869784414465e-05,
+      "loss": 0.0009,
+      "step": 11190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3253469054592597e-05,
+      "loss": 0.0006,
+      "step": 11191
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3251068324770732e-05,
+      "loss": 0.0011,
+      "step": 11192
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3248667594948867e-05,
+      "loss": 0.0019,
+      "step": 11193
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3246266865127e-05,
+      "loss": 0.0011,
+      "step": 11194
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3243866135305134e-05,
+      "loss": 0.0005,
+      "step": 11195
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3241465405483266e-05,
+      "loss": 0.0006,
+      "step": 11196
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.32390646756614e-05,
+      "loss": 0.0007,
+      "step": 11197
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3236663945839536e-05,
+      "loss": 0.0007,
+      "step": 11198
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.323426321601767e-05,
+      "loss": 0.0008,
+      "step": 11199
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3231862486195803e-05,
+      "loss": 0.0015,
+      "step": 11200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.322946175637394e-05,
+      "loss": 0.0005,
+      "step": 11201
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.322706102655207e-05,
+      "loss": 0.8565,
+      "step": 11202
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.322466029673021e-05,
+      "loss": 0.0004,
+      "step": 11203
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.322225956690834e-05,
+      "loss": 0.002,
+      "step": 11204
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3219858837086476e-05,
+      "loss": 0.0008,
+      "step": 11205
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3217458107264608e-05,
+      "loss": 0.0008,
+      "step": 11206
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3215057377442746e-05,
+      "loss": 0.0013,
+      "step": 11207
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3212656647620878e-05,
+      "loss": 0.0007,
+      "step": 11208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3210255917799013e-05,
+      "loss": 0.0009,
+      "step": 11209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3207855187977145e-05,
+      "loss": 0.0007,
+      "step": 11210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.320545445815528e-05,
+      "loss": 0.0007,
+      "step": 11211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3203053728333416e-05,
+      "loss": 0.0008,
+      "step": 11212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3200652998511547e-05,
+      "loss": 0.0007,
+      "step": 11213
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3198252268689683e-05,
+      "loss": 0.0012,
+      "step": 11214
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3195851538867814e-05,
+      "loss": 0.0006,
+      "step": 11215
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3193450809045953e-05,
+      "loss": 0.0006,
+      "step": 11216
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3191050079224085e-05,
+      "loss": 0.0006,
+      "step": 11217
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.318864934940222e-05,
+      "loss": 0.0007,
+      "step": 11218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3186248619580352e-05,
+      "loss": 0.4011,
+      "step": 11219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3183847889758487e-05,
+      "loss": 0.0005,
+      "step": 11220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3181447159936622e-05,
+      "loss": 0.0011,
+      "step": 11221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3179046430114757e-05,
+      "loss": 0.5021,
+      "step": 11222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.317664570029289e-05,
+      "loss": 0.0007,
+      "step": 11223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3174244970471024e-05,
+      "loss": 0.0007,
+      "step": 11224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.317184424064916e-05,
+      "loss": 0.0005,
+      "step": 11225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3169443510827295e-05,
+      "loss": 0.0009,
+      "step": 11226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3167042781005426e-05,
+      "loss": 0.0006,
+      "step": 11227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.316464205118356e-05,
+      "loss": 0.0007,
+      "step": 11228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3162241321361693e-05,
+      "loss": 0.0007,
+      "step": 11229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3159840591539832e-05,
+      "loss": 0.0006,
+      "step": 11230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3157439861717964e-05,
+      "loss": 0.0009,
+      "step": 11231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.31550391318961e-05,
+      "loss": 0.0009,
+      "step": 11232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.315263840207423e-05,
+      "loss": 0.0005,
+      "step": 11233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3150237672252366e-05,
+      "loss": 0.0008,
+      "step": 11234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.31478369424305e-05,
+      "loss": 0.0008,
+      "step": 11235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3145436212608633e-05,
+      "loss": 0.0005,
+      "step": 11236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3143035482786768e-05,
+      "loss": 0.0007,
+      "step": 11237
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.31406347529649e-05,
+      "loss": 0.0006,
+      "step": 11238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.313823402314304e-05,
+      "loss": 0.0009,
+      "step": 11239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.313583329332117e-05,
+      "loss": 0.0009,
+      "step": 11240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3133432563499306e-05,
+      "loss": 0.0011,
+      "step": 11241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3131031833677437e-05,
+      "loss": 0.0007,
+      "step": 11242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3128631103855573e-05,
+      "loss": 0.0008,
+      "step": 11243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3126230374033708e-05,
+      "loss": 0.0013,
+      "step": 11244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3123829644211843e-05,
+      "loss": 0.0007,
+      "step": 11245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3121428914389975e-05,
+      "loss": 0.0008,
+      "step": 11246
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.311902818456811e-05,
+      "loss": 1.7717,
+      "step": 11247
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3116627454746245e-05,
+      "loss": 0.0007,
+      "step": 11248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.311422672492438e-05,
+      "loss": 0.0008,
+      "step": 11249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3111825995102512e-05,
+      "loss": 0.009,
+      "step": 11250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3109425265280647e-05,
+      "loss": 0.0008,
+      "step": 11251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.310702453545878e-05,
+      "loss": 0.0008,
+      "step": 11252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3104623805636914e-05,
+      "loss": 0.0006,
+      "step": 11253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.310222307581505e-05,
+      "loss": 0.0008,
+      "step": 11254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.309982234599318e-05,
+      "loss": 0.0009,
+      "step": 11255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3097421616171316e-05,
+      "loss": 0.0006,
+      "step": 11256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.309502088634945e-05,
+      "loss": 0.0008,
+      "step": 11257
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3092620156527587e-05,
+      "loss": 0.0003,
+      "step": 11258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.309021942670572e-05,
+      "loss": 0.0005,
+      "step": 11259
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3087818696883854e-05,
+      "loss": 0.0011,
+      "step": 11260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3085417967061986e-05,
+      "loss": 0.001,
+      "step": 11261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3083017237240124e-05,
+      "loss": 0.9475,
+      "step": 11262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3080616507418256e-05,
+      "loss": 0.585,
+      "step": 11263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.307821577759639e-05,
+      "loss": 0.4391,
+      "step": 11264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3075815047774523e-05,
+      "loss": 0.0012,
+      "step": 11265
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3073414317952658e-05,
+      "loss": 0.0897,
+      "step": 11266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3071013588130793e-05,
+      "loss": 0.001,
+      "step": 11267
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.306861285830893e-05,
+      "loss": 0.0008,
+      "step": 11268
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.306621212848706e-05,
+      "loss": 0.0009,
+      "step": 11269
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3063811398665196e-05,
+      "loss": 0.001,
+      "step": 11270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.306141066884333e-05,
+      "loss": 0.0011,
+      "step": 11271
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3059009939021466e-05,
+      "loss": 0.0009,
+      "step": 11272
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3056609209199598e-05,
+      "loss": 0.001,
+      "step": 11273
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3054208479377733e-05,
+      "loss": 0.0011,
+      "step": 11274
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3051807749555865e-05,
+      "loss": 0.0009,
+      "step": 11275
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3049407019734e-05,
+      "loss": 0.0011,
+      "step": 11276
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3047006289912135e-05,
+      "loss": 0.0011,
+      "step": 11277
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3044605560090267e-05,
+      "loss": 0.001,
+      "step": 11278
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3042204830268402e-05,
+      "loss": 0.0011,
+      "step": 11279
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3039804100446537e-05,
+      "loss": 0.0009,
+      "step": 11280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3037403370624672e-05,
+      "loss": 0.0007,
+      "step": 11281
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3035002640802804e-05,
+      "loss": 0.0022,
+      "step": 11282
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.303260191098094e-05,
+      "loss": 0.0011,
+      "step": 11283
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.303020118115907e-05,
+      "loss": 0.0018,
+      "step": 11284
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.302780045133721e-05,
+      "loss": 0.001,
+      "step": 11285
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.302539972151534e-05,
+      "loss": 0.1481,
+      "step": 11286
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3022998991693477e-05,
+      "loss": 0.0011,
+      "step": 11287
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.302059826187161e-05,
+      "loss": 0.001,
+      "step": 11288
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3018197532049747e-05,
+      "loss": 0.0013,
+      "step": 11289
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.301579680222788e-05,
+      "loss": 0.001,
+      "step": 11290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3013396072406014e-05,
+      "loss": 0.0015,
+      "step": 11291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3010995342584146e-05,
+      "loss": 0.8025,
+      "step": 11292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.300859461276228e-05,
+      "loss": 0.7813,
+      "step": 11293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3006193882940416e-05,
+      "loss": 0.0013,
+      "step": 11294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3003793153118548e-05,
+      "loss": 0.0017,
+      "step": 11295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3001392423296683e-05,
+      "loss": 0.0015,
+      "step": 11296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2998991693474815e-05,
+      "loss": 0.0013,
+      "step": 11297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.299659096365295e-05,
+      "loss": 0.0013,
+      "step": 11298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2994190233831085e-05,
+      "loss": 0.0019,
+      "step": 11299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.299178950400922e-05,
+      "loss": 0.8131,
+      "step": 11300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2989388774187352e-05,
+      "loss": 0.0013,
+      "step": 11301
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2986988044365488e-05,
+      "loss": 0.0019,
+      "step": 11302
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2984587314543623e-05,
+      "loss": 0.0014,
+      "step": 11303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2982186584721758e-05,
+      "loss": 0.0025,
+      "step": 11304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.297978585489989e-05,
+      "loss": 0.0018,
+      "step": 11305
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2977385125078025e-05,
+      "loss": 0.0015,
+      "step": 11306
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2974984395256157e-05,
+      "loss": 0.1103,
+      "step": 11307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2972583665434295e-05,
+      "loss": 0.0017,
+      "step": 11308
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2970182935612427e-05,
+      "loss": 0.0018,
+      "step": 11309
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2967782205790562e-05,
+      "loss": 0.1429,
+      "step": 11310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2965381475968694e-05,
+      "loss": 0.0016,
+      "step": 11311
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.296298074614683e-05,
+      "loss": 0.0018,
+      "step": 11312
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2960580016324965e-05,
+      "loss": 0.0034,
+      "step": 11313
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.29581792865031e-05,
+      "loss": 0.0019,
+      "step": 11314
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.295577855668123e-05,
+      "loss": 0.0019,
+      "step": 11315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2953377826859367e-05,
+      "loss": 0.8119,
+      "step": 11316
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2950977097037502e-05,
+      "loss": 0.0013,
+      "step": 11317
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2948576367215634e-05,
+      "loss": 0.7274,
+      "step": 11318
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.294617563739377e-05,
+      "loss": 0.0014,
+      "step": 11319
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.29437749075719e-05,
+      "loss": 0.0012,
+      "step": 11320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2941374177750036e-05,
+      "loss": 0.0018,
+      "step": 11321
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.293897344792817e-05,
+      "loss": 0.0016,
+      "step": 11322
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2936572718106306e-05,
+      "loss": 0.0019,
+      "step": 11323
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2934171988284438e-05,
+      "loss": 0.002,
+      "step": 11324
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2931771258462573e-05,
+      "loss": 0.0016,
+      "step": 11325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.292937052864071e-05,
+      "loss": 0.002,
+      "step": 11326
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2926969798818844e-05,
+      "loss": 0.0017,
+      "step": 11327
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2924569068996975e-05,
+      "loss": 0.7688,
+      "step": 11328
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.292216833917511e-05,
+      "loss": 0.0009,
+      "step": 11329
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2919767609353242e-05,
+      "loss": 0.0019,
+      "step": 11330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.291736687953138e-05,
+      "loss": 0.7612,
+      "step": 11331
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2914966149709513e-05,
+      "loss": 0.0027,
+      "step": 11332
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2912565419887648e-05,
+      "loss": 0.0029,
+      "step": 11333
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.291016469006578e-05,
+      "loss": 0.0017,
+      "step": 11334
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2907763960243915e-05,
+      "loss": 0.003,
+      "step": 11335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.290536323042205e-05,
+      "loss": 0.0019,
+      "step": 11336
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2902962500600182e-05,
+      "loss": 0.0017,
+      "step": 11337
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2900561770778317e-05,
+      "loss": 0.0034,
+      "step": 11338
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.289816104095645e-05,
+      "loss": 0.0034,
+      "step": 11339
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2895760311134588e-05,
+      "loss": 0.003,
+      "step": 11340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.289335958131272e-05,
+      "loss": 0.0027,
+      "step": 11341
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2890958851490855e-05,
+      "loss": 0.0025,
+      "step": 11342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2888558121668986e-05,
+      "loss": 0.0021,
+      "step": 11343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2886157391847125e-05,
+      "loss": 0.0022,
+      "step": 11344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2883756662025257e-05,
+      "loss": 0.0022,
+      "step": 11345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2881355932203392e-05,
+      "loss": 0.6986,
+      "step": 11346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2878955202381524e-05,
+      "loss": 0.0022,
+      "step": 11347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.287655447255966e-05,
+      "loss": 0.7123,
+      "step": 11348
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2874153742737794e-05,
+      "loss": 0.0035,
+      "step": 11349
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.287175301291593e-05,
+      "loss": 0.0034,
+      "step": 11350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.286935228309406e-05,
+      "loss": 0.0024,
+      "step": 11351
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2866951553272196e-05,
+      "loss": 0.0038,
+      "step": 11352
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2864550823450328e-05,
+      "loss": 0.002,
+      "step": 11353
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2862150093628463e-05,
+      "loss": 0.0027,
+      "step": 11354
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.28597493638066e-05,
+      "loss": 0.0029,
+      "step": 11355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2857348633984734e-05,
+      "loss": 0.0025,
+      "step": 11356
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2854947904162865e-05,
+      "loss": 0.0036,
+      "step": 11357
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2852547174341e-05,
+      "loss": 0.0021,
+      "step": 11358
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2850146444519136e-05,
+      "loss": 0.0048,
+      "step": 11359
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2847745714697268e-05,
+      "loss": 0.7397,
+      "step": 11360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2845344984875403e-05,
+      "loss": 0.0026,
+      "step": 11361
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2842944255053535e-05,
+      "loss": 0.0014,
+      "step": 11362
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2840543525231673e-05,
+      "loss": 0.0024,
+      "step": 11363
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2838142795409805e-05,
+      "loss": 0.0026,
+      "step": 11364
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.283574206558794e-05,
+      "loss": 0.0027,
+      "step": 11365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2833341335766072e-05,
+      "loss": 0.0022,
+      "step": 11366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.283094060594421e-05,
+      "loss": 0.6642,
+      "step": 11367
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2828539876122342e-05,
+      "loss": 0.0025,
+      "step": 11368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2826139146300478e-05,
+      "loss": 0.2332,
+      "step": 11369
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.282373841647861e-05,
+      "loss": 0.0021,
+      "step": 11370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2821337686656745e-05,
+      "loss": 0.0015,
+      "step": 11371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.281893695683488e-05,
+      "loss": 0.002,
+      "step": 11372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2816536227013015e-05,
+      "loss": 0.0035,
+      "step": 11373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2814135497191147e-05,
+      "loss": 0.002,
+      "step": 11374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2811734767369282e-05,
+      "loss": 0.0013,
+      "step": 11375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2809334037547417e-05,
+      "loss": 0.0027,
+      "step": 11376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.280693330772555e-05,
+      "loss": 0.0029,
+      "step": 11377
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2804532577903684e-05,
+      "loss": 0.0025,
+      "step": 11378
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2802131848081816e-05,
+      "loss": 0.0008,
+      "step": 11379
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.279973111825995e-05,
+      "loss": 0.2131,
+      "step": 11380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2797330388438086e-05,
+      "loss": 0.0032,
+      "step": 11381
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.279492965861622e-05,
+      "loss": 0.0016,
+      "step": 11382
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2792528928794353e-05,
+      "loss": 0.1773,
+      "step": 11383
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.279012819897249e-05,
+      "loss": 0.0019,
+      "step": 11384
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.278772746915062e-05,
+      "loss": 0.0031,
+      "step": 11385
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.278532673932876e-05,
+      "loss": 0.0011,
+      "step": 11386
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.278292600950689e-05,
+      "loss": 0.0044,
+      "step": 11387
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2780525279685026e-05,
+      "loss": 0.0024,
+      "step": 11388
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2778124549863158e-05,
+      "loss": 0.0015,
+      "step": 11389
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2775723820041296e-05,
+      "loss": 0.0026,
+      "step": 11390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2773323090219428e-05,
+      "loss": 0.0015,
+      "step": 11391
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2770922360397563e-05,
+      "loss": 0.0012,
+      "step": 11392
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2768521630575695e-05,
+      "loss": 0.002,
+      "step": 11393
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.276612090075383e-05,
+      "loss": 0.0017,
+      "step": 11394
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2763720170931965e-05,
+      "loss": 0.0015,
+      "step": 11395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.27613194411101e-05,
+      "loss": 0.0021,
+      "step": 11396
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2758918711288232e-05,
+      "loss": 0.0022,
+      "step": 11397
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2756517981466367e-05,
+      "loss": 0.0021,
+      "step": 11398
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2754117251644503e-05,
+      "loss": 0.002,
+      "step": 11399
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2751716521822634e-05,
+      "loss": 0.7784,
+      "step": 11400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.274931579200077e-05,
+      "loss": 0.0887,
+      "step": 11401
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.27469150621789e-05,
+      "loss": 0.0014,
+      "step": 11402
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2744514332357037e-05,
+      "loss": 1.4024,
+      "step": 11403
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2742113602535172e-05,
+      "loss": 0.0026,
+      "step": 11404
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2739712872713307e-05,
+      "loss": 0.0026,
+      "step": 11405
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.273731214289144e-05,
+      "loss": 0.0018,
+      "step": 11406
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2734911413069574e-05,
+      "loss": 0.6983,
+      "step": 11407
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.273251068324771e-05,
+      "loss": 0.6932,
+      "step": 11408
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2730109953425844e-05,
+      "loss": 0.7245,
+      "step": 11409
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2727709223603976e-05,
+      "loss": 0.0031,
+      "step": 11410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.272530849378211e-05,
+      "loss": 0.0029,
+      "step": 11411
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2722907763960243e-05,
+      "loss": 0.0029,
+      "step": 11412
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2720507034138382e-05,
+      "loss": 0.0034,
+      "step": 11413
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2718106304316514e-05,
+      "loss": 0.6579,
+      "step": 11414
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.271570557449465e-05,
+      "loss": 0.0041,
+      "step": 11415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.271330484467278e-05,
+      "loss": 0.0032,
+      "step": 11416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2710904114850916e-05,
+      "loss": 0.0031,
+      "step": 11417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.270850338502905e-05,
+      "loss": 0.0063,
+      "step": 11418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2706102655207183e-05,
+      "loss": 0.621,
+      "step": 11419
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2703701925385318e-05,
+      "loss": 0.0059,
+      "step": 11420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.270130119556345e-05,
+      "loss": 0.0062,
+      "step": 11421
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2698900465741588e-05,
+      "loss": 0.003,
+      "step": 11422
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.269649973591972e-05,
+      "loss": 0.0053,
+      "step": 11423
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2694099006097855e-05,
+      "loss": 0.0043,
+      "step": 11424
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2691698276275987e-05,
+      "loss": 0.0065,
+      "step": 11425
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2689297546454122e-05,
+      "loss": 0.0692,
+      "step": 11426
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2686896816632257e-05,
+      "loss": 0.0023,
+      "step": 11427
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2684496086810393e-05,
+      "loss": 0.0029,
+      "step": 11428
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2682095356988524e-05,
+      "loss": 0.0048,
+      "step": 11429
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.267969462716666e-05,
+      "loss": 0.0045,
+      "step": 11430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2677293897344795e-05,
+      "loss": 0.0059,
+      "step": 11431
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.267489316752293e-05,
+      "loss": 0.0186,
+      "step": 11432
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2672492437701062e-05,
+      "loss": 0.0055,
+      "step": 11433
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2670091707879197e-05,
+      "loss": 0.0034,
+      "step": 11434
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.266769097805733e-05,
+      "loss": 0.5651,
+      "step": 11435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2665290248235464e-05,
+      "loss": 0.0044,
+      "step": 11436
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.26628895184136e-05,
+      "loss": 0.0034,
+      "step": 11437
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2660488788591734e-05,
+      "loss": 0.0026,
+      "step": 11438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2658088058769866e-05,
+      "loss": 0.0031,
+      "step": 11439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2655687328948e-05,
+      "loss": 0.0036,
+      "step": 11440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2653286599126137e-05,
+      "loss": 0.002,
+      "step": 11441
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.265088586930427e-05,
+      "loss": 0.0057,
+      "step": 11442
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2648485139482404e-05,
+      "loss": 0.0036,
+      "step": 11443
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2646084409660535e-05,
+      "loss": 0.0028,
+      "step": 11444
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2643683679838674e-05,
+      "loss": 0.0035,
+      "step": 11445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2641282950016806e-05,
+      "loss": 0.6339,
+      "step": 11446
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.263888222019494e-05,
+      "loss": 0.0062,
+      "step": 11447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2636481490373073e-05,
+      "loss": 0.0028,
+      "step": 11448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2634080760551208e-05,
+      "loss": 0.0034,
+      "step": 11449
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2631680030729343e-05,
+      "loss": 0.0029,
+      "step": 11450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2629279300907478e-05,
+      "loss": 0.0017,
+      "step": 11451
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.262687857108561e-05,
+      "loss": 0.0037,
+      "step": 11452
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2624477841263745e-05,
+      "loss": 0.0023,
+      "step": 11453
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.262207711144188e-05,
+      "loss": 0.0035,
+      "step": 11454
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2619676381620016e-05,
+      "loss": 0.0014,
+      "step": 11455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2617275651798147e-05,
+      "loss": 0.0036,
+      "step": 11456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2614874921976283e-05,
+      "loss": 0.0036,
+      "step": 11457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2612474192154414e-05,
+      "loss": 0.001,
+      "step": 11458
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.261007346233255e-05,
+      "loss": 0.0036,
+      "step": 11459
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2607672732510685e-05,
+      "loss": 0.0031,
+      "step": 11460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2605272002688817e-05,
+      "loss": 0.0032,
+      "step": 11461
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2602871272866952e-05,
+      "loss": 0.0025,
+      "step": 11462
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2600470543045087e-05,
+      "loss": 0.0021,
+      "step": 11463
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2598069813223222e-05,
+      "loss": 0.0022,
+      "step": 11464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2595669083401354e-05,
+      "loss": 0.0023,
+      "step": 11465
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.259326835357949e-05,
+      "loss": 0.0025,
+      "step": 11466
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.259086762375762e-05,
+      "loss": 0.0039,
+      "step": 11467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.258846689393576e-05,
+      "loss": 0.0044,
+      "step": 11468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.258606616411389e-05,
+      "loss": 0.003,
+      "step": 11469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2583665434292027e-05,
+      "loss": 0.0012,
+      "step": 11470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.258126470447016e-05,
+      "loss": 0.003,
+      "step": 11471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2578863974648297e-05,
+      "loss": 0.0026,
+      "step": 11472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.257646324482643e-05,
+      "loss": 0.0021,
+      "step": 11473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2574062515004564e-05,
+      "loss": 0.0018,
+      "step": 11474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2571661785182696e-05,
+      "loss": 0.0017,
+      "step": 11475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.256926105536083e-05,
+      "loss": 0.0012,
+      "step": 11476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2566860325538966e-05,
+      "loss": 0.7427,
+      "step": 11477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2564459595717098e-05,
+      "loss": 0.6988,
+      "step": 11478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2562058865895233e-05,
+      "loss": 0.0014,
+      "step": 11479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2559658136073368e-05,
+      "loss": 0.0027,
+      "step": 11480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.25572574062515e-05,
+      "loss": 0.0022,
+      "step": 11481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2554856676429635e-05,
+      "loss": 0.0029,
+      "step": 11482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.255245594660777e-05,
+      "loss": 0.0024,
+      "step": 11483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2550055216785902e-05,
+      "loss": 0.0018,
+      "step": 11484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2547654486964037e-05,
+      "loss": 0.0022,
+      "step": 11485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2545253757142173e-05,
+      "loss": 0.0028,
+      "step": 11486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2542853027320308e-05,
+      "loss": 0.0029,
+      "step": 11487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.254045229749844e-05,
+      "loss": 0.0022,
+      "step": 11488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2538051567676575e-05,
+      "loss": 0.0026,
+      "step": 11489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2535650837854707e-05,
+      "loss": 0.0038,
+      "step": 11490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2533250108032845e-05,
+      "loss": 0.0012,
+      "step": 11491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2530849378210977e-05,
+      "loss": 0.0018,
+      "step": 11492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2528448648389112e-05,
+      "loss": 0.0029,
+      "step": 11493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2526047918567244e-05,
+      "loss": 0.6581,
+      "step": 11494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2523647188745383e-05,
+      "loss": 0.002,
+      "step": 11495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2521246458923514e-05,
+      "loss": 1.0044,
+      "step": 11496
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.251884572910165e-05,
+      "loss": 0.0016,
+      "step": 11497
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.251644499927978e-05,
+      "loss": 0.0016,
+      "step": 11498
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2514044269457916e-05,
+      "loss": 0.0013,
+      "step": 11499
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.251164353963605e-05,
+      "loss": 0.0027,
+      "step": 11500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2509242809814183e-05,
+      "loss": 0.0022,
+      "step": 11501
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.250684207999232e-05,
+      "loss": 0.0025,
+      "step": 11502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.250444135017045e-05,
+      "loss": 0.0023,
+      "step": 11503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.250204062034859e-05,
+      "loss": 0.0036,
+      "step": 11504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.249963989052672e-05,
+      "loss": 0.0026,
+      "step": 11505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2497239160704856e-05,
+      "loss": 0.0027,
+      "step": 11506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2494838430882988e-05,
+      "loss": 0.0016,
+      "step": 11507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2492437701061123e-05,
+      "loss": 0.0017,
+      "step": 11508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2490036971239258e-05,
+      "loss": 0.0017,
+      "step": 11509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2487636241417393e-05,
+      "loss": 0.0015,
+      "step": 11510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2485235511595525e-05,
+      "loss": 0.0018,
+      "step": 11511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.248283478177366e-05,
+      "loss": 0.0018,
+      "step": 11512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2480434051951792e-05,
+      "loss": 0.0025,
+      "step": 11513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.247803332212993e-05,
+      "loss": 0.0022,
+      "step": 11514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2475632592308063e-05,
+      "loss": 0.0025,
+      "step": 11515
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2473231862486198e-05,
+      "loss": 0.0022,
+      "step": 11516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.247083113266433e-05,
+      "loss": 0.0029,
+      "step": 11517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2468430402842465e-05,
+      "loss": 0.0015,
+      "step": 11518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.24660296730206e-05,
+      "loss": 0.0013,
+      "step": 11519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2463628943198735e-05,
+      "loss": 0.0026,
+      "step": 11520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2461228213376867e-05,
+      "loss": 0.0027,
+      "step": 11521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2458827483555002e-05,
+      "loss": 0.0013,
+      "step": 11522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2456426753733137e-05,
+      "loss": 0.0014,
+      "step": 11523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.245402602391127e-05,
+      "loss": 0.0009,
+      "step": 11524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2451625294089404e-05,
+      "loss": 0.0017,
+      "step": 11525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2449224564267536e-05,
+      "loss": 0.0015,
+      "step": 11526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2446823834445675e-05,
+      "loss": 0.0025,
+      "step": 11527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2444423104623806e-05,
+      "loss": 0.0022,
+      "step": 11528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.244202237480194e-05,
+      "loss": 0.0009,
+      "step": 11529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2439621644980073e-05,
+      "loss": 0.0022,
+      "step": 11530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.243722091515821e-05,
+      "loss": 0.0015,
+      "step": 11531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2434820185336344e-05,
+      "loss": 0.0021,
+      "step": 11532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.243241945551448e-05,
+      "loss": 0.002,
+      "step": 11533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.243001872569261e-05,
+      "loss": 0.0015,
+      "step": 11534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2427617995870746e-05,
+      "loss": 0.0012,
+      "step": 11535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.242521726604888e-05,
+      "loss": 0.0018,
+      "step": 11536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2422816536227016e-05,
+      "loss": 0.0021,
+      "step": 11537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2420415806405148e-05,
+      "loss": 0.0011,
+      "step": 11538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2418015076583283e-05,
+      "loss": 0.0008,
+      "step": 11539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2415614346761415e-05,
+      "loss": 0.0015,
+      "step": 11540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.241321361693955e-05,
+      "loss": 0.0017,
+      "step": 11541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2410812887117686e-05,
+      "loss": 0.0012,
+      "step": 11542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2408412157295817e-05,
+      "loss": 0.0007,
+      "step": 11543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2406011427473953e-05,
+      "loss": 0.0013,
+      "step": 11544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2403610697652084e-05,
+      "loss": 0.648,
+      "step": 11545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2401209967830223e-05,
+      "loss": 0.0017,
+      "step": 11546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2398809238008355e-05,
+      "loss": 0.001,
+      "step": 11547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.239640850818649e-05,
+      "loss": 0.0008,
+      "step": 11548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2394007778364622e-05,
+      "loss": 0.0015,
+      "step": 11549
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.239160704854276e-05,
+      "loss": 0.0015,
+      "step": 11550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2389206318720892e-05,
+      "loss": 0.0015,
+      "step": 11551
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2386805588899027e-05,
+      "loss": 0.0014,
+      "step": 11552
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.238440485907716e-05,
+      "loss": 0.002,
+      "step": 11553
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2382004129255294e-05,
+      "loss": 0.0017,
+      "step": 11554
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.237960339943343e-05,
+      "loss": 0.0019,
+      "step": 11555
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2377202669611565e-05,
+      "loss": 0.0016,
+      "step": 11556
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2374801939789696e-05,
+      "loss": 0.0012,
+      "step": 11557
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.237240120996783e-05,
+      "loss": 0.0006,
+      "step": 11558
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2370000480145967e-05,
+      "loss": 0.0009,
+      "step": 11559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.23675997503241e-05,
+      "loss": 0.0013,
+      "step": 11560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2365199020502234e-05,
+      "loss": 0.0015,
+      "step": 11561
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.236279829068037e-05,
+      "loss": 0.0019,
+      "step": 11562
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.23603975608585e-05,
+      "loss": 0.0009,
+      "step": 11563
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2357996831036636e-05,
+      "loss": 0.0011,
+      "step": 11564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.235559610121477e-05,
+      "loss": 0.0018,
+      "step": 11565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2353195371392903e-05,
+      "loss": 0.042,
+      "step": 11566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2350794641571038e-05,
+      "loss": 0.0009,
+      "step": 11567
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2348393911749173e-05,
+      "loss": 0.799,
+      "step": 11568
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.234599318192731e-05,
+      "loss": 0.0015,
+      "step": 11569
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.234359245210544e-05,
+      "loss": 0.0008,
+      "step": 11570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2341191722283576e-05,
+      "loss": 0.0008,
+      "step": 11571
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2338790992461707e-05,
+      "loss": 0.001,
+      "step": 11572
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2336390262639846e-05,
+      "loss": 0.0017,
+      "step": 11573
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2333989532817978e-05,
+      "loss": 0.0011,
+      "step": 11574
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2331588802996113e-05,
+      "loss": 0.0013,
+      "step": 11575
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2329188073174245e-05,
+      "loss": 0.0011,
+      "step": 11576
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.232678734335238e-05,
+      "loss": 0.0015,
+      "step": 11577
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2324386613530515e-05,
+      "loss": 0.0009,
+      "step": 11578
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.232198588370865e-05,
+      "loss": 0.0012,
+      "step": 11579
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2319585153886782e-05,
+      "loss": 0.0014,
+      "step": 11580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2317184424064917e-05,
+      "loss": 0.0013,
+      "step": 11581
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2314783694243052e-05,
+      "loss": 0.0023,
+      "step": 11582
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2312382964421184e-05,
+      "loss": 0.0011,
+      "step": 11583
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.230998223459932e-05,
+      "loss": 0.0015,
+      "step": 11584
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.230758150477745e-05,
+      "loss": 0.0012,
+      "step": 11585
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2305180774955586e-05,
+      "loss": 0.0015,
+      "step": 11586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.230278004513372e-05,
+      "loss": 0.9638,
+      "step": 11587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2300379315311857e-05,
+      "loss": 0.0025,
+      "step": 11588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.229797858548999e-05,
+      "loss": 0.0012,
+      "step": 11589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2295577855668124e-05,
+      "loss": 0.0016,
+      "step": 11590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.229317712584626e-05,
+      "loss": 0.0013,
+      "step": 11591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2290776396024394e-05,
+      "loss": 0.001,
+      "step": 11592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2288375666202526e-05,
+      "loss": 0.0011,
+      "step": 11593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.228597493638066e-05,
+      "loss": 0.0005,
+      "step": 11594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2283574206558793e-05,
+      "loss": 0.0012,
+      "step": 11595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.228117347673693e-05,
+      "loss": 0.0011,
+      "step": 11596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2278772746915063e-05,
+      "loss": 0.0012,
+      "step": 11597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.22763720170932e-05,
+      "loss": 0.0007,
+      "step": 11598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.227397128727133e-05,
+      "loss": 0.0012,
+      "step": 11599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2271570557449465e-05,
+      "loss": 0.0019,
+      "step": 11600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.22691698276276e-05,
+      "loss": 0.0009,
+      "step": 11601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2266769097805732e-05,
+      "loss": 0.0007,
+      "step": 11602
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2264368367983868e-05,
+      "loss": 0.0012,
+      "step": 11603
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2261967638162003e-05,
+      "loss": 0.0011,
+      "step": 11604
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2259566908340138e-05,
+      "loss": 0.0008,
+      "step": 11605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.225716617851827e-05,
+      "loss": 0.0015,
+      "step": 11606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2254765448696405e-05,
+      "loss": 0.0008,
+      "step": 11607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2252364718874537e-05,
+      "loss": 0.0013,
+      "step": 11608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2249963989052672e-05,
+      "loss": 0.0011,
+      "step": 11609
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2247563259230807e-05,
+      "loss": 0.0007,
+      "step": 11610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2245162529408942e-05,
+      "loss": 0.0017,
+      "step": 11611
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2242761799587074e-05,
+      "loss": 0.0008,
+      "step": 11612
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.224036106976521e-05,
+      "loss": 0.0013,
+      "step": 11613
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2237960339943345e-05,
+      "loss": 0.0011,
+      "step": 11614
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.223555961012148e-05,
+      "loss": 0.0009,
+      "step": 11615
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.223315888029961e-05,
+      "loss": 0.0011,
+      "step": 11616
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2230758150477747e-05,
+      "loss": 0.0007,
+      "step": 11617
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.222835742065588e-05,
+      "loss": 0.0012,
+      "step": 11618
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2225956690834017e-05,
+      "loss": 0.0009,
+      "step": 11619
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.222355596101215e-05,
+      "loss": 0.0012,
+      "step": 11620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2221155231190284e-05,
+      "loss": 0.0007,
+      "step": 11621
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2218754501368416e-05,
+      "loss": 0.0014,
+      "step": 11622
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.221635377154655e-05,
+      "loss": 0.0006,
+      "step": 11623
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2213953041724686e-05,
+      "loss": 0.0006,
+      "step": 11624
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2211552311902818e-05,
+      "loss": 0.0009,
+      "step": 11625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2209151582080953e-05,
+      "loss": 0.0013,
+      "step": 11626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2206750852259085e-05,
+      "loss": 0.0012,
+      "step": 11627
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2204350122437224e-05,
+      "loss": 0.001,
+      "step": 11628
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2201949392615355e-05,
+      "loss": 0.0008,
+      "step": 11629
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.219954866279349e-05,
+      "loss": 0.0006,
+      "step": 11630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2197147932971622e-05,
+      "loss": 0.0008,
+      "step": 11631
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.219474720314976e-05,
+      "loss": 0.0007,
+      "step": 11632
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2192346473327893e-05,
+      "loss": 0.0014,
+      "step": 11633
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2189945743506028e-05,
+      "loss": 0.0007,
+      "step": 11634
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.218754501368416e-05,
+      "loss": 0.0006,
+      "step": 11635
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2185144283862295e-05,
+      "loss": 0.001,
+      "step": 11636
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.218274355404043e-05,
+      "loss": 0.0005,
+      "step": 11637
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2180342824218565e-05,
+      "loss": 0.001,
+      "step": 11638
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2177942094396697e-05,
+      "loss": 0.001,
+      "step": 11639
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2175541364574832e-05,
+      "loss": 0.001,
+      "step": 11640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2173140634752964e-05,
+      "loss": 0.0015,
+      "step": 11641
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.21707399049311e-05,
+      "loss": 0.0009,
+      "step": 11642
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2168339175109235e-05,
+      "loss": 0.0009,
+      "step": 11643
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.216593844528737e-05,
+      "loss": 0.001,
+      "step": 11644
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.21635377154655e-05,
+      "loss": 0.0007,
+      "step": 11645
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2161136985643637e-05,
+      "loss": 0.0004,
+      "step": 11646
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2158736255821772e-05,
+      "loss": 0.0005,
+      "step": 11647
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2156335525999904e-05,
+      "loss": 0.001,
+      "step": 11648
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.215393479617804e-05,
+      "loss": 0.0008,
+      "step": 11649
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.215153406635617e-05,
+      "loss": 0.0007,
+      "step": 11650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.214913333653431e-05,
+      "loss": 0.0007,
+      "step": 11651
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.214673260671244e-05,
+      "loss": 0.0008,
+      "step": 11652
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2144331876890576e-05,
+      "loss": 0.0006,
+      "step": 11653
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2141931147068708e-05,
+      "loss": 0.0007,
+      "step": 11654
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2139530417246847e-05,
+      "loss": 0.0005,
+      "step": 11655
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.213712968742498e-05,
+      "loss": 0.0009,
+      "step": 11656
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2134728957603114e-05,
+      "loss": 0.0004,
+      "step": 11657
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2132328227781245e-05,
+      "loss": 0.0012,
+      "step": 11658
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.212992749795938e-05,
+      "loss": 0.0008,
+      "step": 11659
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2127526768137516e-05,
+      "loss": 0.0007,
+      "step": 11660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.212512603831565e-05,
+      "loss": 0.0008,
+      "step": 11661
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2122725308493783e-05,
+      "loss": 0.001,
+      "step": 11662
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2120324578671918e-05,
+      "loss": 0.0007,
+      "step": 11663
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.211792384885005e-05,
+      "loss": 0.0011,
+      "step": 11664
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2115523119028185e-05,
+      "loss": 0.001,
+      "step": 11665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.211312238920632e-05,
+      "loss": 0.0008,
+      "step": 11666
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2110721659384452e-05,
+      "loss": 0.0009,
+      "step": 11667
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2108320929562587e-05,
+      "loss": 0.0008,
+      "step": 11668
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2105920199740722e-05,
+      "loss": 0.0006,
+      "step": 11669
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2103519469918858e-05,
+      "loss": 0.8771,
+      "step": 11670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.210111874009699e-05,
+      "loss": 0.0007,
+      "step": 11671
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2098718010275125e-05,
+      "loss": 0.0006,
+      "step": 11672
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2096317280453256e-05,
+      "loss": 0.0009,
+      "step": 11673
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2093916550631395e-05,
+      "loss": 0.0008,
+      "step": 11674
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2091515820809527e-05,
+      "loss": 0.0007,
+      "step": 11675
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2089115090987662e-05,
+      "loss": 0.001,
+      "step": 11676
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2086714361165794e-05,
+      "loss": 0.0005,
+      "step": 11677
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2084313631343932e-05,
+      "loss": 0.0007,
+      "step": 11678
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2081912901522064e-05,
+      "loss": 0.0009,
+      "step": 11679
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.20795121717002e-05,
+      "loss": 0.875,
+      "step": 11680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.207711144187833e-05,
+      "loss": 0.0007,
+      "step": 11681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2074710712056466e-05,
+      "loss": 0.0008,
+      "step": 11682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.20723099822346e-05,
+      "loss": 0.0005,
+      "step": 11683
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2069909252412733e-05,
+      "loss": 0.0006,
+      "step": 11684
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.206750852259087e-05,
+      "loss": 0.0011,
+      "step": 11685
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2065107792769004e-05,
+      "loss": 0.0006,
+      "step": 11686
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.206270706294714e-05,
+      "loss": 0.0007,
+      "step": 11687
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.206030633312527e-05,
+      "loss": 0.0006,
+      "step": 11688
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2057905603303406e-05,
+      "loss": 0.003,
+      "step": 11689
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2055504873481538e-05,
+      "loss": 0.001,
+      "step": 11690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2053104143659673e-05,
+      "loss": 0.0009,
+      "step": 11691
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2050703413837808e-05,
+      "loss": 0.0006,
+      "step": 11692
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2048302684015943e-05,
+      "loss": 0.0008,
+      "step": 11693
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2045901954194075e-05,
+      "loss": 0.0007,
+      "step": 11694
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.204350122437221e-05,
+      "loss": 0.001,
+      "step": 11695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2041100494550342e-05,
+      "loss": 0.0007,
+      "step": 11696
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.203869976472848e-05,
+      "loss": 0.0007,
+      "step": 11697
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2036299034906612e-05,
+      "loss": 0.0012,
+      "step": 11698
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2033898305084748e-05,
+      "loss": 0.0012,
+      "step": 11699
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.203149757526288e-05,
+      "loss": 0.001,
+      "step": 11700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2029096845441018e-05,
+      "loss": 0.7032,
+      "step": 11701
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.202669611561915e-05,
+      "loss": 0.0005,
+      "step": 11702
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2024295385797285e-05,
+      "loss": 0.001,
+      "step": 11703
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2021894655975417e-05,
+      "loss": 0.0008,
+      "step": 11704
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2019493926153552e-05,
+      "loss": 0.0009,
+      "step": 11705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2017093196331687e-05,
+      "loss": 0.0005,
+      "step": 11706
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.201469246650982e-05,
+      "loss": 0.0007,
+      "step": 11707
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2012291736687954e-05,
+      "loss": 0.0011,
+      "step": 11708
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2009891006866086e-05,
+      "loss": 0.0007,
+      "step": 11709
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2007490277044224e-05,
+      "loss": 0.001,
+      "step": 11710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2005089547222356e-05,
+      "loss": 0.001,
+      "step": 11711
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.200268881740049e-05,
+      "loss": 0.0006,
+      "step": 11712
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2000288087578623e-05,
+      "loss": 0.001,
+      "step": 11713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.199788735775676e-05,
+      "loss": 0.0008,
+      "step": 11714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1995486627934894e-05,
+      "loss": 0.0007,
+      "step": 11715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.199308589811303e-05,
+      "loss": 0.001,
+      "step": 11716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.199068516829116e-05,
+      "loss": 0.0006,
+      "step": 11717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1988284438469296e-05,
+      "loss": 0.0006,
+      "step": 11718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.198588370864743e-05,
+      "loss": 0.001,
+      "step": 11719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1983482978825566e-05,
+      "loss": 0.88,
+      "step": 11720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1981082249003698e-05,
+      "loss": 0.0009,
+      "step": 11721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1978681519181833e-05,
+      "loss": 0.0008,
+      "step": 11722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1976280789359965e-05,
+      "loss": 0.8172,
+      "step": 11723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.19738800595381e-05,
+      "loss": 0.0011,
+      "step": 11724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1971479329716235e-05,
+      "loss": 0.0011,
+      "step": 11725
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1969078599894367e-05,
+      "loss": 0.0005,
+      "step": 11726
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1966677870072502e-05,
+      "loss": 0.0008,
+      "step": 11727
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1964277140250637e-05,
+      "loss": 0.001,
+      "step": 11728
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1961876410428773e-05,
+      "loss": 0.0009,
+      "step": 11729
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1959475680606904e-05,
+      "loss": 0.0008,
+      "step": 11730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.195707495078504e-05,
+      "loss": 0.0012,
+      "step": 11731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.195467422096317e-05,
+      "loss": 0.0006,
+      "step": 11732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.195227349114131e-05,
+      "loss": 0.0008,
+      "step": 11733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1949872761319442e-05,
+      "loss": 0.8556,
+      "step": 11734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1947472031497577e-05,
+      "loss": 0.0008,
+      "step": 11735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.194507130167571e-05,
+      "loss": 0.0097,
+      "step": 11736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1942670571853844e-05,
+      "loss": 0.0006,
+      "step": 11737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.194026984203198e-05,
+      "loss": 0.0009,
+      "step": 11738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1937869112210114e-05,
+      "loss": 0.0011,
+      "step": 11739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1935468382388246e-05,
+      "loss": 0.001,
+      "step": 11740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.193306765256638e-05,
+      "loss": 0.0014,
+      "step": 11741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1930666922744517e-05,
+      "loss": 0.7871,
+      "step": 11742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1928266192922652e-05,
+      "loss": 0.0012,
+      "step": 11743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1925865463100784e-05,
+      "loss": 0.0011,
+      "step": 11744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.192346473327892e-05,
+      "loss": 0.0008,
+      "step": 11745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.192106400345705e-05,
+      "loss": 0.0016,
+      "step": 11746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1918663273635186e-05,
+      "loss": 0.0198,
+      "step": 11747
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.191626254381332e-05,
+      "loss": 0.0272,
+      "step": 11748
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1913861813991453e-05,
+      "loss": 0.0117,
+      "step": 11749
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1911461084169588e-05,
+      "loss": 0.003,
+      "step": 11750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1909060354347723e-05,
+      "loss": 0.0591,
+      "step": 11751
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1906659624525858e-05,
+      "loss": 0.0051,
+      "step": 11752
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.190425889470399e-05,
+      "loss": 0.0019,
+      "step": 11753
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1901858164882125e-05,
+      "loss": 0.0021,
+      "step": 11754
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1899457435060257e-05,
+      "loss": 0.0021,
+      "step": 11755
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1897056705238396e-05,
+      "loss": 0.0008,
+      "step": 11756
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1894655975416527e-05,
+      "loss": 0.0055,
+      "step": 11757
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1892255245594663e-05,
+      "loss": 0.0016,
+      "step": 11758
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1889854515772794e-05,
+      "loss": 0.0083,
+      "step": 11759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.188745378595093e-05,
+      "loss": 0.0066,
+      "step": 11760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1885053056129065e-05,
+      "loss": 0.0016,
+      "step": 11761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.18826523263072e-05,
+      "loss": 0.0005,
+      "step": 11762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1880251596485332e-05,
+      "loss": 0.0007,
+      "step": 11763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1877850866663467e-05,
+      "loss": 0.0003,
+      "step": 11764
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1875450136841602e-05,
+      "loss": 0.0023,
+      "step": 11765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1873049407019734e-05,
+      "loss": 0.0014,
+      "step": 11766
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.187064867719787e-05,
+      "loss": 0.0008,
+      "step": 11767
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1868247947376004e-05,
+      "loss": 0.0007,
+      "step": 11768
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1865847217554136e-05,
+      "loss": 0.0009,
+      "step": 11769
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.186344648773227e-05,
+      "loss": 0.0025,
+      "step": 11770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1861045757910407e-05,
+      "loss": 0.0007,
+      "step": 11771
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.185864502808854e-05,
+      "loss": 0.0008,
+      "step": 11772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1856244298266674e-05,
+      "loss": 0.0006,
+      "step": 11773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.185384356844481e-05,
+      "loss": 0.0009,
+      "step": 11774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1851442838622944e-05,
+      "loss": 0.0011,
+      "step": 11775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1849042108801076e-05,
+      "loss": 0.001,
+      "step": 11776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.184664137897921e-05,
+      "loss": 0.0015,
+      "step": 11777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1844240649157343e-05,
+      "loss": 0.0009,
+      "step": 11778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.184183991933548e-05,
+      "loss": 0.0005,
+      "step": 11779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1839439189513613e-05,
+      "loss": 0.0009,
+      "step": 11780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1837038459691748e-05,
+      "loss": 0.0007,
+      "step": 11781
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.183463772986988e-05,
+      "loss": 0.0015,
+      "step": 11782
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.183223700004802e-05,
+      "loss": 0.0012,
+      "step": 11783
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.182983627022615e-05,
+      "loss": 0.0008,
+      "step": 11784
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1827435540404286e-05,
+      "loss": 0.0084,
+      "step": 11785
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1825034810582417e-05,
+      "loss": 0.001,
+      "step": 11786
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1822634080760553e-05,
+      "loss": 0.0012,
+      "step": 11787
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1820233350938688e-05,
+      "loss": 0.0007,
+      "step": 11788
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.181783262111682e-05,
+      "loss": 0.0006,
+      "step": 11789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1815431891294955e-05,
+      "loss": 0.0011,
+      "step": 11790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1813031161473087e-05,
+      "loss": 0.0003,
+      "step": 11791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1810630431651222e-05,
+      "loss": 0.0009,
+      "step": 11792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1808229701829357e-05,
+      "loss": 0.0056,
+      "step": 11793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1805828972007492e-05,
+      "loss": 1.005,
+      "step": 11794
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1803428242185624e-05,
+      "loss": 0.0008,
+      "step": 11795
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.180102751236376e-05,
+      "loss": 0.0005,
+      "step": 11796
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1798626782541894e-05,
+      "loss": 0.0009,
+      "step": 11797
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.179622605272003e-05,
+      "loss": 0.0011,
+      "step": 11798
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.179382532289816e-05,
+      "loss": 0.0013,
+      "step": 11799
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1791424593076297e-05,
+      "loss": 0.7489,
+      "step": 11800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.178902386325443e-05,
+      "loss": 0.0009,
+      "step": 11801
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1786623133432567e-05,
+      "loss": 0.0005,
+      "step": 11802
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.17842224036107e-05,
+      "loss": 0.0007,
+      "step": 11803
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1781821673788834e-05,
+      "loss": 0.0009,
+      "step": 11804
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1779420943966966e-05,
+      "loss": 0.0009,
+      "step": 11805
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.17770202141451e-05,
+      "loss": 0.0009,
+      "step": 11806
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1774619484323236e-05,
+      "loss": 0.0014,
+      "step": 11807
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1772218754501368e-05,
+      "loss": 0.001,
+      "step": 11808
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1769818024679503e-05,
+      "loss": 0.001,
+      "step": 11809
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1767417294857638e-05,
+      "loss": 0.8165,
+      "step": 11810
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1765016565035773e-05,
+      "loss": 0.0009,
+      "step": 11811
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1762615835213905e-05,
+      "loss": 0.0011,
+      "step": 11812
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.176021510539204e-05,
+      "loss": 0.6641,
+      "step": 11813
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1757814375570172e-05,
+      "loss": 0.001,
+      "step": 11814
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.175541364574831e-05,
+      "loss": 0.001,
+      "step": 11815
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1753012915926443e-05,
+      "loss": 0.0012,
+      "step": 11816
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1750612186104578e-05,
+      "loss": 0.0035,
+      "step": 11817
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.174821145628271e-05,
+      "loss": 0.0016,
+      "step": 11818
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1745810726460845e-05,
+      "loss": 0.0011,
+      "step": 11819
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.174340999663898e-05,
+      "loss": 0.0014,
+      "step": 11820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1741009266817115e-05,
+      "loss": 0.0019,
+      "step": 11821
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1738608536995247e-05,
+      "loss": 0.0032,
+      "step": 11822
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1736207807173382e-05,
+      "loss": 0.0009,
+      "step": 11823
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1733807077351514e-05,
+      "loss": 0.0012,
+      "step": 11824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1731406347529653e-05,
+      "loss": 0.0012,
+      "step": 11825
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1729005617707784e-05,
+      "loss": 0.001,
+      "step": 11826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.172660488788592e-05,
+      "loss": 0.0016,
+      "step": 11827
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.172420415806405e-05,
+      "loss": 0.0015,
+      "step": 11828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1721803428242186e-05,
+      "loss": 0.0041,
+      "step": 11829
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.171940269842032e-05,
+      "loss": 0.0008,
+      "step": 11830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1717001968598453e-05,
+      "loss": 0.0009,
+      "step": 11831
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.171460123877659e-05,
+      "loss": 0.0009,
+      "step": 11832
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.171220050895472e-05,
+      "loss": 0.0013,
+      "step": 11833
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.170979977913286e-05,
+      "loss": 0.0009,
+      "step": 11834
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.170739904931099e-05,
+      "loss": 0.4341,
+      "step": 11835
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1704998319489126e-05,
+      "loss": 0.0013,
+      "step": 11836
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1702597589667258e-05,
+      "loss": 0.0012,
+      "step": 11837
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1700196859845396e-05,
+      "loss": 0.001,
+      "step": 11838
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1697796130023528e-05,
+      "loss": 0.0008,
+      "step": 11839
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1695395400201663e-05,
+      "loss": 0.0007,
+      "step": 11840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1692994670379795e-05,
+      "loss": 0.0011,
+      "step": 11841
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.169059394055793e-05,
+      "loss": 0.0008,
+      "step": 11842
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1688193210736066e-05,
+      "loss": 0.8692,
+      "step": 11843
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.16857924809142e-05,
+      "loss": 0.0009,
+      "step": 11844
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1683391751092333e-05,
+      "loss": 0.0009,
+      "step": 11845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1680991021270468e-05,
+      "loss": 0.001,
+      "step": 11846
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1678590291448603e-05,
+      "loss": 0.0011,
+      "step": 11847
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1676189561626735e-05,
+      "loss": 0.001,
+      "step": 11848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.167378883180487e-05,
+      "loss": 0.2676,
+      "step": 11849
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1671388101983002e-05,
+      "loss": 0.0009,
+      "step": 11850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1668987372161137e-05,
+      "loss": 0.001,
+      "step": 11851
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1666586642339272e-05,
+      "loss": 0.0031,
+      "step": 11852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1664185912517407e-05,
+      "loss": 0.001,
+      "step": 11853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.166178518269554e-05,
+      "loss": 0.0007,
+      "step": 11854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1659384452873674e-05,
+      "loss": 0.0008,
+      "step": 11855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1656983723051806e-05,
+      "loss": 0.0018,
+      "step": 11856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1654582993229945e-05,
+      "loss": 0.0015,
+      "step": 11857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1652182263408076e-05,
+      "loss": 0.0006,
+      "step": 11858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.164978153358621e-05,
+      "loss": 0.0011,
+      "step": 11859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1647380803764343e-05,
+      "loss": 0.0008,
+      "step": 11860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1644980073942482e-05,
+      "loss": 0.0009,
+      "step": 11861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1642579344120614e-05,
+      "loss": 0.0007,
+      "step": 11862
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.164017861429875e-05,
+      "loss": 0.0009,
+      "step": 11863
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.163777788447688e-05,
+      "loss": 0.0006,
+      "step": 11864
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1635377154655016e-05,
+      "loss": 0.0009,
+      "step": 11865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.163297642483315e-05,
+      "loss": 0.0011,
+      "step": 11866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1630575695011286e-05,
+      "loss": 0.0009,
+      "step": 11867
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1628174965189418e-05,
+      "loss": 0.0026,
+      "step": 11868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1625774235367553e-05,
+      "loss": 0.0008,
+      "step": 11869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.162337350554569e-05,
+      "loss": 0.0015,
+      "step": 11870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.162097277572382e-05,
+      "loss": 0.0006,
+      "step": 11871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1618572045901956e-05,
+      "loss": 0.0011,
+      "step": 11872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1616171316080087e-05,
+      "loss": 0.0009,
+      "step": 11873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1613770586258223e-05,
+      "loss": 0.0006,
+      "step": 11874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1611369856436358e-05,
+      "loss": 0.0011,
+      "step": 11875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1608969126614493e-05,
+      "loss": 0.0007,
+      "step": 11876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1606568396792625e-05,
+      "loss": 0.001,
+      "step": 11877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.160416766697076e-05,
+      "loss": 0.0008,
+      "step": 11878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1601766937148895e-05,
+      "loss": 0.0011,
+      "step": 11879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.159936620732703e-05,
+      "loss": 0.0009,
+      "step": 11880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1596965477505162e-05,
+      "loss": 0.0005,
+      "step": 11881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1594564747683297e-05,
+      "loss": 0.0007,
+      "step": 11882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.159216401786143e-05,
+      "loss": 0.0009,
+      "step": 11883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1589763288039568e-05,
+      "loss": 0.0006,
+      "step": 11884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.15873625582177e-05,
+      "loss": 0.0038,
+      "step": 11885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1584961828395835e-05,
+      "loss": 0.0011,
+      "step": 11886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1582561098573966e-05,
+      "loss": 0.0006,
+      "step": 11887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.15801603687521e-05,
+      "loss": 0.0007,
+      "step": 11888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1577759638930237e-05,
+      "loss": 0.0006,
+      "step": 11889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.157535890910837e-05,
+      "loss": 0.0007,
+      "step": 11890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1572958179286504e-05,
+      "loss": 0.8218,
+      "step": 11891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.157055744946464e-05,
+      "loss": 0.0005,
+      "step": 11892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1568156719642774e-05,
+      "loss": 0.0012,
+      "step": 11893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1565755989820906e-05,
+      "loss": 0.0007,
+      "step": 11894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.156335525999904e-05,
+      "loss": 0.0005,
+      "step": 11895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1560954530177173e-05,
+      "loss": 0.0008,
+      "step": 11896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1558553800355308e-05,
+      "loss": 0.8083,
+      "step": 11897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1556153070533443e-05,
+      "loss": 0.0012,
+      "step": 11898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.155375234071158e-05,
+      "loss": 0.001,
+      "step": 11899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.155135161088971e-05,
+      "loss": 0.0009,
+      "step": 11900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1548950881067846e-05,
+      "loss": 0.0354,
+      "step": 11901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.154655015124598e-05,
+      "loss": 0.0007,
+      "step": 11902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1544149421424116e-05,
+      "loss": 0.0007,
+      "step": 11903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1541748691602248e-05,
+      "loss": 0.0008,
+      "step": 11904
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1539347961780383e-05,
+      "loss": 0.0013,
+      "step": 11905
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1536947231958515e-05,
+      "loss": 0.0013,
+      "step": 11906
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1534546502136653e-05,
+      "loss": 0.001,
+      "step": 11907
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1532145772314785e-05,
+      "loss": 0.0011,
+      "step": 11908
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.152974504249292e-05,
+      "loss": 0.0009,
+      "step": 11909
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1527344312671052e-05,
+      "loss": 0.0009,
+      "step": 11910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1524943582849187e-05,
+      "loss": 0.0011,
+      "step": 11911
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1522542853027322e-05,
+      "loss": 0.6731,
+      "step": 11912
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1520142123205454e-05,
+      "loss": 0.0013,
+      "step": 11913
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.151774139338359e-05,
+      "loss": 0.0006,
+      "step": 11914
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.151534066356172e-05,
+      "loss": 0.0004,
+      "step": 11915
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.151293993373986e-05,
+      "loss": 0.1829,
+      "step": 11916
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.151053920391799e-05,
+      "loss": 0.0004,
+      "step": 11917
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1508138474096127e-05,
+      "loss": 1.0469,
+      "step": 11918
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.150573774427426e-05,
+      "loss": 0.0012,
+      "step": 11919
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1503337014452394e-05,
+      "loss": 0.1562,
+      "step": 11920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.150093628463053e-05,
+      "loss": 0.0011,
+      "step": 11921
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1498535554808664e-05,
+      "loss": 0.0004,
+      "step": 11922
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1496134824986796e-05,
+      "loss": 0.0008,
+      "step": 11923
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.149373409516493e-05,
+      "loss": 0.0007,
+      "step": 11924
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1491333365343066e-05,
+      "loss": 0.0005,
+      "step": 11925
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.14889326355212e-05,
+      "loss": 0.0011,
+      "step": 11926
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1486531905699333e-05,
+      "loss": 0.7725,
+      "step": 11927
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.148413117587747e-05,
+      "loss": 0.0054,
+      "step": 11928
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.14817304460556e-05,
+      "loss": 0.0013,
+      "step": 11929
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1479329716233735e-05,
+      "loss": 0.0006,
+      "step": 11930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.147692898641187e-05,
+      "loss": 0.0008,
+      "step": 11931
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1474528256590002e-05,
+      "loss": 0.001,
+      "step": 11932
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1472127526768138e-05,
+      "loss": 0.0374,
+      "step": 11933
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1469726796946273e-05,
+      "loss": 0.0013,
+      "step": 11934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1467326067124408e-05,
+      "loss": 0.0009,
+      "step": 11935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.146492533730254e-05,
+      "loss": 0.0009,
+      "step": 11936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1462524607480675e-05,
+      "loss": 0.0012,
+      "step": 11937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1460123877658807e-05,
+      "loss": 0.0009,
+      "step": 11938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1457723147836945e-05,
+      "loss": 0.0007,
+      "step": 11939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1455322418015077e-05,
+      "loss": 0.0007,
+      "step": 11940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1452921688193212e-05,
+      "loss": 0.0011,
+      "step": 11941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1450520958371344e-05,
+      "loss": 0.0009,
+      "step": 11942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1448120228549483e-05,
+      "loss": 0.0034,
+      "step": 11943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1445719498727615e-05,
+      "loss": 0.0008,
+      "step": 11944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.144331876890575e-05,
+      "loss": 0.3213,
+      "step": 11945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.144091803908388e-05,
+      "loss": 0.0013,
+      "step": 11946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1438517309262017e-05,
+      "loss": 0.6084,
+      "step": 11947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1436116579440152e-05,
+      "loss": 0.0012,
+      "step": 11948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1433715849618287e-05,
+      "loss": 0.0007,
+      "step": 11949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.143131511979642e-05,
+      "loss": 1.0324,
+      "step": 11950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1428914389974554e-05,
+      "loss": 0.0008,
+      "step": 11951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1426513660152686e-05,
+      "loss": 0.0013,
+      "step": 11952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.142411293033082e-05,
+      "loss": 0.0012,
+      "step": 11953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1421712200508956e-05,
+      "loss": 0.0013,
+      "step": 11954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1419311470687088e-05,
+      "loss": 0.2829,
+      "step": 11955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1416910740865223e-05,
+      "loss": 0.0009,
+      "step": 11956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.141451001104336e-05,
+      "loss": 0.0014,
+      "step": 11957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1412109281221494e-05,
+      "loss": 0.0009,
+      "step": 11958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1409708551399625e-05,
+      "loss": 0.0012,
+      "step": 11959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.140730782157776e-05,
+      "loss": 0.0008,
+      "step": 11960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1404907091755892e-05,
+      "loss": 0.0012,
+      "step": 11961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.140250636193403e-05,
+      "loss": 0.0012,
+      "step": 11962
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1400105632112163e-05,
+      "loss": 0.8221,
+      "step": 11963
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1397704902290298e-05,
+      "loss": 0.0009,
+      "step": 11964
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.139530417246843e-05,
+      "loss": 0.0007,
+      "step": 11965
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.139290344264657e-05,
+      "loss": 0.2456,
+      "step": 11966
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.13905027128247e-05,
+      "loss": 0.0015,
+      "step": 11967
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1388101983002835e-05,
+      "loss": 0.0008,
+      "step": 11968
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1385701253180967e-05,
+      "loss": 0.0013,
+      "step": 11969
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1383300523359102e-05,
+      "loss": 0.0007,
+      "step": 11970
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1380899793537238e-05,
+      "loss": 0.0022,
+      "step": 11971
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.137849906371537e-05,
+      "loss": 0.0015,
+      "step": 11972
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1376098333893505e-05,
+      "loss": 0.0011,
+      "step": 11973
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1373697604071636e-05,
+      "loss": 0.0013,
+      "step": 11974
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1371296874249775e-05,
+      "loss": 0.0011,
+      "step": 11975
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1368896144427907e-05,
+      "loss": 0.0012,
+      "step": 11976
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1366495414606042e-05,
+      "loss": 0.0009,
+      "step": 11977
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1364094684784174e-05,
+      "loss": 0.0015,
+      "step": 11978
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.136169395496231e-05,
+      "loss": 0.0012,
+      "step": 11979
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1359293225140444e-05,
+      "loss": 0.7891,
+      "step": 11980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.135689249531858e-05,
+      "loss": 0.0008,
+      "step": 11981
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.135449176549671e-05,
+      "loss": 0.0014,
+      "step": 11982
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1352091035674846e-05,
+      "loss": 0.7306,
+      "step": 11983
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1349690305852978e-05,
+      "loss": 0.0014,
+      "step": 11984
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1347289576031117e-05,
+      "loss": 0.0007,
+      "step": 11985
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.134488884620925e-05,
+      "loss": 0.0019,
+      "step": 11986
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1342488116387384e-05,
+      "loss": 0.0012,
+      "step": 11987
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1340087386565515e-05,
+      "loss": 0.0008,
+      "step": 11988
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.133768665674365e-05,
+      "loss": 0.0013,
+      "step": 11989
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1335285926921786e-05,
+      "loss": 0.0008,
+      "step": 11990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.133288519709992e-05,
+      "loss": 0.0016,
+      "step": 11991
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1330484467278053e-05,
+      "loss": 0.0013,
+      "step": 11992
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1328083737456188e-05,
+      "loss": 0.0014,
+      "step": 11993
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1325683007634323e-05,
+      "loss": 0.0012,
+      "step": 11994
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1323282277812455e-05,
+      "loss": 0.001,
+      "step": 11995
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.132088154799059e-05,
+      "loss": 0.0015,
+      "step": 11996
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1318480818168722e-05,
+      "loss": 0.1309,
+      "step": 11997
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.131608008834686e-05,
+      "loss": 0.0007,
+      "step": 11998
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1313679358524992e-05,
+      "loss": 0.0013,
+      "step": 11999
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1311278628703128e-05,
+      "loss": 0.0005,
+      "step": 12000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.130887789888126e-05,
+      "loss": 0.0022,
+      "step": 12001
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1306477169059395e-05,
+      "loss": 0.0012,
+      "step": 12002
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.130407643923753e-05,
+      "loss": 0.0021,
+      "step": 12003
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1301675709415665e-05,
+      "loss": 0.0007,
+      "step": 12004
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1299274979593797e-05,
+      "loss": 0.0013,
+      "step": 12005
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1296874249771932e-05,
+      "loss": 0.0305,
+      "step": 12006
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1294473519950064e-05,
+      "loss": 0.0012,
+      "step": 12007
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1292072790128202e-05,
+      "loss": 0.0017,
+      "step": 12008
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1289672060306334e-05,
+      "loss": 0.0578,
+      "step": 12009
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.128727133048447e-05,
+      "loss": 0.0007,
+      "step": 12010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.12848706006626e-05,
+      "loss": 0.0013,
+      "step": 12011
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1282469870840736e-05,
+      "loss": 0.0015,
+      "step": 12012
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.128006914101887e-05,
+      "loss": 0.0013,
+      "step": 12013
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1277668411197003e-05,
+      "loss": 0.0013,
+      "step": 12014
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.127526768137514e-05,
+      "loss": 0.0015,
+      "step": 12015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1272866951553274e-05,
+      "loss": 0.0005,
+      "step": 12016
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.127046622173141e-05,
+      "loss": 0.0016,
+      "step": 12017
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.126806549190954e-05,
+      "loss": 0.0345,
+      "step": 12018
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1265664762087676e-05,
+      "loss": 0.0014,
+      "step": 12019
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1263264032265808e-05,
+      "loss": 0.8514,
+      "step": 12020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1260863302443946e-05,
+      "loss": 0.7865,
+      "step": 12021
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1258462572622078e-05,
+      "loss": 0.0011,
+      "step": 12022
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1256061842800213e-05,
+      "loss": 0.0004,
+      "step": 12023
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1253661112978345e-05,
+      "loss": 0.0017,
+      "step": 12024
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.125126038315648e-05,
+      "loss": 0.0011,
+      "step": 12025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1248859653334615e-05,
+      "loss": 0.0018,
+      "step": 12026
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.124645892351275e-05,
+      "loss": 0.0009,
+      "step": 12027
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1244058193690882e-05,
+      "loss": 0.0021,
+      "step": 12028
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1241657463869017e-05,
+      "loss": 0.0013,
+      "step": 12029
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1239256734047153e-05,
+      "loss": 0.0013,
+      "step": 12030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1236856004225288e-05,
+      "loss": 0.0009,
+      "step": 12031
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.123445527440342e-05,
+      "loss": 0.0235,
+      "step": 12032
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1232054544581555e-05,
+      "loss": 0.0014,
+      "step": 12033
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1229653814759687e-05,
+      "loss": 0.0011,
+      "step": 12034
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1227253084937822e-05,
+      "loss": 0.0052,
+      "step": 12035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1224852355115957e-05,
+      "loss": 0.0024,
+      "step": 12036
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.122245162529409e-05,
+      "loss": 0.0016,
+      "step": 12037
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1220050895472224e-05,
+      "loss": 0.0014,
+      "step": 12038
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1217650165650356e-05,
+      "loss": 0.0008,
+      "step": 12039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1215249435828494e-05,
+      "loss": 0.0015,
+      "step": 12040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1212848706006626e-05,
+      "loss": 0.0012,
+      "step": 12041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.121044797618476e-05,
+      "loss": 0.002,
+      "step": 12042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1208047246362893e-05,
+      "loss": 0.0011,
+      "step": 12043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1205646516541032e-05,
+      "loss": 0.0006,
+      "step": 12044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1203245786719164e-05,
+      "loss": 0.0017,
+      "step": 12045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.12008450568973e-05,
+      "loss": 0.0013,
+      "step": 12046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.119844432707543e-05,
+      "loss": 0.001,
+      "step": 12047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1196043597253566e-05,
+      "loss": 0.002,
+      "step": 12048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.11936428674317e-05,
+      "loss": 0.446,
+      "step": 12049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1191242137609836e-05,
+      "loss": 0.0009,
+      "step": 12050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1188841407787968e-05,
+      "loss": 0.0013,
+      "step": 12051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1186440677966103e-05,
+      "loss": 0.0011,
+      "step": 12052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1184039948144238e-05,
+      "loss": 0.0014,
+      "step": 12053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.118163921832237e-05,
+      "loss": 0.0008,
+      "step": 12054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1179238488500505e-05,
+      "loss": 0.001,
+      "step": 12055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1176837758678637e-05,
+      "loss": 0.0012,
+      "step": 12056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1174437028856772e-05,
+      "loss": 0.0011,
+      "step": 12057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1172036299034907e-05,
+      "loss": 0.0012,
+      "step": 12058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1169635569213043e-05,
+      "loss": 0.0011,
+      "step": 12059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1167234839391174e-05,
+      "loss": 0.0009,
+      "step": 12060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.116483410956931e-05,
+      "loss": 0.0014,
+      "step": 12061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1162433379747445e-05,
+      "loss": 0.0015,
+      "step": 12062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.116003264992558e-05,
+      "loss": 0.283,
+      "step": 12063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1157631920103712e-05,
+      "loss": 0.0011,
+      "step": 12064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1155231190281847e-05,
+      "loss": 0.0006,
+      "step": 12065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.115283046045998e-05,
+      "loss": 0.0017,
+      "step": 12066
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1150429730638117e-05,
+      "loss": 0.0013,
+      "step": 12067
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.114802900081625e-05,
+      "loss": 0.3828,
+      "step": 12068
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1145628270994384e-05,
+      "loss": 0.0008,
+      "step": 12069
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1143227541172516e-05,
+      "loss": 0.0013,
+      "step": 12070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.114082681135065e-05,
+      "loss": 0.0018,
+      "step": 12071
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1138426081528787e-05,
+      "loss": 0.0015,
+      "step": 12072
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1136025351706922e-05,
+      "loss": 0.0016,
+      "step": 12073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1133624621885054e-05,
+      "loss": 0.0005,
+      "step": 12074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.113122389206319e-05,
+      "loss": 0.001,
+      "step": 12075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1128823162241324e-05,
+      "loss": 0.0011,
+      "step": 12076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1126422432419456e-05,
+      "loss": 0.0016,
+      "step": 12077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.112402170259759e-05,
+      "loss": 0.0009,
+      "step": 12078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1121620972775723e-05,
+      "loss": 0.1823,
+      "step": 12079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1119220242953858e-05,
+      "loss": 0.0009,
+      "step": 12080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1116819513131993e-05,
+      "loss": 0.0012,
+      "step": 12081
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1114418783310128e-05,
+      "loss": 0.001,
+      "step": 12082
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.111201805348826e-05,
+      "loss": 0.0013,
+      "step": 12083
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1109617323666395e-05,
+      "loss": 0.0008,
+      "step": 12084
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.110721659384453e-05,
+      "loss": 0.0012,
+      "step": 12085
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1104815864022666e-05,
+      "loss": 0.0007,
+      "step": 12086
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1102415134200797e-05,
+      "loss": 0.001,
+      "step": 12087
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1100014404378933e-05,
+      "loss": 0.0011,
+      "step": 12088
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1097613674557064e-05,
+      "loss": 0.001,
+      "step": 12089
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1095212944735203e-05,
+      "loss": 0.1585,
+      "step": 12090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1092812214913335e-05,
+      "loss": 0.0015,
+      "step": 12091
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.109041148509147e-05,
+      "loss": 0.001,
+      "step": 12092
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1088010755269602e-05,
+      "loss": 0.0009,
+      "step": 12093
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1085610025447737e-05,
+      "loss": 0.0009,
+      "step": 12094
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1083209295625872e-05,
+      "loss": 0.001,
+      "step": 12095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1080808565804004e-05,
+      "loss": 0.001,
+      "step": 12096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107840783598214e-05,
+      "loss": 0.0009,
+      "step": 12097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107600710616027e-05,
+      "loss": 0.001,
+      "step": 12098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107360637633841e-05,
+      "loss": 0.0008,
+      "step": 12099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.107120564651654e-05,
+      "loss": 0.0456,
+      "step": 12100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1068804916694677e-05,
+      "loss": 0.0014,
+      "step": 12101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.106640418687281e-05,
+      "loss": 0.0217,
+      "step": 12102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1064003457050944e-05,
+      "loss": 0.0009,
+      "step": 12103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.106160272722908e-05,
+      "loss": 0.001,
+      "step": 12104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1059201997407214e-05,
+      "loss": 0.001,
+      "step": 12105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1056801267585346e-05,
+      "loss": 0.0007,
+      "step": 12106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.105440053776348e-05,
+      "loss": 0.0009,
+      "step": 12107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1051999807941616e-05,
+      "loss": 0.0007,
+      "step": 12108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.104959907811975e-05,
+      "loss": 0.0008,
+      "step": 12109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1047198348297883e-05,
+      "loss": 0.0008,
+      "step": 12110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1044797618476018e-05,
+      "loss": 0.0005,
+      "step": 12111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.104239688865415e-05,
+      "loss": 0.001,
+      "step": 12112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1039996158832285e-05,
+      "loss": 0.0012,
+      "step": 12113
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.103759542901042e-05,
+      "loss": 0.0004,
+      "step": 12114
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1035194699188556e-05,
+      "loss": 0.0006,
+      "step": 12115
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1032793969366687e-05,
+      "loss": 0.001,
+      "step": 12116
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1030393239544823e-05,
+      "loss": 0.0008,
+      "step": 12117
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1027992509722958e-05,
+      "loss": 0.0063,
+      "step": 12118
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.102559177990109e-05,
+      "loss": 0.0009,
+      "step": 12119
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1023191050079225e-05,
+      "loss": 0.0008,
+      "step": 12120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1020790320257357e-05,
+      "loss": 0.001,
+      "step": 12121
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1018389590435495e-05,
+      "loss": 0.0035,
+      "step": 12122
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1015988860613627e-05,
+      "loss": 0.0007,
+      "step": 12123
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1013588130791762e-05,
+      "loss": 0.0011,
+      "step": 12124
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1011187400969894e-05,
+      "loss": 0.0007,
+      "step": 12125
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1008786671148033e-05,
+      "loss": 0.0007,
+      "step": 12126
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1006385941326164e-05,
+      "loss": 0.0029,
+      "step": 12127
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.10039852115043e-05,
+      "loss": 0.0008,
+      "step": 12128
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.100158448168243e-05,
+      "loss": 0.5378,
+      "step": 12129
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0999183751860566e-05,
+      "loss": 0.001,
+      "step": 12130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.09967830220387e-05,
+      "loss": 0.0009,
+      "step": 12131
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0994382292216837e-05,
+      "loss": 0.0008,
+      "step": 12132
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.099198156239497e-05,
+      "loss": 0.0006,
+      "step": 12133
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0989580832573104e-05,
+      "loss": 0.0006,
+      "step": 12134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0987180102751236e-05,
+      "loss": 0.0008,
+      "step": 12135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.098477937292937e-05,
+      "loss": 0.0009,
+      "step": 12136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0982378643107506e-05,
+      "loss": 0.0007,
+      "step": 12137
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0979977913285638e-05,
+      "loss": 0.0007,
+      "step": 12138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0977577183463773e-05,
+      "loss": 0.0007,
+      "step": 12139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0975176453641908e-05,
+      "loss": 0.0005,
+      "step": 12140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0972775723820043e-05,
+      "loss": 0.0006,
+      "step": 12141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0970374993998175e-05,
+      "loss": 0.0016,
+      "step": 12142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.096797426417631e-05,
+      "loss": 0.0008,
+      "step": 12143
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0965573534354442e-05,
+      "loss": 0.0041,
+      "step": 12144
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.096317280453258e-05,
+      "loss": 0.0008,
+      "step": 12145
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0960772074710713e-05,
+      "loss": 0.0007,
+      "step": 12146
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0958371344888848e-05,
+      "loss": 0.0009,
+      "step": 12147
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.095597061506698e-05,
+      "loss": 0.0006,
+      "step": 12148
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0953569885245118e-05,
+      "loss": 0.0007,
+      "step": 12149
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.095116915542325e-05,
+      "loss": 0.0039,
+      "step": 12150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0948768425601385e-05,
+      "loss": 0.0006,
+      "step": 12151
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0946367695779517e-05,
+      "loss": 0.0005,
+      "step": 12152
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0943966965957652e-05,
+      "loss": 0.0007,
+      "step": 12153
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0941566236135787e-05,
+      "loss": 0.0009,
+      "step": 12154
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0939165506313922e-05,
+      "loss": 0.0009,
+      "step": 12155
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0936764776492054e-05,
+      "loss": 0.0007,
+      "step": 12156
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.093436404667019e-05,
+      "loss": 0.0008,
+      "step": 12157
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0931963316848325e-05,
+      "loss": 0.0008,
+      "step": 12158
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0929562587026456e-05,
+      "loss": 0.0004,
+      "step": 12159
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.092716185720459e-05,
+      "loss": 0.8647,
+      "step": 12160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0924761127382723e-05,
+      "loss": 0.0032,
+      "step": 12161
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.092236039756086e-05,
+      "loss": 0.0008,
+      "step": 12162
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0919959667738994e-05,
+      "loss": 0.0008,
+      "step": 12163
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.091755893791713e-05,
+      "loss": 0.0005,
+      "step": 12164
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.091515820809526e-05,
+      "loss": 0.0009,
+      "step": 12165
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0912757478273396e-05,
+      "loss": 0.0004,
+      "step": 12166
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0910356748451528e-05,
+      "loss": 0.0006,
+      "step": 12167
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0907956018629666e-05,
+      "loss": 0.0007,
+      "step": 12168
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0905555288807798e-05,
+      "loss": 0.0011,
+      "step": 12169
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0903154558985933e-05,
+      "loss": 0.0006,
+      "step": 12170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0900753829164065e-05,
+      "loss": 0.0007,
+      "step": 12171
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0898353099342204e-05,
+      "loss": 0.0005,
+      "step": 12172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0895952369520336e-05,
+      "loss": 0.0005,
+      "step": 12173
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.089355163969847e-05,
+      "loss": 0.0017,
+      "step": 12174
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0891150909876603e-05,
+      "loss": 0.0007,
+      "step": 12175
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0888750180054738e-05,
+      "loss": 0.0005,
+      "step": 12176
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0886349450232873e-05,
+      "loss": 0.0008,
+      "step": 12177
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0883948720411005e-05,
+      "loss": 0.0012,
+      "step": 12178
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.088154799058914e-05,
+      "loss": 0.0004,
+      "step": 12179
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0879147260767272e-05,
+      "loss": 0.0006,
+      "step": 12180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.087674653094541e-05,
+      "loss": 0.0005,
+      "step": 12181
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0874345801123542e-05,
+      "loss": 0.0004,
+      "step": 12182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0871945071301677e-05,
+      "loss": 0.0005,
+      "step": 12183
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.086954434147981e-05,
+      "loss": 0.001,
+      "step": 12184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0867143611657944e-05,
+      "loss": 0.0005,
+      "step": 12185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.086474288183608e-05,
+      "loss": 0.0006,
+      "step": 12186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0862342152014215e-05,
+      "loss": 0.0007,
+      "step": 12187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0859941422192346e-05,
+      "loss": 0.0006,
+      "step": 12188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.085754069237048e-05,
+      "loss": 0.0005,
+      "step": 12189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0855139962548617e-05,
+      "loss": 0.0005,
+      "step": 12190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0852739232726752e-05,
+      "loss": 0.0005,
+      "step": 12191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0850338502904884e-05,
+      "loss": 0.0005,
+      "step": 12192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.084793777308302e-05,
+      "loss": 0.0006,
+      "step": 12193
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.084553704326115e-05,
+      "loss": 0.0003,
+      "step": 12194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0843136313439286e-05,
+      "loss": 0.0008,
+      "step": 12195
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.084073558361742e-05,
+      "loss": 0.0031,
+      "step": 12196
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0838334853795556e-05,
+      "loss": 0.0006,
+      "step": 12197
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0835934123973688e-05,
+      "loss": 0.5182,
+      "step": 12198
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0833533394151823e-05,
+      "loss": 0.0006,
+      "step": 12199
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.083113266432996e-05,
+      "loss": 0.0007,
+      "step": 12200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.082873193450809e-05,
+      "loss": 0.0006,
+      "step": 12201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0826331204686226e-05,
+      "loss": 0.0004,
+      "step": 12202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0823930474864357e-05,
+      "loss": 0.0005,
+      "step": 12203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0821529745042496e-05,
+      "loss": 0.0008,
+      "step": 12204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0819129015220628e-05,
+      "loss": 0.0006,
+      "step": 12205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0816728285398763e-05,
+      "loss": 0.0005,
+      "step": 12206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0814327555576895e-05,
+      "loss": 0.6741,
+      "step": 12207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.081192682575503e-05,
+      "loss": 0.0004,
+      "step": 12208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0809526095933165e-05,
+      "loss": 0.004,
+      "step": 12209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.08071253661113e-05,
+      "loss": 0.0006,
+      "step": 12210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0804724636289432e-05,
+      "loss": 0.0007,
+      "step": 12211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0802323906467567e-05,
+      "loss": 0.0005,
+      "step": 12212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0799923176645702e-05,
+      "loss": 0.0042,
+      "step": 12213
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0797522446823838e-05,
+      "loss": 0.0007,
+      "step": 12214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.079512171700197e-05,
+      "loss": 0.0004,
+      "step": 12215
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0792720987180105e-05,
+      "loss": 0.0004,
+      "step": 12216
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0790320257358236e-05,
+      "loss": 0.4763,
+      "step": 12217
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.078791952753637e-05,
+      "loss": 0.0007,
+      "step": 12218
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0785518797714507e-05,
+      "loss": 0.0005,
+      "step": 12219
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.078311806789264e-05,
+      "loss": 0.0006,
+      "step": 12220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0780717338070774e-05,
+      "loss": 0.0005,
+      "step": 12221
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.077831660824891e-05,
+      "loss": 0.0004,
+      "step": 12222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0775915878427044e-05,
+      "loss": 0.0044,
+      "step": 12223
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0773515148605176e-05,
+      "loss": 0.8511,
+      "step": 12224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.077111441878331e-05,
+      "loss": 0.0007,
+      "step": 12225
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0768713688961443e-05,
+      "loss": 0.3129,
+      "step": 12226
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.076631295913958e-05,
+      "loss": 1.058,
+      "step": 12227
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0763912229317713e-05,
+      "loss": 0.0009,
+      "step": 12228
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.076151149949585e-05,
+      "loss": 0.4024,
+      "step": 12229
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.075911076967398e-05,
+      "loss": 0.0004,
+      "step": 12230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0756710039852115e-05,
+      "loss": 0.0008,
+      "step": 12231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.075430931003025e-05,
+      "loss": 0.0008,
+      "step": 12232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0751908580208386e-05,
+      "loss": 0.0008,
+      "step": 12233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0749507850386518e-05,
+      "loss": 0.0009,
+      "step": 12234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0747107120564653e-05,
+      "loss": 0.0006,
+      "step": 12235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0744706390742788e-05,
+      "loss": 0.0012,
+      "step": 12236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.074230566092092e-05,
+      "loss": 0.0012,
+      "step": 12237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0739904931099055e-05,
+      "loss": 0.0009,
+      "step": 12238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.073750420127719e-05,
+      "loss": 0.001,
+      "step": 12239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0735103471455322e-05,
+      "loss": 0.0008,
+      "step": 12240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0732702741633457e-05,
+      "loss": 0.0008,
+      "step": 12241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0730302011811592e-05,
+      "loss": 0.0008,
+      "step": 12242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0727901281989724e-05,
+      "loss": 0.8529,
+      "step": 12243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.072550055216786e-05,
+      "loss": 0.002,
+      "step": 12244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0723099822345995e-05,
+      "loss": 0.001,
+      "step": 12245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.072069909252413e-05,
+      "loss": 0.001,
+      "step": 12246
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.071829836270226e-05,
+      "loss": 0.0008,
+      "step": 12247
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0715897632880397e-05,
+      "loss": 0.0008,
+      "step": 12248
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.071349690305853e-05,
+      "loss": 0.0134,
+      "step": 12249
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0711096173236667e-05,
+      "loss": 0.001,
+      "step": 12250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.07086954434148e-05,
+      "loss": 0.0012,
+      "step": 12251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0706294713592934e-05,
+      "loss": 0.8147,
+      "step": 12252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0703893983771066e-05,
+      "loss": 0.6496,
+      "step": 12253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0701493253949205e-05,
+      "loss": 0.0209,
+      "step": 12254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0699092524127336e-05,
+      "loss": 0.001,
+      "step": 12255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.069669179430547e-05,
+      "loss": 0.2547,
+      "step": 12256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0694291064483603e-05,
+      "loss": 0.0009,
+      "step": 12257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.069189033466174e-05,
+      "loss": 0.2788,
+      "step": 12258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0689489604839874e-05,
+      "loss": 0.0015,
+      "step": 12259
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0687088875018005e-05,
+      "loss": 0.0025,
+      "step": 12260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.068468814519614e-05,
+      "loss": 0.0013,
+      "step": 12261
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0682287415374272e-05,
+      "loss": 0.0034,
+      "step": 12262
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0679886685552408e-05,
+      "loss": 0.0011,
+      "step": 12263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0677485955730543e-05,
+      "loss": 0.0955,
+      "step": 12264
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0675085225908678e-05,
+      "loss": 0.0014,
+      "step": 12265
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.067268449608681e-05,
+      "loss": 0.0023,
+      "step": 12266
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0670283766264945e-05,
+      "loss": 0.0015,
+      "step": 12267
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.066788303644308e-05,
+      "loss": 0.0011,
+      "step": 12268
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0665482306621215e-05,
+      "loss": 0.0013,
+      "step": 12269
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0663081576799347e-05,
+      "loss": 0.001,
+      "step": 12270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0660680846977482e-05,
+      "loss": 0.0008,
+      "step": 12271
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0658280117155614e-05,
+      "loss": 0.0009,
+      "step": 12272
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0655879387333753e-05,
+      "loss": 0.0424,
+      "step": 12273
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0653478657511885e-05,
+      "loss": 0.1525,
+      "step": 12274
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.065107792769002e-05,
+      "loss": 0.0011,
+      "step": 12275
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.064867719786815e-05,
+      "loss": 0.0009,
+      "step": 12276
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0646276468046287e-05,
+      "loss": 0.0033,
+      "step": 12277
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0643875738224422e-05,
+      "loss": 0.0005,
+      "step": 12278
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0641475008402557e-05,
+      "loss": 0.0009,
+      "step": 12279
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.063907427858069e-05,
+      "loss": 0.0011,
+      "step": 12280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0636673548758824e-05,
+      "loss": 0.0017,
+      "step": 12281
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.063427281893696e-05,
+      "loss": 0.0012,
+      "step": 12282
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.063187208911509e-05,
+      "loss": 0.001,
+      "step": 12283
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0629471359293226e-05,
+      "loss": 0.001,
+      "step": 12284
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0627070629471358e-05,
+      "loss": 0.001,
+      "step": 12285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0624669899649497e-05,
+      "loss": 0.0007,
+      "step": 12286
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.062226916982763e-05,
+      "loss": 0.0012,
+      "step": 12287
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0619868440005764e-05,
+      "loss": 0.001,
+      "step": 12288
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0617467710183895e-05,
+      "loss": 0.001,
+      "step": 12289
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.061506698036203e-05,
+      "loss": 0.0012,
+      "step": 12290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0612666250540166e-05,
+      "loss": 0.4722,
+      "step": 12291
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.06102655207183e-05,
+      "loss": 0.0008,
+      "step": 12292
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0607864790896433e-05,
+      "loss": 0.0008,
+      "step": 12293
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0605464061074568e-05,
+      "loss": 0.0017,
+      "step": 12294
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.06030633312527e-05,
+      "loss": 0.0011,
+      "step": 12295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.060066260143084e-05,
+      "loss": 0.0006,
+      "step": 12296
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.059826187160897e-05,
+      "loss": 0.001,
+      "step": 12297
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0595861141787105e-05,
+      "loss": 0.0357,
+      "step": 12298
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0593460411965237e-05,
+      "loss": 0.0006,
+      "step": 12299
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0591059682143372e-05,
+      "loss": 0.0413,
+      "step": 12300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0588658952321508e-05,
+      "loss": 0.0313,
+      "step": 12301
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.058625822249964e-05,
+      "loss": 0.001,
+      "step": 12302
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0583857492677775e-05,
+      "loss": 0.5863,
+      "step": 12303
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0581456762855906e-05,
+      "loss": 0.0007,
+      "step": 12304
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0579056033034045e-05,
+      "loss": 0.0011,
+      "step": 12305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0576655303212177e-05,
+      "loss": 0.0012,
+      "step": 12306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0574254573390312e-05,
+      "loss": 0.0009,
+      "step": 12307
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0571853843568444e-05,
+      "loss": 0.001,
+      "step": 12308
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0569453113746582e-05,
+      "loss": 0.0009,
+      "step": 12309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0567052383924714e-05,
+      "loss": 0.001,
+      "step": 12310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.056465165410285e-05,
+      "loss": 0.0008,
+      "step": 12311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.056225092428098e-05,
+      "loss": 0.0007,
+      "step": 12312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0559850194459116e-05,
+      "loss": 0.0007,
+      "step": 12313
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.055744946463725e-05,
+      "loss": 0.001,
+      "step": 12314
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0555048734815387e-05,
+      "loss": 0.3875,
+      "step": 12315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.055264800499352e-05,
+      "loss": 0.0011,
+      "step": 12316
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0550247275171654e-05,
+      "loss": 0.0008,
+      "step": 12317
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.054784654534979e-05,
+      "loss": 0.0007,
+      "step": 12318
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.054544581552792e-05,
+      "loss": 0.0006,
+      "step": 12319
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0543045085706056e-05,
+      "loss": 0.0011,
+      "step": 12320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.054064435588419e-05,
+      "loss": 0.0005,
+      "step": 12321
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0538243626062323e-05,
+      "loss": 0.0009,
+      "step": 12322
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0535842896240458e-05,
+      "loss": 0.0008,
+      "step": 12323
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0533442166418593e-05,
+      "loss": 0.985,
+      "step": 12324
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0531041436596725e-05,
+      "loss": 0.0008,
+      "step": 12325
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.052864070677486e-05,
+      "loss": 0.0008,
+      "step": 12326
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0526239976952992e-05,
+      "loss": 0.0007,
+      "step": 12327
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.052383924713113e-05,
+      "loss": 0.0009,
+      "step": 12328
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0521438517309262e-05,
+      "loss": 0.0008,
+      "step": 12329
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0519037787487398e-05,
+      "loss": 0.0009,
+      "step": 12330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.051663705766553e-05,
+      "loss": 0.001,
+      "step": 12331
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0514236327843668e-05,
+      "loss": 0.0008,
+      "step": 12332
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.05118355980218e-05,
+      "loss": 0.0009,
+      "step": 12333
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0509434868199935e-05,
+      "loss": 0.0006,
+      "step": 12334
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0507034138378067e-05,
+      "loss": 0.0007,
+      "step": 12335
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0504633408556202e-05,
+      "loss": 0.0009,
+      "step": 12336
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0502232678734337e-05,
+      "loss": 0.0009,
+      "step": 12337
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0499831948912472e-05,
+      "loss": 0.001,
+      "step": 12338
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0497431219090604e-05,
+      "loss": 0.001,
+      "step": 12339
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.049503048926874e-05,
+      "loss": 0.0009,
+      "step": 12340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0492629759446874e-05,
+      "loss": 0.0008,
+      "step": 12341
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0490229029625006e-05,
+      "loss": 0.001,
+      "step": 12342
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.048782829980314e-05,
+      "loss": 0.0007,
+      "step": 12343
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0485427569981273e-05,
+      "loss": 0.0006,
+      "step": 12344
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.048302684015941e-05,
+      "loss": 0.0008,
+      "step": 12345
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0480626110337544e-05,
+      "loss": 0.001,
+      "step": 12346
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.047822538051568e-05,
+      "loss": 0.0009,
+      "step": 12347
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.047582465069381e-05,
+      "loss": 0.0008,
+      "step": 12348
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0473423920871946e-05,
+      "loss": 0.001,
+      "step": 12349
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0471023191050078e-05,
+      "loss": 0.0007,
+      "step": 12350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0468622461228216e-05,
+      "loss": 0.0009,
+      "step": 12351
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0466221731406348e-05,
+      "loss": 0.0008,
+      "step": 12352
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0463821001584483e-05,
+      "loss": 0.0007,
+      "step": 12353
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0461420271762615e-05,
+      "loss": 0.0007,
+      "step": 12354
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0459019541940754e-05,
+      "loss": 0.0007,
+      "step": 12355
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0456618812118885e-05,
+      "loss": 0.0008,
+      "step": 12356
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.045421808229702e-05,
+      "loss": 0.0006,
+      "step": 12357
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0451817352475152e-05,
+      "loss": 0.0009,
+      "step": 12358
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0449416622653287e-05,
+      "loss": 0.0009,
+      "step": 12359
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0447015892831423e-05,
+      "loss": 0.0011,
+      "step": 12360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0444615163009554e-05,
+      "loss": 0.0008,
+      "step": 12361
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.044221443318769e-05,
+      "loss": 0.0009,
+      "step": 12362
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0439813703365825e-05,
+      "loss": 0.0007,
+      "step": 12363
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.043741297354396e-05,
+      "loss": 0.9371,
+      "step": 12364
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0435012243722092e-05,
+      "loss": 0.0007,
+      "step": 12365
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0432611513900227e-05,
+      "loss": 0.0007,
+      "step": 12366
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.043021078407836e-05,
+      "loss": 0.0005,
+      "step": 12367
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0427810054256494e-05,
+      "loss": 0.0016,
+      "step": 12368
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.042540932443463e-05,
+      "loss": 0.0008,
+      "step": 12369
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0423008594612764e-05,
+      "loss": 0.0007,
+      "step": 12370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0420607864790896e-05,
+      "loss": 0.001,
+      "step": 12371
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.041820713496903e-05,
+      "loss": 0.0008,
+      "step": 12372
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0415806405147167e-05,
+      "loss": 0.0009,
+      "step": 12373
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0413405675325302e-05,
+      "loss": 0.0006,
+      "step": 12374
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0411004945503434e-05,
+      "loss": 0.0007,
+      "step": 12375
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.040860421568157e-05,
+      "loss": 0.0005,
+      "step": 12376
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.04062034858597e-05,
+      "loss": 0.0006,
+      "step": 12377
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.040380275603784e-05,
+      "loss": 0.0006,
+      "step": 12378
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.040140202621597e-05,
+      "loss": 0.0005,
+      "step": 12379
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0399001296394106e-05,
+      "loss": 0.0005,
+      "step": 12380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0396600566572238e-05,
+      "loss": 0.0006,
+      "step": 12381
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0394199836750373e-05,
+      "loss": 0.0006,
+      "step": 12382
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0391799106928508e-05,
+      "loss": 0.0005,
+      "step": 12383
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.038939837710664e-05,
+      "loss": 0.0007,
+      "step": 12384
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0386997647284775e-05,
+      "loss": 0.0006,
+      "step": 12385
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0384596917462907e-05,
+      "loss": 0.0006,
+      "step": 12386
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0382196187641046e-05,
+      "loss": 0.0005,
+      "step": 12387
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0379795457819177e-05,
+      "loss": 0.0006,
+      "step": 12388
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0377394727997313e-05,
+      "loss": 0.0008,
+      "step": 12389
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0374993998175444e-05,
+      "loss": 0.0007,
+      "step": 12390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.037259326835358e-05,
+      "loss": 0.0006,
+      "step": 12391
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0370192538531715e-05,
+      "loss": 0.5749,
+      "step": 12392
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.036779180870985e-05,
+      "loss": 0.0005,
+      "step": 12393
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0365391078887982e-05,
+      "loss": 0.0008,
+      "step": 12394
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0362990349066117e-05,
+      "loss": 0.0006,
+      "step": 12395
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0360589619244252e-05,
+      "loss": 0.0005,
+      "step": 12396
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0358188889422387e-05,
+      "loss": 0.0004,
+      "step": 12397
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.035578815960052e-05,
+      "loss": 0.0008,
+      "step": 12398
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0353387429778654e-05,
+      "loss": 0.0006,
+      "step": 12399
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0350986699956786e-05,
+      "loss": 0.0005,
+      "step": 12400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.034858597013492e-05,
+      "loss": 0.0007,
+      "step": 12401
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0346185240313057e-05,
+      "loss": 0.0005,
+      "step": 12402
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0343784510491192e-05,
+      "loss": 0.0008,
+      "step": 12403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0341383780669324e-05,
+      "loss": 0.0006,
+      "step": 12404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.033898305084746e-05,
+      "loss": 0.0006,
+      "step": 12405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0336582321025594e-05,
+      "loss": 0.0006,
+      "step": 12406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0334181591203726e-05,
+      "loss": 0.0006,
+      "step": 12407
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.033178086138186e-05,
+      "loss": 0.9426,
+      "step": 12408
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0329380131559993e-05,
+      "loss": 0.0005,
+      "step": 12409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.032697940173813e-05,
+      "loss": 0.0005,
+      "step": 12410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0324578671916263e-05,
+      "loss": 0.0007,
+      "step": 12411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0322177942094398e-05,
+      "loss": 0.0009,
+      "step": 12412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.031977721227253e-05,
+      "loss": 0.0008,
+      "step": 12413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0317376482450665e-05,
+      "loss": 0.8935,
+      "step": 12414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.03149757526288e-05,
+      "loss": 0.0006,
+      "step": 12415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0312575022806936e-05,
+      "loss": 0.0004,
+      "step": 12416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0310174292985067e-05,
+      "loss": 0.0007,
+      "step": 12417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0307773563163203e-05,
+      "loss": 0.0006,
+      "step": 12418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0305372833341338e-05,
+      "loss": 0.0008,
+      "step": 12419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0302972103519473e-05,
+      "loss": 0.0007,
+      "step": 12420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0300571373697605e-05,
+      "loss": 0.0008,
+      "step": 12421
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.029817064387574e-05,
+      "loss": 0.0028,
+      "step": 12422
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0295769914053872e-05,
+      "loss": 0.0004,
+      "step": 12423
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0293369184232007e-05,
+      "loss": 0.0032,
+      "step": 12424
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0290968454410142e-05,
+      "loss": 0.0008,
+      "step": 12425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0288567724588274e-05,
+      "loss": 0.0007,
+      "step": 12426
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.028616699476641e-05,
+      "loss": 0.0008,
+      "step": 12427
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0283766264944544e-05,
+      "loss": 0.0006,
+      "step": 12428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.028136553512268e-05,
+      "loss": 0.0008,
+      "step": 12429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.027896480530081e-05,
+      "loss": 0.0009,
+      "step": 12430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0276564075478947e-05,
+      "loss": 0.0018,
+      "step": 12431
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.027416334565708e-05,
+      "loss": 0.0005,
+      "step": 12432
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0271762615835217e-05,
+      "loss": 0.0007,
+      "step": 12433
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.026936188601335e-05,
+      "loss": 0.0007,
+      "step": 12434
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0266961156191484e-05,
+      "loss": 0.0005,
+      "step": 12435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0264560426369616e-05,
+      "loss": 0.0006,
+      "step": 12436
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0262159696547754e-05,
+      "loss": 0.0008,
+      "step": 12437
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0259758966725886e-05,
+      "loss": 0.0005,
+      "step": 12438
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.025735823690402e-05,
+      "loss": 0.0007,
+      "step": 12439
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0254957507082153e-05,
+      "loss": 0.0009,
+      "step": 12440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0252556777260288e-05,
+      "loss": 0.0006,
+      "step": 12441
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0250156047438423e-05,
+      "loss": 0.0006,
+      "step": 12442
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0247755317616555e-05,
+      "loss": 0.8351,
+      "step": 12443
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.024535458779469e-05,
+      "loss": 0.0007,
+      "step": 12444
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0242953857972826e-05,
+      "loss": 0.0005,
+      "step": 12445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0240553128150957e-05,
+      "loss": 0.0007,
+      "step": 12446
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0238152398329093e-05,
+      "loss": 0.0006,
+      "step": 12447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0235751668507228e-05,
+      "loss": 0.0005,
+      "step": 12448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.023335093868536e-05,
+      "loss": 0.0006,
+      "step": 12449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0230950208863495e-05,
+      "loss": 0.0009,
+      "step": 12450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.022854947904163e-05,
+      "loss": 0.9808,
+      "step": 12451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0226148749219765e-05,
+      "loss": 0.0006,
+      "step": 12452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0223748019397897e-05,
+      "loss": 0.0006,
+      "step": 12453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0221347289576032e-05,
+      "loss": 0.001,
+      "step": 12454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0218946559754164e-05,
+      "loss": 0.0008,
+      "step": 12455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0216545829932303e-05,
+      "loss": 0.0007,
+      "step": 12456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0214145100110434e-05,
+      "loss": 0.0009,
+      "step": 12457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.021174437028857e-05,
+      "loss": 0.0009,
+      "step": 12458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.02093436404667e-05,
+      "loss": 0.0008,
+      "step": 12459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.020694291064484e-05,
+      "loss": 0.0005,
+      "step": 12460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.020454218082297e-05,
+      "loss": 0.0007,
+      "step": 12461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0202141451001107e-05,
+      "loss": 0.0008,
+      "step": 12462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.019974072117924e-05,
+      "loss": 0.4785,
+      "step": 12463
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0197339991357374e-05,
+      "loss": 0.001,
+      "step": 12464
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.019493926153551e-05,
+      "loss": 0.0006,
+      "step": 12465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.019253853171364e-05,
+      "loss": 0.001,
+      "step": 12466
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0190137801891776e-05,
+      "loss": 0.0121,
+      "step": 12467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0187737072069908e-05,
+      "loss": 0.0009,
+      "step": 12468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0185336342248046e-05,
+      "loss": 0.001,
+      "step": 12469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0182935612426178e-05,
+      "loss": 0.0007,
+      "step": 12470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0180534882604313e-05,
+      "loss": 0.0009,
+      "step": 12471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0178134152782445e-05,
+      "loss": 0.0008,
+      "step": 12472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.017573342296058e-05,
+      "loss": 0.0011,
+      "step": 12473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0173332693138716e-05,
+      "loss": 0.0011,
+      "step": 12474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.017093196331685e-05,
+      "loss": 0.001,
+      "step": 12475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0168531233494983e-05,
+      "loss": 0.001,
+      "step": 12476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0166130503673118e-05,
+      "loss": 0.0009,
+      "step": 12477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.016372977385125e-05,
+      "loss": 0.0008,
+      "step": 12478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0161329044029388e-05,
+      "loss": 0.001,
+      "step": 12479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.015892831420752e-05,
+      "loss": 0.0008,
+      "step": 12480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0156527584385655e-05,
+      "loss": 0.001,
+      "step": 12481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0154126854563787e-05,
+      "loss": 0.0006,
+      "step": 12482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0151726124741922e-05,
+      "loss": 0.0011,
+      "step": 12483
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0149325394920057e-05,
+      "loss": 0.0008,
+      "step": 12484
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.014692466509819e-05,
+      "loss": 0.0007,
+      "step": 12485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0144523935276324e-05,
+      "loss": 0.0006,
+      "step": 12486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.014212320545446e-05,
+      "loss": 0.9107,
+      "step": 12487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0139722475632595e-05,
+      "loss": 0.001,
+      "step": 12488
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0137321745810726e-05,
+      "loss": 0.0008,
+      "step": 12489
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.013492101598886e-05,
+      "loss": 0.0007,
+      "step": 12490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0132520286166993e-05,
+      "loss": 0.0008,
+      "step": 12491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0130119556345132e-05,
+      "loss": 0.0008,
+      "step": 12492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0127718826523264e-05,
+      "loss": 0.0007,
+      "step": 12493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.01253180967014e-05,
+      "loss": 0.0009,
+      "step": 12494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.012291736687953e-05,
+      "loss": 0.001,
+      "step": 12495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0120516637057666e-05,
+      "loss": 0.0009,
+      "step": 12496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.01181159072358e-05,
+      "loss": 0.0005,
+      "step": 12497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0115715177413936e-05,
+      "loss": 0.0069,
+      "step": 12498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0113314447592068e-05,
+      "loss": 0.3699,
+      "step": 12499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0110913717770203e-05,
+      "loss": 0.0012,
+      "step": 12500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010851298794834e-05,
+      "loss": 0.0011,
+      "step": 12501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0106112258126474e-05,
+      "loss": 0.0007,
+      "step": 12502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0103711528304606e-05,
+      "loss": 0.001,
+      "step": 12503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010131079848274e-05,
+      "loss": 0.0009,
+      "step": 12504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0098910068660873e-05,
+      "loss": 0.001,
+      "step": 12505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0096509338839008e-05,
+      "loss": 0.0012,
+      "step": 12506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0094108609017143e-05,
+      "loss": 0.001,
+      "step": 12507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0091707879195275e-05,
+      "loss": 0.0008,
+      "step": 12508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.008930714937341e-05,
+      "loss": 0.0011,
+      "step": 12509
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0086906419551542e-05,
+      "loss": 0.8337,
+      "step": 12510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.008450568972968e-05,
+      "loss": 0.0224,
+      "step": 12511
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0082104959907812e-05,
+      "loss": 0.0012,
+      "step": 12512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0079704230085947e-05,
+      "loss": 0.0013,
+      "step": 12513
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.007730350026408e-05,
+      "loss": 0.0009,
+      "step": 12514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0074902770442218e-05,
+      "loss": 0.0083,
+      "step": 12515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.007250204062035e-05,
+      "loss": 0.0012,
+      "step": 12516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0070101310798485e-05,
+      "loss": 0.001,
+      "step": 12517
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0067700580976616e-05,
+      "loss": 0.019,
+      "step": 12518
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.006529985115475e-05,
+      "loss": 0.0008,
+      "step": 12519
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0062899121332887e-05,
+      "loss": 0.0011,
+      "step": 12520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0060498391511022e-05,
+      "loss": 0.7784,
+      "step": 12521
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0058097661689154e-05,
+      "loss": 0.0013,
+      "step": 12522
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.005569693186729e-05,
+      "loss": 0.0009,
+      "step": 12523
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0053296202045424e-05,
+      "loss": 0.0009,
+      "step": 12524
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0050895472223556e-05,
+      "loss": 0.001,
+      "step": 12525
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.004849474240169e-05,
+      "loss": 0.0011,
+      "step": 12526
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0046094012579826e-05,
+      "loss": 0.0013,
+      "step": 12527
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0043693282757958e-05,
+      "loss": 0.936,
+      "step": 12528
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0041292552936093e-05,
+      "loss": 0.001,
+      "step": 12529
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.003889182311423e-05,
+      "loss": 0.0013,
+      "step": 12530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.003649109329236e-05,
+      "loss": 0.0007,
+      "step": 12531
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0034090363470496e-05,
+      "loss": 0.0011,
+      "step": 12532
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.003168963364863e-05,
+      "loss": 0.0061,
+      "step": 12533
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0029288903826766e-05,
+      "loss": 0.0012,
+      "step": 12534
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0026888174004898e-05,
+      "loss": 0.0008,
+      "step": 12535
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0024487444183033e-05,
+      "loss": 0.0013,
+      "step": 12536
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0022086714361165e-05,
+      "loss": 0.0012,
+      "step": 12537
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0019685984539303e-05,
+      "loss": 0.0012,
+      "step": 12538
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0017285254717435e-05,
+      "loss": 0.0013,
+      "step": 12539
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.001488452489557e-05,
+      "loss": 0.0018,
+      "step": 12540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0012483795073702e-05,
+      "loss": 0.001,
+      "step": 12541
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0010083065251837e-05,
+      "loss": 0.001,
+      "step": 12542
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0007682335429972e-05,
+      "loss": 0.0015,
+      "step": 12543
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0005281605608108e-05,
+      "loss": 0.0009,
+      "step": 12544
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.000288087578624e-05,
+      "loss": 0.0015,
+      "step": 12545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0000480145964375e-05,
+      "loss": 0.001,
+      "step": 12546
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999807941614251e-05,
+      "loss": 0.001,
+      "step": 12547
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999567868632064e-05,
+      "loss": 0.0012,
+      "step": 12548
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9993277956498777e-05,
+      "loss": 0.0006,
+      "step": 12549
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999087722667691e-05,
+      "loss": 0.5425,
+      "step": 12550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9988476496855044e-05,
+      "loss": 0.0009,
+      "step": 12551
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.998607576703318e-05,
+      "loss": 0.001,
+      "step": 12552
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9983675037211314e-05,
+      "loss": 0.0019,
+      "step": 12553
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9981274307389446e-05,
+      "loss": 0.0009,
+      "step": 12554
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.997887357756758e-05,
+      "loss": 0.0012,
+      "step": 12555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9976472847745716e-05,
+      "loss": 0.0015,
+      "step": 12556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.997407211792385e-05,
+      "loss": 0.001,
+      "step": 12557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9971671388101983e-05,
+      "loss": 0.0015,
+      "step": 12558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.996927065828012e-05,
+      "loss": 0.0016,
+      "step": 12559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.996686992845825e-05,
+      "loss": 0.0009,
+      "step": 12560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.996446919863639e-05,
+      "loss": 0.0007,
+      "step": 12561
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.996206846881452e-05,
+      "loss": 0.0009,
+      "step": 12562
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9959667738992656e-05,
+      "loss": 0.0009,
+      "step": 12563
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9957267009170788e-05,
+      "loss": 0.0009,
+      "step": 12564
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9954866279348923e-05,
+      "loss": 0.001,
+      "step": 12565
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9952465549527058e-05,
+      "loss": 0.0008,
+      "step": 12566
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.995006481970519e-05,
+      "loss": 0.0008,
+      "step": 12567
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9947664089883325e-05,
+      "loss": 0.0052,
+      "step": 12568
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.994526336006146e-05,
+      "loss": 0.0008,
+      "step": 12569
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9942862630239595e-05,
+      "loss": 0.0008,
+      "step": 12570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9940461900417727e-05,
+      "loss": 0.0023,
+      "step": 12571
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9938061170595862e-05,
+      "loss": 0.0076,
+      "step": 12572
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9935660440773994e-05,
+      "loss": 0.0009,
+      "step": 12573
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.993325971095213e-05,
+      "loss": 0.0012,
+      "step": 12574
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9930858981130265e-05,
+      "loss": 0.0064,
+      "step": 12575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.99284582513084e-05,
+      "loss": 0.0012,
+      "step": 12576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.992605752148653e-05,
+      "loss": 0.0008,
+      "step": 12577
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9923656791664667e-05,
+      "loss": 0.001,
+      "step": 12578
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9921256061842802e-05,
+      "loss": 0.0013,
+      "step": 12579
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9918855332020937e-05,
+      "loss": 0.0008,
+      "step": 12580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.991645460219907e-05,
+      "loss": 0.0011,
+      "step": 12581
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9914053872377204e-05,
+      "loss": 0.0008,
+      "step": 12582
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9911653142555336e-05,
+      "loss": 0.0011,
+      "step": 12583
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9909252412733474e-05,
+      "loss": 0.0008,
+      "step": 12584
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9906851682911606e-05,
+      "loss": 0.0009,
+      "step": 12585
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.990445095308974e-05,
+      "loss": 0.8766,
+      "step": 12586
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9902050223267873e-05,
+      "loss": 0.0005,
+      "step": 12587
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.989964949344601e-05,
+      "loss": 0.0007,
+      "step": 12588
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9897248763624144e-05,
+      "loss": 0.0011,
+      "step": 12589
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9894848033802275e-05,
+      "loss": 0.8,
+      "step": 12590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.989244730398041e-05,
+      "loss": 0.0011,
+      "step": 12591
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9890046574158542e-05,
+      "loss": 0.0009,
+      "step": 12592
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988764584433668e-05,
+      "loss": 0.0009,
+      "step": 12593
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9885245114514813e-05,
+      "loss": 0.0032,
+      "step": 12594
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9882844384692948e-05,
+      "loss": 0.0012,
+      "step": 12595
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988044365487108e-05,
+      "loss": 0.0009,
+      "step": 12596
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.987804292504922e-05,
+      "loss": 0.0012,
+      "step": 12597
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.987564219522735e-05,
+      "loss": 0.001,
+      "step": 12598
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9873241465405485e-05,
+      "loss": 0.0008,
+      "step": 12599
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9870840735583617e-05,
+      "loss": 0.001,
+      "step": 12600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9868440005761752e-05,
+      "loss": 0.001,
+      "step": 12601
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9866039275939888e-05,
+      "loss": 0.0009,
+      "step": 12602
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9863638546118023e-05,
+      "loss": 0.001,
+      "step": 12603
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9861237816296155e-05,
+      "loss": 0.0007,
+      "step": 12604
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.985883708647429e-05,
+      "loss": 0.0067,
+      "step": 12605
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.985643635665242e-05,
+      "loss": 0.0018,
+      "step": 12606
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9854035626830557e-05,
+      "loss": 0.001,
+      "step": 12607
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9851634897008692e-05,
+      "loss": 0.0011,
+      "step": 12608
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9849234167186824e-05,
+      "loss": 0.001,
+      "step": 12609
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.984683343736496e-05,
+      "loss": 0.0004,
+      "step": 12610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9844432707543094e-05,
+      "loss": 0.0005,
+      "step": 12611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.984203197772123e-05,
+      "loss": 0.0012,
+      "step": 12612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.983963124789936e-05,
+      "loss": 0.001,
+      "step": 12613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9837230518077496e-05,
+      "loss": 0.0028,
+      "step": 12614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9834829788255628e-05,
+      "loss": 0.0008,
+      "step": 12615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9832429058433767e-05,
+      "loss": 0.0006,
+      "step": 12616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.98300283286119e-05,
+      "loss": 0.0006,
+      "step": 12617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9827627598790034e-05,
+      "loss": 0.0008,
+      "step": 12618
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9825226868968165e-05,
+      "loss": 0.0009,
+      "step": 12619
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9822826139146304e-05,
+      "loss": 0.0007,
+      "step": 12620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9820425409324436e-05,
+      "loss": 0.0006,
+      "step": 12621
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.981802467950257e-05,
+      "loss": 0.0006,
+      "step": 12622
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9815623949680703e-05,
+      "loss": 0.0022,
+      "step": 12623
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9813223219858838e-05,
+      "loss": 0.4327,
+      "step": 12624
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9810822490036973e-05,
+      "loss": 0.0008,
+      "step": 12625
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980842176021511e-05,
+      "loss": 0.0011,
+      "step": 12626
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980602103039324e-05,
+      "loss": 0.0009,
+      "step": 12627
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9803620300571375e-05,
+      "loss": 0.001,
+      "step": 12628
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.980121957074951e-05,
+      "loss": 0.0007,
+      "step": 12629
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9798818840927642e-05,
+      "loss": 0.0008,
+      "step": 12630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9796418111105778e-05,
+      "loss": 0.0007,
+      "step": 12631
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.979401738128391e-05,
+      "loss": 0.0009,
+      "step": 12632
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9791616651462045e-05,
+      "loss": 0.0007,
+      "step": 12633
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.978921592164018e-05,
+      "loss": 0.6489,
+      "step": 12634
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9786815191818315e-05,
+      "loss": 0.0007,
+      "step": 12635
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9784414461996447e-05,
+      "loss": 0.0023,
+      "step": 12636
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9782013732174582e-05,
+      "loss": 0.0009,
+      "step": 12637
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9779613002352714e-05,
+      "loss": 0.0007,
+      "step": 12638
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9777212272530852e-05,
+      "loss": 0.0007,
+      "step": 12639
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9774811542708984e-05,
+      "loss": 0.0009,
+      "step": 12640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.977241081288712e-05,
+      "loss": 0.0043,
+      "step": 12641
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.977001008306525e-05,
+      "loss": 0.0011,
+      "step": 12642
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.976760935324339e-05,
+      "loss": 0.0006,
+      "step": 12643
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.976520862342152e-05,
+      "loss": 0.0005,
+      "step": 12644
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9762807893599657e-05,
+      "loss": 0.0016,
+      "step": 12645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.976040716377779e-05,
+      "loss": 0.0008,
+      "step": 12646
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9758006433955924e-05,
+      "loss": 0.3746,
+      "step": 12647
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.975560570413406e-05,
+      "loss": 0.0037,
+      "step": 12648
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.975320497431219e-05,
+      "loss": 0.001,
+      "step": 12649
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9750804244490326e-05,
+      "loss": 0.0006,
+      "step": 12650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.974840351466846e-05,
+      "loss": 1.4673,
+      "step": 12651
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9746002784846596e-05,
+      "loss": 0.0034,
+      "step": 12652
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9743602055024728e-05,
+      "loss": 0.0028,
+      "step": 12653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9741201325202863e-05,
+      "loss": 0.0033,
+      "step": 12654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9738800595380995e-05,
+      "loss": 0.0009,
+      "step": 12655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.973639986555913e-05,
+      "loss": 0.0011,
+      "step": 12656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9733999135737265e-05,
+      "loss": 0.0022,
+      "step": 12657
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.97315984059154e-05,
+      "loss": 0.0026,
+      "step": 12658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9729197676093532e-05,
+      "loss": 0.0026,
+      "step": 12659
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9726796946271667e-05,
+      "loss": 0.0059,
+      "step": 12660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9724396216449803e-05,
+      "loss": 0.001,
+      "step": 12661
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9721995486627938e-05,
+      "loss": 0.0009,
+      "step": 12662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.971959475680607e-05,
+      "loss": 0.0007,
+      "step": 12663
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9717194026984205e-05,
+      "loss": 0.0035,
+      "step": 12664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9714793297162337e-05,
+      "loss": 0.0008,
+      "step": 12665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9712392567340475e-05,
+      "loss": 0.0198,
+      "step": 12666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9709991837518607e-05,
+      "loss": 0.0181,
+      "step": 12667
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9707591107696742e-05,
+      "loss": 0.9775,
+      "step": 12668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9705190377874874e-05,
+      "loss": 0.0011,
+      "step": 12669
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.970278964805301e-05,
+      "loss": 0.001,
+      "step": 12670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9700388918231144e-05,
+      "loss": 0.0006,
+      "step": 12671
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9697988188409276e-05,
+      "loss": 0.0008,
+      "step": 12672
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.969558745858741e-05,
+      "loss": 0.0011,
+      "step": 12673
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9693186728765543e-05,
+      "loss": 0.0004,
+      "step": 12674
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9690785998943682e-05,
+      "loss": 0.0007,
+      "step": 12675
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9688385269121814e-05,
+      "loss": 0.0007,
+      "step": 12676
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968598453929995e-05,
+      "loss": 0.0015,
+      "step": 12677
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.968358380947808e-05,
+      "loss": 0.0006,
+      "step": 12678
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9681183079656216e-05,
+      "loss": 0.0019,
+      "step": 12679
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.967878234983435e-05,
+      "loss": 0.0007,
+      "step": 12680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9676381620012486e-05,
+      "loss": 0.0008,
+      "step": 12681
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9673980890190618e-05,
+      "loss": 0.0005,
+      "step": 12682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9671580160368753e-05,
+      "loss": 0.616,
+      "step": 12683
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9669179430546888e-05,
+      "loss": 0.0018,
+      "step": 12684
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9666778700725023e-05,
+      "loss": 0.0006,
+      "step": 12685
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9664377970903155e-05,
+      "loss": 0.0007,
+      "step": 12686
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.966197724108129e-05,
+      "loss": 0.0007,
+      "step": 12687
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9659576511259422e-05,
+      "loss": 0.0021,
+      "step": 12688
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9657175781437557e-05,
+      "loss": 0.0005,
+      "step": 12689
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9654775051615693e-05,
+      "loss": 0.0061,
+      "step": 12690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9652374321793824e-05,
+      "loss": 0.0033,
+      "step": 12691
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.964997359197196e-05,
+      "loss": 0.0019,
+      "step": 12692
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9647572862150095e-05,
+      "loss": 0.0019,
+      "step": 12693
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.964517213232823e-05,
+      "loss": 0.0033,
+      "step": 12694
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9642771402506362e-05,
+      "loss": 0.0007,
+      "step": 12695
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9640370672684497e-05,
+      "loss": 0.0005,
+      "step": 12696
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.963796994286263e-05,
+      "loss": 0.0005,
+      "step": 12697
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9635569213040767e-05,
+      "loss": 0.0009,
+      "step": 12698
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.96331684832189e-05,
+      "loss": 0.0005,
+      "step": 12699
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9630767753397034e-05,
+      "loss": 0.0006,
+      "step": 12700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9628367023575166e-05,
+      "loss": 0.0007,
+      "step": 12701
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.96259662937533e-05,
+      "loss": 0.0007,
+      "step": 12702
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9623565563931437e-05,
+      "loss": 0.0008,
+      "step": 12703
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9621164834109572e-05,
+      "loss": 0.0005,
+      "step": 12704
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9618764104287704e-05,
+      "loss": 0.0006,
+      "step": 12705
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.961636337446584e-05,
+      "loss": 0.0005,
+      "step": 12706
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9613962644643974e-05,
+      "loss": 0.0007,
+      "step": 12707
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.961156191482211e-05,
+      "loss": 0.0009,
+      "step": 12708
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.960916118500024e-05,
+      "loss": 0.0008,
+      "step": 12709
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9606760455178376e-05,
+      "loss": 0.0042,
+      "step": 12710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9604359725356508e-05,
+      "loss": 0.0004,
+      "step": 12711
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9601958995534643e-05,
+      "loss": 0.0005,
+      "step": 12712
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9599558265712778e-05,
+      "loss": 0.0005,
+      "step": 12713
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.959715753589091e-05,
+      "loss": 0.0006,
+      "step": 12714
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9594756806069045e-05,
+      "loss": 0.0005,
+      "step": 12715
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.959235607624718e-05,
+      "loss": 0.0005,
+      "step": 12716
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9589955346425316e-05,
+      "loss": 0.0005,
+      "step": 12717
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9587554616603447e-05,
+      "loss": 0.0004,
+      "step": 12718
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9585153886781583e-05,
+      "loss": 0.0031,
+      "step": 12719
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9582753156959714e-05,
+      "loss": 0.0071,
+      "step": 12720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9580352427137853e-05,
+      "loss": 0.0006,
+      "step": 12721
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9577951697315985e-05,
+      "loss": 0.4869,
+      "step": 12722
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.957555096749412e-05,
+      "loss": 0.0006,
+      "step": 12723
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9573150237672252e-05,
+      "loss": 0.0013,
+      "step": 12724
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9570749507850387e-05,
+      "loss": 0.0036,
+      "step": 12725
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9568348778028522e-05,
+      "loss": 0.0047,
+      "step": 12726
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9565948048206657e-05,
+      "loss": 0.0006,
+      "step": 12727
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.956354731838479e-05,
+      "loss": 0.0006,
+      "step": 12728
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9561146588562924e-05,
+      "loss": 0.0003,
+      "step": 12729
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955874585874106e-05,
+      "loss": 0.0006,
+      "step": 12730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955634512891919e-05,
+      "loss": 0.0006,
+      "step": 12731
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9553944399097327e-05,
+      "loss": 0.0005,
+      "step": 12732
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955154366927546e-05,
+      "loss": 0.0006,
+      "step": 12733
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9549142939453594e-05,
+      "loss": 0.0005,
+      "step": 12734
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.954674220963173e-05,
+      "loss": 0.0007,
+      "step": 12735
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9544341479809864e-05,
+      "loss": 0.0007,
+      "step": 12736
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9541940749987996e-05,
+      "loss": 0.0008,
+      "step": 12737
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.953954002016613e-05,
+      "loss": 0.0007,
+      "step": 12738
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9537139290344266e-05,
+      "loss": 0.0005,
+      "step": 12739
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.95347385605224e-05,
+      "loss": 0.004,
+      "step": 12740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9532337830700533e-05,
+      "loss": 0.0005,
+      "step": 12741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9529937100878668e-05,
+      "loss": 0.0005,
+      "step": 12742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.95275363710568e-05,
+      "loss": 0.0006,
+      "step": 12743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.952513564123494e-05,
+      "loss": 0.0005,
+      "step": 12744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.952273491141307e-05,
+      "loss": 0.0004,
+      "step": 12745
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9520334181591206e-05,
+      "loss": 0.0007,
+      "step": 12746
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9517933451769337e-05,
+      "loss": 0.0007,
+      "step": 12747
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9515532721947473e-05,
+      "loss": 0.0005,
+      "step": 12748
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9513131992125608e-05,
+      "loss": 0.0005,
+      "step": 12749
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9510731262303743e-05,
+      "loss": 0.0005,
+      "step": 12750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9508330532481875e-05,
+      "loss": 0.0004,
+      "step": 12751
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.950592980266001e-05,
+      "loss": 0.0006,
+      "step": 12752
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9503529072838145e-05,
+      "loss": 0.0006,
+      "step": 12753
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9501128343016277e-05,
+      "loss": 0.0007,
+      "step": 12754
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9498727613194412e-05,
+      "loss": 0.0003,
+      "step": 12755
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9496326883372544e-05,
+      "loss": 0.0006,
+      "step": 12756
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.949392615355068e-05,
+      "loss": 0.0004,
+      "step": 12757
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9491525423728814e-05,
+      "loss": 0.0006,
+      "step": 12758
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.948912469390695e-05,
+      "loss": 0.0004,
+      "step": 12759
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.948672396408508e-05,
+      "loss": 0.0004,
+      "step": 12760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9484323234263216e-05,
+      "loss": 0.0005,
+      "step": 12761
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.948192250444135e-05,
+      "loss": 0.0004,
+      "step": 12762
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9479521774619487e-05,
+      "loss": 0.0025,
+      "step": 12763
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.947712104479762e-05,
+      "loss": 0.0006,
+      "step": 12764
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9474720314975754e-05,
+      "loss": 0.0006,
+      "step": 12765
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9472319585153886e-05,
+      "loss": 1.0456,
+      "step": 12766
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9469918855332024e-05,
+      "loss": 0.0005,
+      "step": 12767
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9467518125510156e-05,
+      "loss": 0.0006,
+      "step": 12768
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.946511739568829e-05,
+      "loss": 0.0005,
+      "step": 12769
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9462716665866423e-05,
+      "loss": 0.0006,
+      "step": 12770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9460315936044558e-05,
+      "loss": 0.0004,
+      "step": 12771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9457915206222693e-05,
+      "loss": 0.0005,
+      "step": 12772
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9455514476400825e-05,
+      "loss": 0.0005,
+      "step": 12773
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.945311374657896e-05,
+      "loss": 0.0006,
+      "step": 12774
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9450713016757096e-05,
+      "loss": 0.0042,
+      "step": 12775
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.944831228693523e-05,
+      "loss": 0.0005,
+      "step": 12776
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9445911557113363e-05,
+      "loss": 0.4665,
+      "step": 12777
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9443510827291498e-05,
+      "loss": 0.0007,
+      "step": 12778
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.944111009746963e-05,
+      "loss": 0.0005,
+      "step": 12779
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9438709367647768e-05,
+      "loss": 0.0006,
+      "step": 12780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.94363086378259e-05,
+      "loss": 0.0004,
+      "step": 12781
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9433907908004035e-05,
+      "loss": 0.0005,
+      "step": 12782
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9431507178182167e-05,
+      "loss": 0.0004,
+      "step": 12783
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9429106448360302e-05,
+      "loss": 0.0005,
+      "step": 12784
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9426705718538437e-05,
+      "loss": 0.0005,
+      "step": 12785
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9424304988716572e-05,
+      "loss": 0.0005,
+      "step": 12786
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9421904258894704e-05,
+      "loss": 0.0006,
+      "step": 12787
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941950352907284e-05,
+      "loss": 0.0005,
+      "step": 12788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941710279925097e-05,
+      "loss": 0.0006,
+      "step": 12789
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941470206942911e-05,
+      "loss": 0.0005,
+      "step": 12790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.941230133960724e-05,
+      "loss": 0.0005,
+      "step": 12791
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9409900609785377e-05,
+      "loss": 0.0006,
+      "step": 12792
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.940749987996351e-05,
+      "loss": 0.0005,
+      "step": 12793
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9405099150141644e-05,
+      "loss": 0.0005,
+      "step": 12794
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.940269842031978e-05,
+      "loss": 0.9023,
+      "step": 12795
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.940029769049791e-05,
+      "loss": 0.0005,
+      "step": 12796
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9397896960676046e-05,
+      "loss": 0.0005,
+      "step": 12797
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9395496230854178e-05,
+      "loss": 1.0487,
+      "step": 12798
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9393095501032316e-05,
+      "loss": 0.0006,
+      "step": 12799
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9390694771210448e-05,
+      "loss": 0.0005,
+      "step": 12800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9388294041388583e-05,
+      "loss": 0.3776,
+      "step": 12801
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9385893311566715e-05,
+      "loss": 0.0006,
+      "step": 12802
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9383492581744854e-05,
+      "loss": 0.0033,
+      "step": 12803
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9381091851922986e-05,
+      "loss": 0.0007,
+      "step": 12804
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.937869112210112e-05,
+      "loss": 0.0007,
+      "step": 12805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9376290392279253e-05,
+      "loss": 0.001,
+      "step": 12806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9373889662457388e-05,
+      "loss": 0.0007,
+      "step": 12807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9371488932635523e-05,
+      "loss": 0.0007,
+      "step": 12808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9369088202813658e-05,
+      "loss": 0.0007,
+      "step": 12809
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.936668747299179e-05,
+      "loss": 0.0006,
+      "step": 12810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9364286743169925e-05,
+      "loss": 0.0006,
+      "step": 12811
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.936188601334806e-05,
+      "loss": 0.0007,
+      "step": 12812
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9359485283526192e-05,
+      "loss": 0.0006,
+      "step": 12813
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9357084553704327e-05,
+      "loss": 0.0007,
+      "step": 12814
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.935468382388246e-05,
+      "loss": 0.0005,
+      "step": 12815
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9352283094060594e-05,
+      "loss": 0.0005,
+      "step": 12816
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.934988236423873e-05,
+      "loss": 0.0005,
+      "step": 12817
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9347481634416865e-05,
+      "loss": 0.0164,
+      "step": 12818
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9345080904594996e-05,
+      "loss": 0.0006,
+      "step": 12819
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.934268017477313e-05,
+      "loss": 0.0005,
+      "step": 12820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9340279444951263e-05,
+      "loss": 0.0006,
+      "step": 12821
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9337878715129402e-05,
+      "loss": 0.0005,
+      "step": 12822
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9335477985307534e-05,
+      "loss": 0.0032,
+      "step": 12823
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.933307725548567e-05,
+      "loss": 0.0005,
+      "step": 12824
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.93306765256638e-05,
+      "loss": 0.0005,
+      "step": 12825
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932827579584194e-05,
+      "loss": 1.0137,
+      "step": 12826
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.932587506602007e-05,
+      "loss": 0.0005,
+      "step": 12827
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9323474336198206e-05,
+      "loss": 0.0006,
+      "step": 12828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9321073606376338e-05,
+      "loss": 0.0006,
+      "step": 12829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9318672876554473e-05,
+      "loss": 0.0005,
+      "step": 12830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.931627214673261e-05,
+      "loss": 0.0006,
+      "step": 12831
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9313871416910744e-05,
+      "loss": 0.3976,
+      "step": 12832
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9311470687088876e-05,
+      "loss": 0.0005,
+      "step": 12833
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.930906995726701e-05,
+      "loss": 0.0006,
+      "step": 12834
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9306669227445146e-05,
+      "loss": 0.0007,
+      "step": 12835
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9304268497623278e-05,
+      "loss": 0.0006,
+      "step": 12836
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9301867767801413e-05,
+      "loss": 0.0006,
+      "step": 12837
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9299467037979545e-05,
+      "loss": 0.3934,
+      "step": 12838
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.929706630815768e-05,
+      "loss": 0.0006,
+      "step": 12839
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9294665578335815e-05,
+      "loss": 0.0006,
+      "step": 12840
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.929226484851395e-05,
+      "loss": 0.0029,
+      "step": 12841
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9289864118692082e-05,
+      "loss": 0.0006,
+      "step": 12842
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9287463388870217e-05,
+      "loss": 0.0006,
+      "step": 12843
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9285062659048352e-05,
+      "loss": 0.0007,
+      "step": 12844
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9282661929226488e-05,
+      "loss": 0.0006,
+      "step": 12845
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.928026119940462e-05,
+      "loss": 0.0006,
+      "step": 12846
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9277860469582755e-05,
+      "loss": 0.0006,
+      "step": 12847
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9275459739760886e-05,
+      "loss": 0.0006,
+      "step": 12848
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9273059009939025e-05,
+      "loss": 0.0006,
+      "step": 12849
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9270658280117157e-05,
+      "loss": 0.0007,
+      "step": 12850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9268257550295292e-05,
+      "loss": 0.0006,
+      "step": 12851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9265856820473424e-05,
+      "loss": 0.0007,
+      "step": 12852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.926345609065156e-05,
+      "loss": 0.0006,
+      "step": 12853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9261055360829694e-05,
+      "loss": 0.0006,
+      "step": 12854
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9258654631007826e-05,
+      "loss": 0.0007,
+      "step": 12855
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.925625390118596e-05,
+      "loss": 0.0006,
+      "step": 12856
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9253853171364093e-05,
+      "loss": 0.0024,
+      "step": 12857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.925145244154223e-05,
+      "loss": 0.0006,
+      "step": 12858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9249051711720363e-05,
+      "loss": 0.0007,
+      "step": 12859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.92466509818985e-05,
+      "loss": 0.0006,
+      "step": 12860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.924425025207663e-05,
+      "loss": 0.0006,
+      "step": 12861
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9241849522254765e-05,
+      "loss": 0.0005,
+      "step": 12862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.92394487924329e-05,
+      "loss": 0.0007,
+      "step": 12863
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9237048062611036e-05,
+      "loss": 0.0005,
+      "step": 12864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9234647332789168e-05,
+      "loss": 0.0006,
+      "step": 12865
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9232246602967303e-05,
+      "loss": 0.0032,
+      "step": 12866
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9229845873145438e-05,
+      "loss": 0.0005,
+      "step": 12867
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9227445143323573e-05,
+      "loss": 0.0006,
+      "step": 12868
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9225044413501705e-05,
+      "loss": 0.0006,
+      "step": 12869
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.922264368367984e-05,
+      "loss": 0.0005,
+      "step": 12870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9220242953857972e-05,
+      "loss": 0.0025,
+      "step": 12871
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9217842224036107e-05,
+      "loss": 0.0005,
+      "step": 12872
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9215441494214242e-05,
+      "loss": 0.0006,
+      "step": 12873
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9213040764392378e-05,
+      "loss": 0.0007,
+      "step": 12874
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.921064003457051e-05,
+      "loss": 0.0006,
+      "step": 12875
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9208239304748645e-05,
+      "loss": 0.0006,
+      "step": 12876
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.920583857492678e-05,
+      "loss": 0.0005,
+      "step": 12877
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.920343784510491e-05,
+      "loss": 0.0006,
+      "step": 12878
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9201037115283047e-05,
+      "loss": 0.397,
+      "step": 12879
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.919863638546118e-05,
+      "loss": 0.0005,
+      "step": 12880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9196235655639317e-05,
+      "loss": 0.0006,
+      "step": 12881
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.919383492581745e-05,
+      "loss": 0.0005,
+      "step": 12882
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9191434195995584e-05,
+      "loss": 0.0006,
+      "step": 12883
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9189033466173716e-05,
+      "loss": 0.0007,
+      "step": 12884
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.918663273635185e-05,
+      "loss": 0.0006,
+      "step": 12885
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9184232006529986e-05,
+      "loss": 0.0006,
+      "step": 12886
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.918183127670812e-05,
+      "loss": 0.0006,
+      "step": 12887
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9179430546886253e-05,
+      "loss": 0.0006,
+      "step": 12888
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.917702981706439e-05,
+      "loss": 0.0006,
+      "step": 12889
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9174629087242524e-05,
+      "loss": 0.2885,
+      "step": 12890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.917222835742066e-05,
+      "loss": 0.0005,
+      "step": 12891
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.916982762759879e-05,
+      "loss": 0.0005,
+      "step": 12892
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9167426897776926e-05,
+      "loss": 0.0005,
+      "step": 12893
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9165026167955058e-05,
+      "loss": 0.0006,
+      "step": 12894
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9162625438133193e-05,
+      "loss": 0.0005,
+      "step": 12895
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9160224708311328e-05,
+      "loss": 0.0006,
+      "step": 12896
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.915782397848946e-05,
+      "loss": 0.0005,
+      "step": 12897
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9155423248667595e-05,
+      "loss": 0.0005,
+      "step": 12898
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.915302251884573e-05,
+      "loss": 0.0005,
+      "step": 12899
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9150621789023865e-05,
+      "loss": 0.0005,
+      "step": 12900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9148221059201997e-05,
+      "loss": 0.0005,
+      "step": 12901
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9145820329380132e-05,
+      "loss": 0.9735,
+      "step": 12902
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9143419599558264e-05,
+      "loss": 0.0004,
+      "step": 12903
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9141018869736403e-05,
+      "loss": 0.0334,
+      "step": 12904
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9138618139914535e-05,
+      "loss": 0.0021,
+      "step": 12905
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.913621741009267e-05,
+      "loss": 0.0006,
+      "step": 12906
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.91338166802708e-05,
+      "loss": 0.475,
+      "step": 12907
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.913141595044894e-05,
+      "loss": 0.0005,
+      "step": 12908
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9129015220627072e-05,
+      "loss": 0.0005,
+      "step": 12909
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9126614490805207e-05,
+      "loss": 0.0007,
+      "step": 12910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.912421376098334e-05,
+      "loss": 0.0005,
+      "step": 12911
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9121813031161474e-05,
+      "loss": 0.0007,
+      "step": 12912
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.911941230133961e-05,
+      "loss": 0.0005,
+      "step": 12913
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9117011571517744e-05,
+      "loss": 0.0005,
+      "step": 12914
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9114610841695876e-05,
+      "loss": 0.0006,
+      "step": 12915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.911221011187401e-05,
+      "loss": 0.0005,
+      "step": 12916
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9109809382052143e-05,
+      "loss": 0.0006,
+      "step": 12917
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.910740865223028e-05,
+      "loss": 0.0005,
+      "step": 12918
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9105007922408414e-05,
+      "loss": 0.0007,
+      "step": 12919
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9102607192586545e-05,
+      "loss": 0.0005,
+      "step": 12920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.910020646276468e-05,
+      "loss": 0.0051,
+      "step": 12921
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9097805732942816e-05,
+      "loss": 0.0005,
+      "step": 12922
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.909540500312095e-05,
+      "loss": 0.0007,
+      "step": 12923
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9093004273299083e-05,
+      "loss": 0.0049,
+      "step": 12924
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9090603543477218e-05,
+      "loss": 0.0004,
+      "step": 12925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.908820281365535e-05,
+      "loss": 0.0006,
+      "step": 12926
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.908580208383349e-05,
+      "loss": 0.0006,
+      "step": 12927
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.908340135401162e-05,
+      "loss": 0.0006,
+      "step": 12928
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9081000624189755e-05,
+      "loss": 0.0033,
+      "step": 12929
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9078599894367887e-05,
+      "loss": 0.0007,
+      "step": 12930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9076199164546026e-05,
+      "loss": 0.0006,
+      "step": 12931
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9073798434724158e-05,
+      "loss": 0.0005,
+      "step": 12932
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9071397704902293e-05,
+      "loss": 0.0007,
+      "step": 12933
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9068996975080425e-05,
+      "loss": 0.0006,
+      "step": 12934
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.906659624525856e-05,
+      "loss": 0.0005,
+      "step": 12935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9064195515436695e-05,
+      "loss": 0.0007,
+      "step": 12936
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9061794785614827e-05,
+      "loss": 0.0006,
+      "step": 12937
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9059394055792962e-05,
+      "loss": 0.0006,
+      "step": 12938
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9056993325971094e-05,
+      "loss": 0.0006,
+      "step": 12939
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9054592596149232e-05,
+      "loss": 0.0006,
+      "step": 12940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9052191866327364e-05,
+      "loss": 0.0006,
+      "step": 12941
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.90497911365055e-05,
+      "loss": 0.0005,
+      "step": 12942
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.904739040668363e-05,
+      "loss": 0.0069,
+      "step": 12943
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9044989676861766e-05,
+      "loss": 0.4648,
+      "step": 12944
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.90425889470399e-05,
+      "loss": 0.0006,
+      "step": 12945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9040188217218037e-05,
+      "loss": 0.0005,
+      "step": 12946
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.903778748739617e-05,
+      "loss": 0.0005,
+      "step": 12947
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9035386757574304e-05,
+      "loss": 0.0006,
+      "step": 12948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9032986027752435e-05,
+      "loss": 0.25,
+      "step": 12949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9030585297930574e-05,
+      "loss": 0.0005,
+      "step": 12950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9028184568108706e-05,
+      "loss": 0.0005,
+      "step": 12951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.902578383828684e-05,
+      "loss": 0.4748,
+      "step": 12952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9023383108464973e-05,
+      "loss": 0.0006,
+      "step": 12953
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9020982378643108e-05,
+      "loss": 0.5016,
+      "step": 12954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9018581648821243e-05,
+      "loss": 0.0006,
+      "step": 12955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.901618091899938e-05,
+      "loss": 0.1994,
+      "step": 12956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.901378018917751e-05,
+      "loss": 0.4004,
+      "step": 12957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9011379459355645e-05,
+      "loss": 0.0005,
+      "step": 12958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.900897872953378e-05,
+      "loss": 0.0005,
+      "step": 12959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9006577999711912e-05,
+      "loss": 0.0005,
+      "step": 12960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9004177269890048e-05,
+      "loss": 0.0005,
+      "step": 12961
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.900177654006818e-05,
+      "loss": 0.0007,
+      "step": 12962
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8999375810246318e-05,
+      "loss": 0.0009,
+      "step": 12963
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.899697508042445e-05,
+      "loss": 0.0006,
+      "step": 12964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8994574350602585e-05,
+      "loss": 0.0145,
+      "step": 12965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8992173620780717e-05,
+      "loss": 0.0005,
+      "step": 12966
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8989772890958852e-05,
+      "loss": 0.0007,
+      "step": 12967
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8987372161136987e-05,
+      "loss": 0.0007,
+      "step": 12968
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8984971431315122e-05,
+      "loss": 0.0007,
+      "step": 12969
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8982570701493254e-05,
+      "loss": 0.0009,
+      "step": 12970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.898016997167139e-05,
+      "loss": 0.0007,
+      "step": 12971
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8977769241849524e-05,
+      "loss": 1.2573,
+      "step": 12972
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.897536851202766e-05,
+      "loss": 0.0007,
+      "step": 12973
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.897296778220579e-05,
+      "loss": 0.0007,
+      "step": 12974
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8970567052383927e-05,
+      "loss": 0.0008,
+      "step": 12975
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.896816632256206e-05,
+      "loss": 0.0006,
+      "step": 12976
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8965765592740194e-05,
+      "loss": 0.0007,
+      "step": 12977
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.896336486291833e-05,
+      "loss": 0.0006,
+      "step": 12978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.896096413309646e-05,
+      "loss": 0.2795,
+      "step": 12979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8958563403274596e-05,
+      "loss": 0.0007,
+      "step": 12980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8956162673452728e-05,
+      "loss": 0.0008,
+      "step": 12981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8953761943630866e-05,
+      "loss": 0.0008,
+      "step": 12982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8951361213808998e-05,
+      "loss": 0.0008,
+      "step": 12983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8948960483987133e-05,
+      "loss": 0.0007,
+      "step": 12984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8946559754165265e-05,
+      "loss": 0.0008,
+      "step": 12985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8944159024343404e-05,
+      "loss": 0.0009,
+      "step": 12986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8941758294521535e-05,
+      "loss": 0.0006,
+      "step": 12987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.893935756469967e-05,
+      "loss": 0.0008,
+      "step": 12988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8936956834877802e-05,
+      "loss": 0.0007,
+      "step": 12989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8934556105055937e-05,
+      "loss": 0.0008,
+      "step": 12990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8932155375234073e-05,
+      "loss": 0.0008,
+      "step": 12991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8929754645412208e-05,
+      "loss": 0.0005,
+      "step": 12992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892735391559034e-05,
+      "loss": 0.0007,
+      "step": 12993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8924953185768475e-05,
+      "loss": 0.0007,
+      "step": 12994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892255245594661e-05,
+      "loss": 0.0009,
+      "step": 12995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8920151726124742e-05,
+      "loss": 0.0006,
+      "step": 12996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8917750996302877e-05,
+      "loss": 0.0007,
+      "step": 12997
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8915350266481012e-05,
+      "loss": 0.0006,
+      "step": 12998
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8912949536659144e-05,
+      "loss": 0.0006,
+      "step": 12999
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.891054880683728e-05,
+      "loss": 0.0008,
+      "step": 13000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8908148077015414e-05,
+      "loss": 0.0006,
+      "step": 13001
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8905747347193546e-05,
+      "loss": 0.0007,
+      "step": 13002
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.890334661737168e-05,
+      "loss": 0.0008,
+      "step": 13003
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8900945887549817e-05,
+      "loss": 0.0006,
+      "step": 13004
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8898545157727952e-05,
+      "loss": 0.0006,
+      "step": 13005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8896144427906084e-05,
+      "loss": 0.0008,
+      "step": 13006
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.889374369808422e-05,
+      "loss": 0.0005,
+      "step": 13007
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.889134296826235e-05,
+      "loss": 0.0006,
+      "step": 13008
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.888894223844049e-05,
+      "loss": 0.0006,
+      "step": 13009
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.888654150861862e-05,
+      "loss": 0.0006,
+      "step": 13010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8884140778796756e-05,
+      "loss": 0.0008,
+      "step": 13011
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8881740048974888e-05,
+      "loss": 0.0007,
+      "step": 13012
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8879339319153023e-05,
+      "loss": 0.0393,
+      "step": 13013
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8876938589331158e-05,
+      "loss": 0.0006,
+      "step": 13014
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8874537859509293e-05,
+      "loss": 0.0006,
+      "step": 13015
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8872137129687425e-05,
+      "loss": 0.0006,
+      "step": 13016
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.886973639986556e-05,
+      "loss": 0.0006,
+      "step": 13017
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8867335670043696e-05,
+      "loss": 0.0007,
+      "step": 13018
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8864934940221827e-05,
+      "loss": 0.0006,
+      "step": 13019
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8862534210399963e-05,
+      "loss": 0.0006,
+      "step": 13020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8860133480578094e-05,
+      "loss": 0.0005,
+      "step": 13021
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.885773275075623e-05,
+      "loss": 0.0006,
+      "step": 13022
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8855332020934365e-05,
+      "loss": 0.0006,
+      "step": 13023
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.88529312911125e-05,
+      "loss": 0.0006,
+      "step": 13024
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8850530561290632e-05,
+      "loss": 0.0005,
+      "step": 13025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8848129831468767e-05,
+      "loss": 0.0006,
+      "step": 13026
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8845729101646902e-05,
+      "loss": 0.0006,
+      "step": 13027
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8843328371825037e-05,
+      "loss": 0.0006,
+      "step": 13028
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.884092764200317e-05,
+      "loss": 0.0006,
+      "step": 13029
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8838526912181304e-05,
+      "loss": 0.0005,
+      "step": 13030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8836126182359436e-05,
+      "loss": 0.0279,
+      "step": 13031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8833725452537575e-05,
+      "loss": 0.0006,
+      "step": 13032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8831324722715707e-05,
+      "loss": 0.0006,
+      "step": 13033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8828923992893842e-05,
+      "loss": 0.0005,
+      "step": 13034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8826523263071974e-05,
+      "loss": 0.0006,
+      "step": 13035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.882412253325011e-05,
+      "loss": 0.0006,
+      "step": 13036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8821721803428244e-05,
+      "loss": 0.0006,
+      "step": 13037
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.881932107360638e-05,
+      "loss": 0.0005,
+      "step": 13038
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.881692034378451e-05,
+      "loss": 0.0005,
+      "step": 13039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8814519613962646e-05,
+      "loss": 0.0005,
+      "step": 13040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.881211888414078e-05,
+      "loss": 0.0004,
+      "step": 13041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8809718154318913e-05,
+      "loss": 0.0005,
+      "step": 13042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8807317424497048e-05,
+      "loss": 0.0006,
+      "step": 13043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.880491669467518e-05,
+      "loss": 0.0006,
+      "step": 13044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8802515964853315e-05,
+      "loss": 0.0006,
+      "step": 13045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.880011523503145e-05,
+      "loss": 0.0005,
+      "step": 13046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8797714505209586e-05,
+      "loss": 0.0005,
+      "step": 13047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8795313775387717e-05,
+      "loss": 0.0005,
+      "step": 13048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8792913045565853e-05,
+      "loss": 0.0005,
+      "step": 13049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8790512315743988e-05,
+      "loss": 0.0005,
+      "step": 13050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8788111585922123e-05,
+      "loss": 0.0005,
+      "step": 13051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8785710856100255e-05,
+      "loss": 0.0004,
+      "step": 13052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.878331012627839e-05,
+      "loss": 0.0005,
+      "step": 13053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8780909396456522e-05,
+      "loss": 0.0263,
+      "step": 13054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.877850866663466e-05,
+      "loss": 0.0006,
+      "step": 13055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8776107936812792e-05,
+      "loss": 0.0004,
+      "step": 13056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8773707206990927e-05,
+      "loss": 0.0005,
+      "step": 13057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.877130647716906e-05,
+      "loss": 0.0005,
+      "step": 13058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8768905747347194e-05,
+      "loss": 0.0004,
+      "step": 13059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876650501752533e-05,
+      "loss": 0.0005,
+      "step": 13060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.876410428770346e-05,
+      "loss": 0.0004,
+      "step": 13061
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8761703557881597e-05,
+      "loss": 0.0006,
+      "step": 13062
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.875930282805973e-05,
+      "loss": 0.0004,
+      "step": 13063
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8756902098237867e-05,
+      "loss": 0.0005,
+      "step": 13064
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8754501368416e-05,
+      "loss": 0.1338,
+      "step": 13065
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8752100638594134e-05,
+      "loss": 0.0004,
+      "step": 13066
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8749699908772266e-05,
+      "loss": 0.0004,
+      "step": 13067
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.87472991789504e-05,
+      "loss": 0.0006,
+      "step": 13068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8744898449128536e-05,
+      "loss": 0.9052,
+      "step": 13069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.874249771930667e-05,
+      "loss": 0.0004,
+      "step": 13070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8740096989484803e-05,
+      "loss": 0.0004,
+      "step": 13071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8737696259662938e-05,
+      "loss": 0.0004,
+      "step": 13072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8735295529841073e-05,
+      "loss": 0.0004,
+      "step": 13073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.873289480001921e-05,
+      "loss": 0.1074,
+      "step": 13074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.873049407019734e-05,
+      "loss": 0.0004,
+      "step": 13075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8728093340375476e-05,
+      "loss": 0.0005,
+      "step": 13076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8725692610553607e-05,
+      "loss": 0.0006,
+      "step": 13077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8723291880731743e-05,
+      "loss": 0.0004,
+      "step": 13078
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8720891150909878e-05,
+      "loss": 0.0005,
+      "step": 13079
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8718490421088013e-05,
+      "loss": 0.001,
+      "step": 13080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8716089691266145e-05,
+      "loss": 0.0004,
+      "step": 13081
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.871368896144428e-05,
+      "loss": 0.0005,
+      "step": 13082
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8711288231622415e-05,
+      "loss": 0.0005,
+      "step": 13083
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8708887501800547e-05,
+      "loss": 0.0004,
+      "step": 13084
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8706486771978682e-05,
+      "loss": 0.0004,
+      "step": 13085
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8704086042156814e-05,
+      "loss": 0.0005,
+      "step": 13086
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8701685312334953e-05,
+      "loss": 0.0005,
+      "step": 13087
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8699284582513084e-05,
+      "loss": 0.0188,
+      "step": 13088
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.869688385269122e-05,
+      "loss": 0.0009,
+      "step": 13089
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.869448312286935e-05,
+      "loss": 0.0004,
+      "step": 13090
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.869208239304749e-05,
+      "loss": 0.0004,
+      "step": 13091
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.868968166322562e-05,
+      "loss": 0.0005,
+      "step": 13092
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8687280933403757e-05,
+      "loss": 0.0005,
+      "step": 13093
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.868488020358189e-05,
+      "loss": 0.0005,
+      "step": 13094
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8682479473760024e-05,
+      "loss": 0.0005,
+      "step": 13095
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.868007874393816e-05,
+      "loss": 0.0004,
+      "step": 13096
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8677678014116294e-05,
+      "loss": 0.0004,
+      "step": 13097
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8675277284294426e-05,
+      "loss": 0.0004,
+      "step": 13098
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.867287655447256e-05,
+      "loss": 0.0005,
+      "step": 13099
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8670475824650693e-05,
+      "loss": 0.2703,
+      "step": 13100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8668075094828828e-05,
+      "loss": 0.0004,
+      "step": 13101
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8665674365006963e-05,
+      "loss": 0.0292,
+      "step": 13102
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8663273635185095e-05,
+      "loss": 0.0005,
+      "step": 13103
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.866087290536323e-05,
+      "loss": 0.0005,
+      "step": 13104
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8658472175541366e-05,
+      "loss": 0.0005,
+      "step": 13105
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.86560714457195e-05,
+      "loss": 0.0005,
+      "step": 13106
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8653670715897633e-05,
+      "loss": 0.0008,
+      "step": 13107
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8651269986075768e-05,
+      "loss": 0.0004,
+      "step": 13108
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.86488692562539e-05,
+      "loss": 0.0004,
+      "step": 13109
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8646468526432038e-05,
+      "loss": 0.0004,
+      "step": 13110
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.864406779661017e-05,
+      "loss": 0.0005,
+      "step": 13111
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8641667066788305e-05,
+      "loss": 0.0003,
+      "step": 13112
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8639266336966437e-05,
+      "loss": 0.0005,
+      "step": 13113
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8636865607144575e-05,
+      "loss": 0.0004,
+      "step": 13114
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8634464877322707e-05,
+      "loss": 0.0005,
+      "step": 13115
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8632064147500842e-05,
+      "loss": 0.0004,
+      "step": 13116
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8629663417678974e-05,
+      "loss": 0.0004,
+      "step": 13117
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.862726268785711e-05,
+      "loss": 0.0004,
+      "step": 13118
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8624861958035245e-05,
+      "loss": 0.0004,
+      "step": 13119
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8622461228213376e-05,
+      "loss": 0.0004,
+      "step": 13120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.862006049839151e-05,
+      "loss": 0.0004,
+      "step": 13121
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8617659768569647e-05,
+      "loss": 0.0004,
+      "step": 13122
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8615259038747782e-05,
+      "loss": 0.0004,
+      "step": 13123
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8612858308925914e-05,
+      "loss": 0.0004,
+      "step": 13124
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.861045757910405e-05,
+      "loss": 0.0004,
+      "step": 13125
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.860805684928218e-05,
+      "loss": 0.0004,
+      "step": 13126
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8605656119460316e-05,
+      "loss": 0.0003,
+      "step": 13127
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.860325538963845e-05,
+      "loss": 1.0608,
+      "step": 13128
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8600854659816586e-05,
+      "loss": 0.0004,
+      "step": 13129
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8598453929994718e-05,
+      "loss": 0.2366,
+      "step": 13130
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8596053200172853e-05,
+      "loss": 0.0004,
+      "step": 13131
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8593652470350985e-05,
+      "loss": 0.0004,
+      "step": 13132
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8591251740529124e-05,
+      "loss": 0.0003,
+      "step": 13133
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8588851010707256e-05,
+      "loss": 0.9855,
+      "step": 13134
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.858645028088539e-05,
+      "loss": 0.022,
+      "step": 13135
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8584049551063523e-05,
+      "loss": 0.0004,
+      "step": 13136
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.858164882124166e-05,
+      "loss": 0.0005,
+      "step": 13137
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8579248091419793e-05,
+      "loss": 0.0004,
+      "step": 13138
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8576847361597928e-05,
+      "loss": 0.0005,
+      "step": 13139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.857444663177606e-05,
+      "loss": 0.0007,
+      "step": 13140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8572045901954195e-05,
+      "loss": 0.0006,
+      "step": 13141
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.856964517213233e-05,
+      "loss": 0.0007,
+      "step": 13142
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8567244442310462e-05,
+      "loss": 0.0007,
+      "step": 13143
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8564843712488597e-05,
+      "loss": 0.9945,
+      "step": 13144
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.856244298266673e-05,
+      "loss": 0.0008,
+      "step": 13145
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8560042252844868e-05,
+      "loss": 0.0925,
+      "step": 13146
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8557641523023e-05,
+      "loss": 0.0006,
+      "step": 13147
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8555240793201135e-05,
+      "loss": 0.234,
+      "step": 13148
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8552840063379266e-05,
+      "loss": 0.0006,
+      "step": 13149
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.85504393335574e-05,
+      "loss": 0.0006,
+      "step": 13150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8548038603735537e-05,
+      "loss": 0.0006,
+      "step": 13151
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8545637873913672e-05,
+      "loss": 0.0007,
+      "step": 13152
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8543237144091804e-05,
+      "loss": 0.0007,
+      "step": 13153
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.854083641426994e-05,
+      "loss": 0.0006,
+      "step": 13154
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8538435684448074e-05,
+      "loss": 0.0008,
+      "step": 13155
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.853603495462621e-05,
+      "loss": 0.0009,
+      "step": 13156
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.853363422480434e-05,
+      "loss": 0.0007,
+      "step": 13157
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8531233494982476e-05,
+      "loss": 0.0007,
+      "step": 13158
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8528832765160608e-05,
+      "loss": 0.0006,
+      "step": 13159
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8526432035338743e-05,
+      "loss": 0.0008,
+      "step": 13160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.852403130551688e-05,
+      "loss": 0.7701,
+      "step": 13161
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8521630575695014e-05,
+      "loss": 0.0007,
+      "step": 13162
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8519229845873146e-05,
+      "loss": 0.001,
+      "step": 13163
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.851682911605128e-05,
+      "loss": 0.0007,
+      "step": 13164
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8514428386229416e-05,
+      "loss": 0.2792,
+      "step": 13165
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8512027656407548e-05,
+      "loss": 0.0008,
+      "step": 13166
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8509626926585683e-05,
+      "loss": 0.0007,
+      "step": 13167
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8507226196763815e-05,
+      "loss": 0.0008,
+      "step": 13168
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8504825466941953e-05,
+      "loss": 0.0011,
+      "step": 13169
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8502424737120085e-05,
+      "loss": 0.0007,
+      "step": 13170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.850002400729822e-05,
+      "loss": 0.0008,
+      "step": 13171
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8497623277476352e-05,
+      "loss": 0.0009,
+      "step": 13172
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8495222547654487e-05,
+      "loss": 0.001,
+      "step": 13173
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8492821817832622e-05,
+      "loss": 0.0006,
+      "step": 13174
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8490421088010758e-05,
+      "loss": 0.0007,
+      "step": 13175
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.848802035818889e-05,
+      "loss": 0.0761,
+      "step": 13176
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8485619628367025e-05,
+      "loss": 0.0007,
+      "step": 13177
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.848321889854516e-05,
+      "loss": 0.0009,
+      "step": 13178
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8480818168723295e-05,
+      "loss": 0.0007,
+      "step": 13179
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8478417438901427e-05,
+      "loss": 0.0009,
+      "step": 13180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8476016709079562e-05,
+      "loss": 0.0009,
+      "step": 13181
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8473615979257694e-05,
+      "loss": 0.0008,
+      "step": 13182
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.847121524943583e-05,
+      "loss": 0.0008,
+      "step": 13183
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8468814519613964e-05,
+      "loss": 0.0008,
+      "step": 13184
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8466413789792096e-05,
+      "loss": 0.0008,
+      "step": 13185
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.846401305997023e-05,
+      "loss": 0.0007,
+      "step": 13186
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8461612330148366e-05,
+      "loss": 0.0009,
+      "step": 13187
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.84592116003265e-05,
+      "loss": 0.0008,
+      "step": 13188
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8456810870504633e-05,
+      "loss": 0.0008,
+      "step": 13189
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.845441014068277e-05,
+      "loss": 0.0007,
+      "step": 13190
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.84520094108609e-05,
+      "loss": 0.0007,
+      "step": 13191
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.844960868103904e-05,
+      "loss": 0.0008,
+      "step": 13192
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.844720795121717e-05,
+      "loss": 0.0009,
+      "step": 13193
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8444807221395306e-05,
+      "loss": 0.0007,
+      "step": 13194
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8442406491573438e-05,
+      "loss": 0.0041,
+      "step": 13195
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8440005761751573e-05,
+      "loss": 0.0008,
+      "step": 13196
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8437605031929708e-05,
+      "loss": 0.0007,
+      "step": 13197
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8435204302107843e-05,
+      "loss": 0.0007,
+      "step": 13198
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8432803572285975e-05,
+      "loss": 0.0008,
+      "step": 13199
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.843040284246411e-05,
+      "loss": 0.0008,
+      "step": 13200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8428002112642245e-05,
+      "loss": 0.8666,
+      "step": 13201
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8425601382820377e-05,
+      "loss": 0.0065,
+      "step": 13202
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8423200652998512e-05,
+      "loss": 0.0009,
+      "step": 13203
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8420799923176648e-05,
+      "loss": 0.001,
+      "step": 13204
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.841839919335478e-05,
+      "loss": 0.0007,
+      "step": 13205
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8415998463532915e-05,
+      "loss": 0.0008,
+      "step": 13206
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.841359773371105e-05,
+      "loss": 0.0399,
+      "step": 13207
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.841119700388918e-05,
+      "loss": 0.0008,
+      "step": 13208
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8408796274067317e-05,
+      "loss": 0.0011,
+      "step": 13209
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8406395544245452e-05,
+      "loss": 0.0007,
+      "step": 13210
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8403994814423587e-05,
+      "loss": 0.0009,
+      "step": 13211
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.840159408460172e-05,
+      "loss": 0.0093,
+      "step": 13212
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8399193354779854e-05,
+      "loss": 0.0206,
+      "step": 13213
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8396792624957986e-05,
+      "loss": 0.0009,
+      "step": 13214
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8394391895136124e-05,
+      "loss": 0.0011,
+      "step": 13215
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8391991165314256e-05,
+      "loss": 0.0005,
+      "step": 13216
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.838959043549239e-05,
+      "loss": 0.0008,
+      "step": 13217
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8387189705670523e-05,
+      "loss": 0.0006,
+      "step": 13218
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8384788975848662e-05,
+      "loss": 0.0007,
+      "step": 13219
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8382388246026794e-05,
+      "loss": 0.0009,
+      "step": 13220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.837998751620493e-05,
+      "loss": 0.0008,
+      "step": 13221
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.837758678638306e-05,
+      "loss": 0.0006,
+      "step": 13222
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8375186056561196e-05,
+      "loss": 0.0006,
+      "step": 13223
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.837278532673933e-05,
+      "loss": 0.0009,
+      "step": 13224
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8370384596917463e-05,
+      "loss": 0.0008,
+      "step": 13225
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8367983867095598e-05,
+      "loss": 0.0007,
+      "step": 13226
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.836558313727373e-05,
+      "loss": 0.9629,
+      "step": 13227
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8363182407451865e-05,
+      "loss": 0.0008,
+      "step": 13228
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.836078167763e-05,
+      "loss": 0.0006,
+      "step": 13229
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8358380947808135e-05,
+      "loss": 0.0005,
+      "step": 13230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8355980217986267e-05,
+      "loss": 0.0009,
+      "step": 13231
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8353579488164402e-05,
+      "loss": 0.0007,
+      "step": 13232
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8351178758342538e-05,
+      "loss": 0.9253,
+      "step": 13233
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8348778028520673e-05,
+      "loss": 0.0008,
+      "step": 13234
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8346377298698805e-05,
+      "loss": 0.0007,
+      "step": 13235
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.834397656887694e-05,
+      "loss": 0.0008,
+      "step": 13236
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.834157583905507e-05,
+      "loss": 0.0009,
+      "step": 13237
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.833917510923321e-05,
+      "loss": 0.0008,
+      "step": 13238
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8336774379411342e-05,
+      "loss": 0.0007,
+      "step": 13239
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8334373649589477e-05,
+      "loss": 0.9426,
+      "step": 13240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.833197291976761e-05,
+      "loss": 0.0008,
+      "step": 13241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8329572189945744e-05,
+      "loss": 0.0007,
+      "step": 13242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.832717146012388e-05,
+      "loss": 0.0008,
+      "step": 13243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.832477073030201e-05,
+      "loss": 0.0006,
+      "step": 13244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8322370000480146e-05,
+      "loss": 0.0006,
+      "step": 13245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.831996927065828e-05,
+      "loss": 0.0009,
+      "step": 13246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8317568540836417e-05,
+      "loss": 0.001,
+      "step": 13247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.831516781101455e-05,
+      "loss": 0.0008,
+      "step": 13248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8312767081192684e-05,
+      "loss": 0.0008,
+      "step": 13249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8310366351370815e-05,
+      "loss": 0.0011,
+      "step": 13250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8307965621548954e-05,
+      "loss": 0.4578,
+      "step": 13251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8305564891727086e-05,
+      "loss": 0.0008,
+      "step": 13252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.830316416190522e-05,
+      "loss": 0.0008,
+      "step": 13253
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8300763432083353e-05,
+      "loss": 0.001,
+      "step": 13254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8298362702261488e-05,
+      "loss": 0.0009,
+      "step": 13255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8295961972439623e-05,
+      "loss": 0.0008,
+      "step": 13256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.829356124261776e-05,
+      "loss": 0.0008,
+      "step": 13257
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.829116051279589e-05,
+      "loss": 0.0009,
+      "step": 13258
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8288759782974025e-05,
+      "loss": 0.001,
+      "step": 13259
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8286359053152157e-05,
+      "loss": 0.0008,
+      "step": 13260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8283958323330296e-05,
+      "loss": 0.0009,
+      "step": 13261
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8281557593508428e-05,
+      "loss": 0.0009,
+      "step": 13262
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8279156863686563e-05,
+      "loss": 0.0134,
+      "step": 13263
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8276756133864695e-05,
+      "loss": 0.0009,
+      "step": 13264
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.827435540404283e-05,
+      "loss": 0.8221,
+      "step": 13265
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8271954674220965e-05,
+      "loss": 0.0008,
+      "step": 13266
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8269553944399097e-05,
+      "loss": 0.0008,
+      "step": 13267
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8267153214577232e-05,
+      "loss": 0.0007,
+      "step": 13268
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8264752484755364e-05,
+      "loss": 0.0009,
+      "step": 13269
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8262351754933502e-05,
+      "loss": 0.0009,
+      "step": 13270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8259951025111634e-05,
+      "loss": 0.0009,
+      "step": 13271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.825755029528977e-05,
+      "loss": 0.0009,
+      "step": 13272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.82551495654679e-05,
+      "loss": 0.0009,
+      "step": 13273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.825274883564604e-05,
+      "loss": 0.3215,
+      "step": 13274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.825034810582417e-05,
+      "loss": 0.001,
+      "step": 13275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8247947376002307e-05,
+      "loss": 0.0007,
+      "step": 13276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824554664618044e-05,
+      "loss": 0.0008,
+      "step": 13277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8243145916358574e-05,
+      "loss": 0.0009,
+      "step": 13278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824074518653671e-05,
+      "loss": 0.0009,
+      "step": 13279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8238344456714844e-05,
+      "loss": 0.0009,
+      "step": 13280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8235943726892976e-05,
+      "loss": 0.0008,
+      "step": 13281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.823354299707111e-05,
+      "loss": 0.001,
+      "step": 13282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8231142267249246e-05,
+      "loss": 0.0009,
+      "step": 13283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8228741537427378e-05,
+      "loss": 0.0009,
+      "step": 13284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8226340807605513e-05,
+      "loss": 0.0008,
+      "step": 13285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.822394007778365e-05,
+      "loss": 0.0008,
+      "step": 13286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.822153934796178e-05,
+      "loss": 0.0009,
+      "step": 13287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8219138618139915e-05,
+      "loss": 0.0009,
+      "step": 13288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.821673788831805e-05,
+      "loss": 0.0011,
+      "step": 13289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8214337158496182e-05,
+      "loss": 0.273,
+      "step": 13290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8211936428674317e-05,
+      "loss": 0.001,
+      "step": 13291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.820953569885245e-05,
+      "loss": 0.0009,
+      "step": 13292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8207134969030588e-05,
+      "loss": 0.0008,
+      "step": 13293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.820473423920872e-05,
+      "loss": 0.001,
+      "step": 13294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8202333509386855e-05,
+      "loss": 0.0008,
+      "step": 13295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8199932779564987e-05,
+      "loss": 0.001,
+      "step": 13296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8197532049743125e-05,
+      "loss": 0.0009,
+      "step": 13297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8195131319921257e-05,
+      "loss": 0.0009,
+      "step": 13298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8192730590099392e-05,
+      "loss": 0.0008,
+      "step": 13299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8190329860277524e-05,
+      "loss": 0.0009,
+      "step": 13300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.818792913045566e-05,
+      "loss": 0.0009,
+      "step": 13301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8185528400633794e-05,
+      "loss": 0.0009,
+      "step": 13302
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.818312767081193e-05,
+      "loss": 0.0008,
+      "step": 13303
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.818072694099006e-05,
+      "loss": 0.0011,
+      "step": 13304
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8178326211168197e-05,
+      "loss": 0.0007,
+      "step": 13305
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8175925481346332e-05,
+      "loss": 0.0009,
+      "step": 13306
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8173524751524464e-05,
+      "loss": 0.0009,
+      "step": 13307
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.81711240217026e-05,
+      "loss": 0.001,
+      "step": 13308
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.816872329188073e-05,
+      "loss": 0.2574,
+      "step": 13309
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8166322562058866e-05,
+      "loss": 0.0009,
+      "step": 13310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8163921832237e-05,
+      "loss": 0.0008,
+      "step": 13311
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8161521102415136e-05,
+      "loss": 0.0007,
+      "step": 13312
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8159120372593268e-05,
+      "loss": 0.0009,
+      "step": 13313
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8156719642771403e-05,
+      "loss": 0.0008,
+      "step": 13314
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8154318912949538e-05,
+      "loss": 0.0012,
+      "step": 13315
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8151918183127673e-05,
+      "loss": 0.0009,
+      "step": 13316
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8149517453305805e-05,
+      "loss": 0.0008,
+      "step": 13317
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.814711672348394e-05,
+      "loss": 0.0008,
+      "step": 13318
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8144715993662072e-05,
+      "loss": 0.001,
+      "step": 13319
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.814231526384021e-05,
+      "loss": 0.0007,
+      "step": 13320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8139914534018343e-05,
+      "loss": 0.0008,
+      "step": 13321
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8137513804196478e-05,
+      "loss": 0.001,
+      "step": 13322
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.813511307437461e-05,
+      "loss": 0.0008,
+      "step": 13323
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8132712344552745e-05,
+      "loss": 0.001,
+      "step": 13324
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.813031161473088e-05,
+      "loss": 0.0008,
+      "step": 13325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8127910884909012e-05,
+      "loss": 0.0009,
+      "step": 13326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8125510155087147e-05,
+      "loss": 0.8099,
+      "step": 13327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8123109425265282e-05,
+      "loss": 0.0006,
+      "step": 13328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8120708695443417e-05,
+      "loss": 0.0008,
+      "step": 13329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.811830796562155e-05,
+      "loss": 0.0008,
+      "step": 13330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8115907235799684e-05,
+      "loss": 0.0008,
+      "step": 13331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8113506505977816e-05,
+      "loss": 0.0008,
+      "step": 13332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.811110577615595e-05,
+      "loss": 0.0008,
+      "step": 13333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8108705046334087e-05,
+      "loss": 0.001,
+      "step": 13334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8106304316512222e-05,
+      "loss": 0.001,
+      "step": 13335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8103903586690354e-05,
+      "loss": 0.0009,
+      "step": 13336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.810150285686849e-05,
+      "loss": 0.001,
+      "step": 13337
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8099102127046624e-05,
+      "loss": 0.0006,
+      "step": 13338
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.809670139722476e-05,
+      "loss": 0.0006,
+      "step": 13339
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.809430066740289e-05,
+      "loss": 0.0007,
+      "step": 13340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8091899937581026e-05,
+      "loss": 0.001,
+      "step": 13341
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8089499207759158e-05,
+      "loss": 0.0009,
+      "step": 13342
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8087098477937296e-05,
+      "loss": 0.0007,
+      "step": 13343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8084697748115428e-05,
+      "loss": 0.0008,
+      "step": 13344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8082297018293563e-05,
+      "loss": 0.0007,
+      "step": 13345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8079896288471695e-05,
+      "loss": 0.0008,
+      "step": 13346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.807749555864983e-05,
+      "loss": 0.293,
+      "step": 13347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8075094828827966e-05,
+      "loss": 0.0007,
+      "step": 13348
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8072694099006097e-05,
+      "loss": 0.0007,
+      "step": 13349
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8070293369184233e-05,
+      "loss": 0.0006,
+      "step": 13350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8067892639362364e-05,
+      "loss": 0.0006,
+      "step": 13351
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8065491909540503e-05,
+      "loss": 0.0007,
+      "step": 13352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8063091179718635e-05,
+      "loss": 0.0008,
+      "step": 13353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.806069044989677e-05,
+      "loss": 0.0009,
+      "step": 13354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8058289720074902e-05,
+      "loss": 0.0007,
+      "step": 13355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8055888990253037e-05,
+      "loss": 0.0007,
+      "step": 13356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8053488260431172e-05,
+      "loss": 0.0007,
+      "step": 13357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8051087530609307e-05,
+      "loss": 0.0006,
+      "step": 13358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.804868680078744e-05,
+      "loss": 0.001,
+      "step": 13359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8046286070965574e-05,
+      "loss": 0.0008,
+      "step": 13360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.804388534114371e-05,
+      "loss": 0.0009,
+      "step": 13361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8041484611321845e-05,
+      "loss": 0.0007,
+      "step": 13362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8039083881499977e-05,
+      "loss": 0.0006,
+      "step": 13363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8036683151678112e-05,
+      "loss": 0.0008,
+      "step": 13364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8034282421856244e-05,
+      "loss": 0.0008,
+      "step": 13365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.803188169203438e-05,
+      "loss": 0.0007,
+      "step": 13366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8029480962212514e-05,
+      "loss": 0.0006,
+      "step": 13367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8027080232390646e-05,
+      "loss": 0.0006,
+      "step": 13368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.802467950256878e-05,
+      "loss": 0.0006,
+      "step": 13369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8022278772746916e-05,
+      "loss": 0.0007,
+      "step": 13370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.801987804292505e-05,
+      "loss": 0.0005,
+      "step": 13371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8017477313103183e-05,
+      "loss": 0.0007,
+      "step": 13372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8015076583281318e-05,
+      "loss": 0.0006,
+      "step": 13373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.801267585345945e-05,
+      "loss": 0.0007,
+      "step": 13374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.801027512363759e-05,
+      "loss": 0.0007,
+      "step": 13375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.800787439381572e-05,
+      "loss": 0.0006,
+      "step": 13376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8005473663993856e-05,
+      "loss": 0.0006,
+      "step": 13377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8003072934171987e-05,
+      "loss": 0.0007,
+      "step": 13378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8000672204350123e-05,
+      "loss": 0.0007,
+      "step": 13379
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7998271474528258e-05,
+      "loss": 0.826,
+      "step": 13380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7995870744706393e-05,
+      "loss": 0.0007,
+      "step": 13381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7993470014884525e-05,
+      "loss": 0.0006,
+      "step": 13382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.799106928506266e-05,
+      "loss": 0.0006,
+      "step": 13383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7988668555240795e-05,
+      "loss": 0.0011,
+      "step": 13384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.798626782541893e-05,
+      "loss": 0.0007,
+      "step": 13385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7983867095597062e-05,
+      "loss": 0.0006,
+      "step": 13386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7981466365775197e-05,
+      "loss": 0.0006,
+      "step": 13387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.797906563595333e-05,
+      "loss": 0.0007,
+      "step": 13388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7976664906131464e-05,
+      "loss": 0.0006,
+      "step": 13389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.79742641763096e-05,
+      "loss": 0.0006,
+      "step": 13390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.797186344648773e-05,
+      "loss": 0.0005,
+      "step": 13391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7969462716665867e-05,
+      "loss": 0.0006,
+      "step": 13392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7967061986844e-05,
+      "loss": 0.0006,
+      "step": 13393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7964661257022137e-05,
+      "loss": 0.0006,
+      "step": 13394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.796226052720027e-05,
+      "loss": 0.0005,
+      "step": 13395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7959859797378404e-05,
+      "loss": 0.0007,
+      "step": 13396
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7957459067556536e-05,
+      "loss": 0.0006,
+      "step": 13397
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7955058337734674e-05,
+      "loss": 0.0005,
+      "step": 13398
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7952657607912806e-05,
+      "loss": 0.0006,
+      "step": 13399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.795025687809094e-05,
+      "loss": 0.0006,
+      "step": 13400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7947856148269073e-05,
+      "loss": 0.0005,
+      "step": 13401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.794545541844721e-05,
+      "loss": 0.9944,
+      "step": 13402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7943054688625343e-05,
+      "loss": 0.0007,
+      "step": 13403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.794065395880348e-05,
+      "loss": 0.0007,
+      "step": 13404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.793825322898161e-05,
+      "loss": 0.0008,
+      "step": 13405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7935852499159746e-05,
+      "loss": 0.753,
+      "step": 13406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.793345176933788e-05,
+      "loss": 0.0005,
+      "step": 13407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7931051039516013e-05,
+      "loss": 0.0006,
+      "step": 13408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7928650309694148e-05,
+      "loss": 0.0006,
+      "step": 13409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7926249579872283e-05,
+      "loss": 0.0006,
+      "step": 13410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7923848850050415e-05,
+      "loss": 0.0005,
+      "step": 13411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.792144812022855e-05,
+      "loss": 0.0005,
+      "step": 13412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7919047390406685e-05,
+      "loss": 0.1124,
+      "step": 13413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7916646660584817e-05,
+      "loss": 0.0006,
+      "step": 13414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7914245930762952e-05,
+      "loss": 1.4221,
+      "step": 13415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7911845200941087e-05,
+      "loss": 0.6272,
+      "step": 13416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7909444471119222e-05,
+      "loss": 0.0008,
+      "step": 13417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7907043741297354e-05,
+      "loss": 0.0007,
+      "step": 13418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.790464301147549e-05,
+      "loss": 0.2872,
+      "step": 13419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.790224228165362e-05,
+      "loss": 0.0007,
+      "step": 13420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789984155183176e-05,
+      "loss": 0.1262,
+      "step": 13421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789744082200989e-05,
+      "loss": 0.0008,
+      "step": 13422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7895040092188027e-05,
+      "loss": 0.0007,
+      "step": 13423
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789263936236616e-05,
+      "loss": 0.0418,
+      "step": 13424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7890238632544297e-05,
+      "loss": 0.0007,
+      "step": 13425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.788783790272243e-05,
+      "loss": 0.0008,
+      "step": 13426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7885437172900564e-05,
+      "loss": 0.0008,
+      "step": 13427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7883036443078696e-05,
+      "loss": 0.0007,
+      "step": 13428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.788063571325683e-05,
+      "loss": 0.0008,
+      "step": 13429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7878234983434966e-05,
+      "loss": 0.0008,
+      "step": 13430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7875834253613098e-05,
+      "loss": 0.0007,
+      "step": 13431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7873433523791233e-05,
+      "loss": 0.0267,
+      "step": 13432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7871032793969365e-05,
+      "loss": 0.0007,
+      "step": 13433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7868632064147504e-05,
+      "loss": 0.0164,
+      "step": 13434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7866231334325636e-05,
+      "loss": 0.0007,
+      "step": 13435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.786383060450377e-05,
+      "loss": 0.0113,
+      "step": 13436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7861429874681903e-05,
+      "loss": 0.9455,
+      "step": 13437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7859029144860038e-05,
+      "loss": 0.001,
+      "step": 13438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7856628415038173e-05,
+      "loss": 0.0154,
+      "step": 13439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7854227685216308e-05,
+      "loss": 0.0089,
+      "step": 13440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.785182695539444e-05,
+      "loss": 0.0015,
+      "step": 13441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7849426225572575e-05,
+      "loss": 0.0034,
+      "step": 13442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7847025495750707e-05,
+      "loss": 0.0019,
+      "step": 13443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7844624765928845e-05,
+      "loss": 0.0015,
+      "step": 13444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7842224036106977e-05,
+      "loss": 0.003,
+      "step": 13445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7839823306285112e-05,
+      "loss": 0.0013,
+      "step": 13446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7837422576463244e-05,
+      "loss": 0.0016,
+      "step": 13447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.783502184664138e-05,
+      "loss": 0.001,
+      "step": 13448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7832621116819515e-05,
+      "loss": 0.0014,
+      "step": 13449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7830220386997646e-05,
+      "loss": 0.0009,
+      "step": 13450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.782781965717578e-05,
+      "loss": 0.006,
+      "step": 13451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7825418927353917e-05,
+      "loss": 0.0008,
+      "step": 13452
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7823018197532052e-05,
+      "loss": 0.8702,
+      "step": 13453
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7820617467710184e-05,
+      "loss": 0.0008,
+      "step": 13454
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.781821673788832e-05,
+      "loss": 0.0009,
+      "step": 13455
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.781581600806645e-05,
+      "loss": 0.0039,
+      "step": 13456
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.781341527824459e-05,
+      "loss": 0.001,
+      "step": 13457
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.781101454842272e-05,
+      "loss": 0.0012,
+      "step": 13458
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7808613818600856e-05,
+      "loss": 0.001,
+      "step": 13459
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7806213088778988e-05,
+      "loss": 0.0007,
+      "step": 13460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7803812358957123e-05,
+      "loss": 0.0011,
+      "step": 13461
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.780141162913526e-05,
+      "loss": 0.0102,
+      "step": 13462
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7799010899313394e-05,
+      "loss": 0.0009,
+      "step": 13463
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7796610169491526e-05,
+      "loss": 0.0011,
+      "step": 13464
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.779420943966966e-05,
+      "loss": 0.001,
+      "step": 13465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7791808709847796e-05,
+      "loss": 0.0008,
+      "step": 13466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.778940798002593e-05,
+      "loss": 0.0011,
+      "step": 13467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7787007250204063e-05,
+      "loss": 0.0008,
+      "step": 13468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7784606520382198e-05,
+      "loss": 0.0009,
+      "step": 13469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.778220579056033e-05,
+      "loss": 0.0014,
+      "step": 13470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7779805060738465e-05,
+      "loss": 0.0011,
+      "step": 13471
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.77774043309166e-05,
+      "loss": 0.0006,
+      "step": 13472
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7775003601094732e-05,
+      "loss": 0.0007,
+      "step": 13473
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7772602871272867e-05,
+      "loss": 0.0009,
+      "step": 13474
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7770202141451e-05,
+      "loss": 0.0025,
+      "step": 13475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7767801411629138e-05,
+      "loss": 0.001,
+      "step": 13476
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.776540068180727e-05,
+      "loss": 0.0007,
+      "step": 13477
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7762999951985405e-05,
+      "loss": 0.001,
+      "step": 13478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7760599222163536e-05,
+      "loss": 0.0023,
+      "step": 13479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7758198492341675e-05,
+      "loss": 0.0006,
+      "step": 13480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7755797762519807e-05,
+      "loss": 0.9224,
+      "step": 13481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7753397032697942e-05,
+      "loss": 0.0007,
+      "step": 13482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7750996302876074e-05,
+      "loss": 0.0008,
+      "step": 13483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.774859557305421e-05,
+      "loss": 0.0008,
+      "step": 13484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7746194843232344e-05,
+      "loss": 0.0007,
+      "step": 13485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.774379411341048e-05,
+      "loss": 0.0028,
+      "step": 13486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.774139338358861e-05,
+      "loss": 0.0031,
+      "step": 13487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7738992653766746e-05,
+      "loss": 0.4204,
+      "step": 13488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.773659192394488e-05,
+      "loss": 0.0008,
+      "step": 13489
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7734191194123013e-05,
+      "loss": 0.001,
+      "step": 13490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.773179046430115e-05,
+      "loss": 0.0008,
+      "step": 13491
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.772938973447928e-05,
+      "loss": 0.0006,
+      "step": 13492
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7726989004657416e-05,
+      "loss": 0.0009,
+      "step": 13493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.772458827483555e-05,
+      "loss": 0.0033,
+      "step": 13494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7722187545013686e-05,
+      "loss": 0.0009,
+      "step": 13495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7719786815191818e-05,
+      "loss": 0.0009,
+      "step": 13496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7717386085369953e-05,
+      "loss": 0.0009,
+      "step": 13497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7714985355548088e-05,
+      "loss": 0.0007,
+      "step": 13498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7712584625726223e-05,
+      "loss": 0.0026,
+      "step": 13499
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7710183895904355e-05,
+      "loss": 0.0017,
+      "step": 13500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.770778316608249e-05,
+      "loss": 0.0009,
+      "step": 13501
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7705382436260622e-05,
+      "loss": 0.0008,
+      "step": 13502
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.770298170643876e-05,
+      "loss": 0.0011,
+      "step": 13503
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7700580976616892e-05,
+      "loss": 0.0009,
+      "step": 13504
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7698180246795028e-05,
+      "loss": 0.0009,
+      "step": 13505
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.769577951697316e-05,
+      "loss": 0.0008,
+      "step": 13506
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7693378787151295e-05,
+      "loss": 0.0008,
+      "step": 13507
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.769097805732943e-05,
+      "loss": 0.001,
+      "step": 13508
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7688577327507565e-05,
+      "loss": 0.9634,
+      "step": 13509
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7686176597685697e-05,
+      "loss": 0.0019,
+      "step": 13510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7683775867863832e-05,
+      "loss": 0.0008,
+      "step": 13511
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7681375138041967e-05,
+      "loss": 0.0008,
+      "step": 13512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.76789744082201e-05,
+      "loss": 0.0018,
+      "step": 13513
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7676573678398234e-05,
+      "loss": 0.0008,
+      "step": 13514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7674172948576366e-05,
+      "loss": 0.0007,
+      "step": 13515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.76717722187545e-05,
+      "loss": 0.0009,
+      "step": 13516
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7669371488932636e-05,
+      "loss": 0.0008,
+      "step": 13517
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.766697075911077e-05,
+      "loss": 0.001,
+      "step": 13518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7664570029288903e-05,
+      "loss": 0.0007,
+      "step": 13519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.766216929946704e-05,
+      "loss": 0.0007,
+      "step": 13520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7659768569645174e-05,
+      "loss": 0.0008,
+      "step": 13521
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.765736783982331e-05,
+      "loss": 0.0008,
+      "step": 13522
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.765496711000144e-05,
+      "loss": 0.0015,
+      "step": 13523
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7652566380179576e-05,
+      "loss": 0.0017,
+      "step": 13524
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7650165650357708e-05,
+      "loss": 0.0008,
+      "step": 13525
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7647764920535846e-05,
+      "loss": 0.0026,
+      "step": 13526
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7645364190713978e-05,
+      "loss": 0.0008,
+      "step": 13527
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7642963460892113e-05,
+      "loss": 0.0007,
+      "step": 13528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7640562731070245e-05,
+      "loss": 0.0007,
+      "step": 13529
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.763816200124838e-05,
+      "loss": 0.0008,
+      "step": 13530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7635761271426515e-05,
+      "loss": 0.0007,
+      "step": 13531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7633360541604647e-05,
+      "loss": 0.0007,
+      "step": 13532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7630959811782782e-05,
+      "loss": 0.0006,
+      "step": 13533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7628559081960918e-05,
+      "loss": 0.0019,
+      "step": 13534
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7626158352139053e-05,
+      "loss": 0.0006,
+      "step": 13535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7623757622317185e-05,
+      "loss": 0.0007,
+      "step": 13536
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.762135689249532e-05,
+      "loss": 0.0006,
+      "step": 13537
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.761895616267345e-05,
+      "loss": 0.0007,
+      "step": 13538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7616555432851587e-05,
+      "loss": 0.0007,
+      "step": 13539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7614154703029722e-05,
+      "loss": 0.0008,
+      "step": 13540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7611753973207857e-05,
+      "loss": 0.0006,
+      "step": 13541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.760935324338599e-05,
+      "loss": 0.9757,
+      "step": 13542
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7606952513564124e-05,
+      "loss": 0.0013,
+      "step": 13543
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.760455178374226e-05,
+      "loss": 0.3707,
+      "step": 13544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7602151053920394e-05,
+      "loss": 0.0008,
+      "step": 13545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7599750324098526e-05,
+      "loss": 0.7636,
+      "step": 13546
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.759734959427666e-05,
+      "loss": 0.0154,
+      "step": 13547
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7594948864454793e-05,
+      "loss": 0.0006,
+      "step": 13548
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7592548134632932e-05,
+      "loss": 0.0005,
+      "step": 13549
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7590147404811064e-05,
+      "loss": 0.0006,
+      "step": 13550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.75877466749892e-05,
+      "loss": 0.0007,
+      "step": 13551
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.758534594516733e-05,
+      "loss": 0.0008,
+      "step": 13552
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7582945215345466e-05,
+      "loss": 0.0016,
+      "step": 13553
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.75805444855236e-05,
+      "loss": 0.0007,
+      "step": 13554
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7578143755701733e-05,
+      "loss": 0.0007,
+      "step": 13555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7575743025879868e-05,
+      "loss": 0.0006,
+      "step": 13556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7573342296058e-05,
+      "loss": 0.0006,
+      "step": 13557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.757094156623614e-05,
+      "loss": 0.0007,
+      "step": 13558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.756854083641427e-05,
+      "loss": 0.0011,
+      "step": 13559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7566140106592405e-05,
+      "loss": 0.672,
+      "step": 13560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7563739376770537e-05,
+      "loss": 0.0007,
+      "step": 13561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7561338646948676e-05,
+      "loss": 0.0028,
+      "step": 13562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7558937917126808e-05,
+      "loss": 0.0016,
+      "step": 13563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7556537187304943e-05,
+      "loss": 0.0005,
+      "step": 13564
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7554136457483075e-05,
+      "loss": 0.0007,
+      "step": 13565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.755173572766121e-05,
+      "loss": 0.0007,
+      "step": 13566
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7549334997839345e-05,
+      "loss": 0.0131,
+      "step": 13567
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.754693426801748e-05,
+      "loss": 0.0007,
+      "step": 13568
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7544533538195612e-05,
+      "loss": 0.0007,
+      "step": 13569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7542132808373747e-05,
+      "loss": 0.0021,
+      "step": 13570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753973207855188e-05,
+      "loss": 0.0006,
+      "step": 13571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7537331348730014e-05,
+      "loss": 0.0006,
+      "step": 13572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753493061890815e-05,
+      "loss": 0.0007,
+      "step": 13573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.753252988908628e-05,
+      "loss": 0.0008,
+      "step": 13574
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7530129159264416e-05,
+      "loss": 0.0007,
+      "step": 13575
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.752772842944255e-05,
+      "loss": 0.0007,
+      "step": 13576
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7525327699620687e-05,
+      "loss": 0.0007,
+      "step": 13577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.752292696979882e-05,
+      "loss": 0.0007,
+      "step": 13578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7520526239976954e-05,
+      "loss": 0.0025,
+      "step": 13579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7518125510155085e-05,
+      "loss": 0.0006,
+      "step": 13580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7515724780333224e-05,
+      "loss": 0.0007,
+      "step": 13581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7513324050511356e-05,
+      "loss": 0.0007,
+      "step": 13582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.751092332068949e-05,
+      "loss": 0.9923,
+      "step": 13583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7508522590867623e-05,
+      "loss": 0.0006,
+      "step": 13584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.750612186104576e-05,
+      "loss": 0.0007,
+      "step": 13585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7503721131223893e-05,
+      "loss": 0.0006,
+      "step": 13586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.750132040140203e-05,
+      "loss": 0.9689,
+      "step": 13587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.749891967158016e-05,
+      "loss": 0.0006,
+      "step": 13588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7496518941758295e-05,
+      "loss": 0.5772,
+      "step": 13589
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.749411821193643e-05,
+      "loss": 0.0008,
+      "step": 13590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7491717482114566e-05,
+      "loss": 0.0025,
+      "step": 13591
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7489316752292698e-05,
+      "loss": 0.0035,
+      "step": 13592
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7486916022470833e-05,
+      "loss": 0.0007,
+      "step": 13593
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7484515292648968e-05,
+      "loss": 0.0008,
+      "step": 13594
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.74821145628271e-05,
+      "loss": 0.0007,
+      "step": 13595
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7479713833005235e-05,
+      "loss": 0.0008,
+      "step": 13596
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7477313103183367e-05,
+      "loss": 0.0007,
+      "step": 13597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7474912373361502e-05,
+      "loss": 0.0009,
+      "step": 13598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7472511643539637e-05,
+      "loss": 0.0008,
+      "step": 13599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7470110913717772e-05,
+      "loss": 0.0008,
+      "step": 13600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7467710183895904e-05,
+      "loss": 0.0058,
+      "step": 13601
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.746530945407404e-05,
+      "loss": 0.003,
+      "step": 13602
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.746290872425217e-05,
+      "loss": 0.0009,
+      "step": 13603
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.746050799443031e-05,
+      "loss": 0.0008,
+      "step": 13604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.745810726460844e-05,
+      "loss": 0.0008,
+      "step": 13605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7455706534786577e-05,
+      "loss": 0.0035,
+      "step": 13606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.745330580496471e-05,
+      "loss": 0.0008,
+      "step": 13607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7450905075142847e-05,
+      "loss": 0.0033,
+      "step": 13608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.744850434532098e-05,
+      "loss": 0.0008,
+      "step": 13609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7446103615499114e-05,
+      "loss": 0.0008,
+      "step": 13610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7443702885677246e-05,
+      "loss": 0.9266,
+      "step": 13611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.744130215585538e-05,
+      "loss": 0.0101,
+      "step": 13612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7438901426033516e-05,
+      "loss": 0.0009,
+      "step": 13613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7436500696211648e-05,
+      "loss": 0.0008,
+      "step": 13614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7434099966389783e-05,
+      "loss": 0.0007,
+      "step": 13615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7431699236567915e-05,
+      "loss": 0.0008,
+      "step": 13616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7429298506746054e-05,
+      "loss": 0.0145,
+      "step": 13617
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7426897776924185e-05,
+      "loss": 0.0009,
+      "step": 13618
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.742449704710232e-05,
+      "loss": 0.001,
+      "step": 13619
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7422096317280452e-05,
+      "loss": 0.0008,
+      "step": 13620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7419695587458587e-05,
+      "loss": 0.0009,
+      "step": 13621
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7417294857636723e-05,
+      "loss": 0.0008,
+      "step": 13622
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7414894127814858e-05,
+      "loss": 0.0008,
+      "step": 13623
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.741249339799299e-05,
+      "loss": 0.0008,
+      "step": 13624
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7410092668171125e-05,
+      "loss": 0.0007,
+      "step": 13625
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.740769193834926e-05,
+      "loss": 0.367,
+      "step": 13626
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7405291208527395e-05,
+      "loss": 0.0011,
+      "step": 13627
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7402890478705527e-05,
+      "loss": 0.0038,
+      "step": 13628
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7400489748883662e-05,
+      "loss": 0.0038,
+      "step": 13629
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7398089019061794e-05,
+      "loss": 0.0009,
+      "step": 13630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.739568828923993e-05,
+      "loss": 0.0011,
+      "step": 13631
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7393287559418064e-05,
+      "loss": 0.0008,
+      "step": 13632
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.73908868295962e-05,
+      "loss": 0.0008,
+      "step": 13633
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.738848609977433e-05,
+      "loss": 0.0008,
+      "step": 13634
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7386085369952467e-05,
+      "loss": 0.0008,
+      "step": 13635
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7383684640130602e-05,
+      "loss": 0.0007,
+      "step": 13636
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7381283910308734e-05,
+      "loss": 0.0008,
+      "step": 13637
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737888318048687e-05,
+      "loss": 0.0007,
+      "step": 13638
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7376482450665e-05,
+      "loss": 0.002,
+      "step": 13639
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737408172084314e-05,
+      "loss": 0.0006,
+      "step": 13640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.737168099102127e-05,
+      "loss": 0.0132,
+      "step": 13641
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7369280261199406e-05,
+      "loss": 0.0007,
+      "step": 13642
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7366879531377538e-05,
+      "loss": 0.0031,
+      "step": 13643
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7364478801555673e-05,
+      "loss": 0.0007,
+      "step": 13644
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7362078071733808e-05,
+      "loss": 0.0007,
+      "step": 13645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7359677341911943e-05,
+      "loss": 0.0009,
+      "step": 13646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7357276612090075e-05,
+      "loss": 0.0007,
+      "step": 13647
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.735487588226821e-05,
+      "loss": 0.0006,
+      "step": 13648
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7352475152446346e-05,
+      "loss": 0.014,
+      "step": 13649
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.735007442262448e-05,
+      "loss": 0.0007,
+      "step": 13650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7347673692802613e-05,
+      "loss": 0.001,
+      "step": 13651
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7345272962980748e-05,
+      "loss": 0.0024,
+      "step": 13652
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.734287223315888e-05,
+      "loss": 0.0092,
+      "step": 13653
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7340471503337015e-05,
+      "loss": 0.002,
+      "step": 13654
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.733807077351515e-05,
+      "loss": 0.0007,
+      "step": 13655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7335670043693282e-05,
+      "loss": 0.0007,
+      "step": 13656
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7333269313871417e-05,
+      "loss": 0.0021,
+      "step": 13657
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7330868584049552e-05,
+      "loss": 0.0007,
+      "step": 13658
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7328467854227687e-05,
+      "loss": 0.0007,
+      "step": 13659
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.732606712440582e-05,
+      "loss": 0.0018,
+      "step": 13660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7323666394583954e-05,
+      "loss": 0.0016,
+      "step": 13661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7321265664762086e-05,
+      "loss": 0.0007,
+      "step": 13662
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7318864934940225e-05,
+      "loss": 0.0007,
+      "step": 13663
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7316464205118357e-05,
+      "loss": 0.0006,
+      "step": 13664
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7314063475296492e-05,
+      "loss": 0.0006,
+      "step": 13665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7311662745474624e-05,
+      "loss": 0.3755,
+      "step": 13666
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.730926201565276e-05,
+      "loss": 0.0006,
+      "step": 13667
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7306861285830894e-05,
+      "loss": 0.0008,
+      "step": 13668
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.730446055600903e-05,
+      "loss": 0.0005,
+      "step": 13669
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.730205982618716e-05,
+      "loss": 0.0006,
+      "step": 13670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7299659096365296e-05,
+      "loss": 0.0013,
+      "step": 13671
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.729725836654343e-05,
+      "loss": 0.0007,
+      "step": 13672
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7294857636721566e-05,
+      "loss": 0.0006,
+      "step": 13673
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7292456906899698e-05,
+      "loss": 0.5424,
+      "step": 13674
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7290056177077833e-05,
+      "loss": 0.0006,
+      "step": 13675
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7287655447255965e-05,
+      "loss": 0.0007,
+      "step": 13676
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.72852547174341e-05,
+      "loss": 0.0006,
+      "step": 13677
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7282853987612236e-05,
+      "loss": 0.0006,
+      "step": 13678
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7280453257790367e-05,
+      "loss": 0.002,
+      "step": 13679
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7278052527968503e-05,
+      "loss": 0.0006,
+      "step": 13680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7275651798146638e-05,
+      "loss": 0.0007,
+      "step": 13681
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7273251068324773e-05,
+      "loss": 0.0006,
+      "step": 13682
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7270850338502905e-05,
+      "loss": 0.948,
+      "step": 13683
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.726844960868104e-05,
+      "loss": 0.0018,
+      "step": 13684
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7266048878859172e-05,
+      "loss": 0.0022,
+      "step": 13685
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.726364814903731e-05,
+      "loss": 0.0007,
+      "step": 13686
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7261247419215442e-05,
+      "loss": 0.0007,
+      "step": 13687
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7258846689393577e-05,
+      "loss": 0.0016,
+      "step": 13688
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.725644595957171e-05,
+      "loss": 0.0066,
+      "step": 13689
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7254045229749848e-05,
+      "loss": 0.0094,
+      "step": 13690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.725164449992798e-05,
+      "loss": 0.0016,
+      "step": 13691
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7249243770106115e-05,
+      "loss": 0.0021,
+      "step": 13692
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7246843040284247e-05,
+      "loss": 0.0008,
+      "step": 13693
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7244442310462382e-05,
+      "loss": 0.0016,
+      "step": 13694
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7242041580640517e-05,
+      "loss": 0.0017,
+      "step": 13695
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.723964085081865e-05,
+      "loss": 0.0006,
+      "step": 13696
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7237240120996784e-05,
+      "loss": 0.0005,
+      "step": 13697
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7234839391174916e-05,
+      "loss": 0.0006,
+      "step": 13698
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.723243866135305e-05,
+      "loss": 0.0009,
+      "step": 13699
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7230037931531186e-05,
+      "loss": 0.0006,
+      "step": 13700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.722763720170932e-05,
+      "loss": 0.0013,
+      "step": 13701
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7225236471887453e-05,
+      "loss": 0.0005,
+      "step": 13702
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7222835742065588e-05,
+      "loss": 0.0027,
+      "step": 13703
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7220435012243723e-05,
+      "loss": 0.0007,
+      "step": 13704
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.721803428242186e-05,
+      "loss": 0.0014,
+      "step": 13705
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.721563355259999e-05,
+      "loss": 0.0007,
+      "step": 13706
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7213232822778126e-05,
+      "loss": 0.0007,
+      "step": 13707
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7210832092956257e-05,
+      "loss": 0.0007,
+      "step": 13708
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7208431363134396e-05,
+      "loss": 0.0008,
+      "step": 13709
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7206030633312528e-05,
+      "loss": 0.0013,
+      "step": 13710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7203629903490663e-05,
+      "loss": 0.0006,
+      "step": 13711
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7201229173668795e-05,
+      "loss": 0.9569,
+      "step": 13712
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.719882844384693e-05,
+      "loss": 0.0006,
+      "step": 13713
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7196427714025065e-05,
+      "loss": 0.0131,
+      "step": 13714
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.71940269842032e-05,
+      "loss": 0.0006,
+      "step": 13715
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7191626254381332e-05,
+      "loss": 0.0005,
+      "step": 13716
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7189225524559467e-05,
+      "loss": 0.0006,
+      "step": 13717
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7186824794737603e-05,
+      "loss": 0.0012,
+      "step": 13718
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7184424064915734e-05,
+      "loss": 0.0007,
+      "step": 13719
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.718202333509387e-05,
+      "loss": 0.0017,
+      "step": 13720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7179622605272e-05,
+      "loss": 0.0016,
+      "step": 13721
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7177221875450136e-05,
+      "loss": 0.0006,
+      "step": 13722
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.717482114562827e-05,
+      "loss": 0.0006,
+      "step": 13723
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7172420415806407e-05,
+      "loss": 0.0007,
+      "step": 13724
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.717001968598454e-05,
+      "loss": 0.0007,
+      "step": 13725
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7167618956162674e-05,
+      "loss": 0.0007,
+      "step": 13726
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.716521822634081e-05,
+      "loss": 0.0007,
+      "step": 13727
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7162817496518944e-05,
+      "loss": 0.0007,
+      "step": 13728
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7160416766697076e-05,
+      "loss": 0.0013,
+      "step": 13729
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.715801603687521e-05,
+      "loss": 0.0007,
+      "step": 13730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7155615307053343e-05,
+      "loss": 0.0007,
+      "step": 13731
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.715321457723148e-05,
+      "loss": 0.0007,
+      "step": 13732
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7150813847409613e-05,
+      "loss": 0.0007,
+      "step": 13733
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.714841311758775e-05,
+      "loss": 0.0013,
+      "step": 13734
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.714601238776588e-05,
+      "loss": 0.0093,
+      "step": 13735
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7143611657944016e-05,
+      "loss": 0.0007,
+      "step": 13736
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.714121092812215e-05,
+      "loss": 0.0007,
+      "step": 13737
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7138810198300283e-05,
+      "loss": 0.4034,
+      "step": 13738
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7136409468478418e-05,
+      "loss": 0.001,
+      "step": 13739
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.713400873865655e-05,
+      "loss": 0.0007,
+      "step": 13740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7131608008834688e-05,
+      "loss": 0.0055,
+      "step": 13741
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.712920727901282e-05,
+      "loss": 0.0007,
+      "step": 13742
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7126806549190955e-05,
+      "loss": 0.0007,
+      "step": 13743
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7124405819369087e-05,
+      "loss": 0.0007,
+      "step": 13744
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7122005089547225e-05,
+      "loss": 0.0007,
+      "step": 13745
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7119604359725357e-05,
+      "loss": 0.0006,
+      "step": 13746
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7117203629903492e-05,
+      "loss": 0.0007,
+      "step": 13747
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7114802900081624e-05,
+      "loss": 0.0006,
+      "step": 13748
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.711240217025976e-05,
+      "loss": 0.0012,
+      "step": 13749
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7110001440437895e-05,
+      "loss": 0.0006,
+      "step": 13750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.710760071061603e-05,
+      "loss": 0.0007,
+      "step": 13751
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.710519998079416e-05,
+      "loss": 0.0007,
+      "step": 13752
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7102799250972297e-05,
+      "loss": 0.0007,
+      "step": 13753
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.710039852115043e-05,
+      "loss": 0.0006,
+      "step": 13754
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7097997791328564e-05,
+      "loss": 0.0006,
+      "step": 13755
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.70955970615067e-05,
+      "loss": 0.0006,
+      "step": 13756
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7093196331684834e-05,
+      "loss": 0.001,
+      "step": 13757
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7090795601862966e-05,
+      "loss": 0.0012,
+      "step": 13758
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.70883948720411e-05,
+      "loss": 0.0006,
+      "step": 13759
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7085994142219236e-05,
+      "loss": 0.0016,
+      "step": 13760
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7083593412397368e-05,
+      "loss": 0.0014,
+      "step": 13761
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7081192682575503e-05,
+      "loss": 0.0006,
+      "step": 13762
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7078791952753635e-05,
+      "loss": 0.0007,
+      "step": 13763
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7076391222931774e-05,
+      "loss": 0.0006,
+      "step": 13764
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7073990493109906e-05,
+      "loss": 0.0006,
+      "step": 13765
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.707158976328804e-05,
+      "loss": 0.001,
+      "step": 13766
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7069189033466173e-05,
+      "loss": 0.0007,
+      "step": 13767
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.706678830364431e-05,
+      "loss": 0.0007,
+      "step": 13768
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7064387573822443e-05,
+      "loss": 0.0011,
+      "step": 13769
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7061986844000578e-05,
+      "loss": 0.0005,
+      "step": 13770
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.705958611417871e-05,
+      "loss": 0.0006,
+      "step": 13771
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7057185384356845e-05,
+      "loss": 0.0006,
+      "step": 13772
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.705478465453498e-05,
+      "loss": 0.0005,
+      "step": 13773
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7052383924713115e-05,
+      "loss": 0.0005,
+      "step": 13774
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7049983194891247e-05,
+      "loss": 0.0005,
+      "step": 13775
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7047582465069382e-05,
+      "loss": 0.0007,
+      "step": 13776
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7045181735247518e-05,
+      "loss": 0.0006,
+      "step": 13777
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.704278100542565e-05,
+      "loss": 0.0006,
+      "step": 13778
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7040380275603785e-05,
+      "loss": 0.0005,
+      "step": 13779
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7037979545781916e-05,
+      "loss": 0.0015,
+      "step": 13780
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.703557881596005e-05,
+      "loss": 0.0013,
+      "step": 13781
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7033178086138187e-05,
+      "loss": 0.001,
+      "step": 13782
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7030777356316322e-05,
+      "loss": 0.6528,
+      "step": 13783
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7028376626494454e-05,
+      "loss": 0.0009,
+      "step": 13784
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.702597589667259e-05,
+      "loss": 0.0022,
+      "step": 13785
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.702357516685072e-05,
+      "loss": 0.0006,
+      "step": 13786
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.702117443702886e-05,
+      "loss": 0.0006,
+      "step": 13787
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.701877370720699e-05,
+      "loss": 0.0005,
+      "step": 13788
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7016372977385126e-05,
+      "loss": 0.0006,
+      "step": 13789
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7013972247563258e-05,
+      "loss": 0.0005,
+      "step": 13790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7011571517741397e-05,
+      "loss": 0.0021,
+      "step": 13791
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.700917078791953e-05,
+      "loss": 0.0007,
+      "step": 13792
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7006770058097664e-05,
+      "loss": 0.0019,
+      "step": 13793
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7004369328275796e-05,
+      "loss": 0.0012,
+      "step": 13794
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.700196859845393e-05,
+      "loss": 0.0005,
+      "step": 13795
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6999567868632066e-05,
+      "loss": 0.0005,
+      "step": 13796
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.69971671388102e-05,
+      "loss": 0.0006,
+      "step": 13797
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6994766408988333e-05,
+      "loss": 0.0005,
+      "step": 13798
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6992365679166468e-05,
+      "loss": 0.0005,
+      "step": 13799
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6989964949344603e-05,
+      "loss": 0.0005,
+      "step": 13800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6987564219522735e-05,
+      "loss": 0.0011,
+      "step": 13801
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.698516348970087e-05,
+      "loss": 0.0011,
+      "step": 13802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6982762759879002e-05,
+      "loss": 0.0005,
+      "step": 13803
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6980362030057137e-05,
+      "loss": 0.0006,
+      "step": 13804
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6977961300235272e-05,
+      "loss": 0.0005,
+      "step": 13805
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6975560570413408e-05,
+      "loss": 0.0008,
+      "step": 13806
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.697315984059154e-05,
+      "loss": 0.0087,
+      "step": 13807
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6970759110769675e-05,
+      "loss": 0.4398,
+      "step": 13808
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.696835838094781e-05,
+      "loss": 0.0005,
+      "step": 13809
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6965957651125945e-05,
+      "loss": 0.001,
+      "step": 13810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6963556921304077e-05,
+      "loss": 0.0028,
+      "step": 13811
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6961156191482212e-05,
+      "loss": 0.0005,
+      "step": 13812
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6958755461660344e-05,
+      "loss": 0.0005,
+      "step": 13813
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6956354731838482e-05,
+      "loss": 0.9722,
+      "step": 13814
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6953954002016614e-05,
+      "loss": 0.9575,
+      "step": 13815
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.695155327219475e-05,
+      "loss": 0.9261,
+      "step": 13816
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 0.0012,
+      "step": 13817
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6946751812551016e-05,
+      "loss": 0.0009,
+      "step": 13818
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.694435108272915e-05,
+      "loss": 0.0006,
+      "step": 13819
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6941950352907283e-05,
+      "loss": 0.0007,
+      "step": 13820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.693954962308542e-05,
+      "loss": 0.0007,
+      "step": 13821
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.693714889326355e-05,
+      "loss": 0.0008,
+      "step": 13822
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.693474816344169e-05,
+      "loss": 0.0017,
+      "step": 13823
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.693234743361982e-05,
+      "loss": 0.9241,
+      "step": 13824
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6929946703797956e-05,
+      "loss": 0.001,
+      "step": 13825
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6927545973976088e-05,
+      "loss": 0.0031,
+      "step": 13826
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6925145244154223e-05,
+      "loss": 0.0008,
+      "step": 13827
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6922744514332358e-05,
+      "loss": 0.0014,
+      "step": 13828
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6920343784510493e-05,
+      "loss": 0.0011,
+      "step": 13829
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6917943054688625e-05,
+      "loss": 0.0011,
+      "step": 13830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.691554232486676e-05,
+      "loss": 0.0007,
+      "step": 13831
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6913141595044895e-05,
+      "loss": 0.0011,
+      "step": 13832
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.691074086522303e-05,
+      "loss": 0.0011,
+      "step": 13833
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6908340135401162e-05,
+      "loss": 0.0008,
+      "step": 13834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6905939405579298e-05,
+      "loss": 0.0019,
+      "step": 13835
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.690353867575743e-05,
+      "loss": 0.0007,
+      "step": 13836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6901137945935565e-05,
+      "loss": 0.0021,
+      "step": 13837
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.68987372161137e-05,
+      "loss": 0.0007,
+      "step": 13838
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6896336486291835e-05,
+      "loss": 0.001,
+      "step": 13839
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6893935756469967e-05,
+      "loss": 0.0012,
+      "step": 13840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6891535026648102e-05,
+      "loss": 0.0011,
+      "step": 13841
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6889134296826237e-05,
+      "loss": 0.0008,
+      "step": 13842
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.688673356700437e-05,
+      "loss": 0.0008,
+      "step": 13843
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6884332837182504e-05,
+      "loss": 0.0008,
+      "step": 13844
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6881932107360636e-05,
+      "loss": 0.001,
+      "step": 13845
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6879531377538774e-05,
+      "loss": 0.0018,
+      "step": 13846
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6877130647716906e-05,
+      "loss": 0.0008,
+      "step": 13847
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.687472991789504e-05,
+      "loss": 0.0008,
+      "step": 13848
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6872329188073173e-05,
+      "loss": 0.0007,
+      "step": 13849
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686992845825131e-05,
+      "loss": 0.0008,
+      "step": 13850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6867527728429444e-05,
+      "loss": 0.0009,
+      "step": 13851
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686512699860758e-05,
+      "loss": 0.001,
+      "step": 13852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686272626878571e-05,
+      "loss": 0.0014,
+      "step": 13853
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6860325538963846e-05,
+      "loss": 0.0008,
+      "step": 13854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.685792480914198e-05,
+      "loss": 0.0008,
+      "step": 13855
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6855524079320116e-05,
+      "loss": 0.0009,
+      "step": 13856
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6853123349498248e-05,
+      "loss": 0.0018,
+      "step": 13857
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6850722619676383e-05,
+      "loss": 0.0007,
+      "step": 13858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6848321889854515e-05,
+      "loss": 0.3613,
+      "step": 13859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.684592116003265e-05,
+      "loss": 0.0011,
+      "step": 13860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6843520430210785e-05,
+      "loss": 0.0007,
+      "step": 13861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6841119700388917e-05,
+      "loss": 0.0008,
+      "step": 13862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6838718970567052e-05,
+      "loss": 0.6412,
+      "step": 13863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6836318240745188e-05,
+      "loss": 0.0008,
+      "step": 13864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6833917510923323e-05,
+      "loss": 0.0007,
+      "step": 13865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6831516781101455e-05,
+      "loss": 0.0011,
+      "step": 13866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682911605127959e-05,
+      "loss": 0.0011,
+      "step": 13867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682671532145772e-05,
+      "loss": 0.0007,
+      "step": 13868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682431459163586e-05,
+      "loss": 0.0012,
+      "step": 13869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6821913861813992e-05,
+      "loss": 0.0013,
+      "step": 13870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6819513131992127e-05,
+      "loss": 0.0009,
+      "step": 13871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.681711240217026e-05,
+      "loss": 0.0008,
+      "step": 13872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6814711672348397e-05,
+      "loss": 0.0009,
+      "step": 13873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.681231094252653e-05,
+      "loss": 0.0011,
+      "step": 13874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6809910212704664e-05,
+      "loss": 0.0009,
+      "step": 13875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6807509482882796e-05,
+      "loss": 0.0011,
+      "step": 13876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.680510875306093e-05,
+      "loss": 0.0017,
+      "step": 13877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6802708023239067e-05,
+      "loss": 0.0008,
+      "step": 13878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.68003072934172e-05,
+      "loss": 0.0009,
+      "step": 13879
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6797906563595334e-05,
+      "loss": 0.0006,
+      "step": 13880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.679550583377347e-05,
+      "loss": 0.0009,
+      "step": 13881
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.67931051039516e-05,
+      "loss": 0.0007,
+      "step": 13882
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6790704374129736e-05,
+      "loss": 0.001,
+      "step": 13883
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.678830364430787e-05,
+      "loss": 0.0007,
+      "step": 13884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6785902914486003e-05,
+      "loss": 0.0009,
+      "step": 13885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6783502184664138e-05,
+      "loss": 0.0084,
+      "step": 13886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6781101454842273e-05,
+      "loss": 0.0194,
+      "step": 13887
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.677870072502041e-05,
+      "loss": 0.0006,
+      "step": 13888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.677629999519854e-05,
+      "loss": 0.001,
+      "step": 13889
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6773899265376675e-05,
+      "loss": 0.0007,
+      "step": 13890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6771498535554807e-05,
+      "loss": 0.0006,
+      "step": 13891
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6769097805732946e-05,
+      "loss": 0.0007,
+      "step": 13892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6766697075911078e-05,
+      "loss": 0.0007,
+      "step": 13893
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6764296346089213e-05,
+      "loss": 0.0007,
+      "step": 13894
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6761895616267345e-05,
+      "loss": 0.0008,
+      "step": 13895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6759494886445483e-05,
+      "loss": 0.0007,
+      "step": 13896
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6757094156623615e-05,
+      "loss": 0.001,
+      "step": 13897
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.675469342680175e-05,
+      "loss": 0.0127,
+      "step": 13898
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6752292696979882e-05,
+      "loss": 0.0007,
+      "step": 13899
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6749891967158017e-05,
+      "loss": 0.0007,
+      "step": 13900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6747491237336152e-05,
+      "loss": 0.0008,
+      "step": 13901
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6745090507514284e-05,
+      "loss": 0.0007,
+      "step": 13902
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.674268977769242e-05,
+      "loss": 0.0012,
+      "step": 13903
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.674028904787055e-05,
+      "loss": 0.0006,
+      "step": 13904
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.673788831804869e-05,
+      "loss": 0.0267,
+      "step": 13905
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.673548758822682e-05,
+      "loss": 0.0006,
+      "step": 13906
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6733086858404957e-05,
+      "loss": 0.0008,
+      "step": 13907
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.673068612858309e-05,
+      "loss": 0.001,
+      "step": 13908
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6728285398761224e-05,
+      "loss": 0.0009,
+      "step": 13909
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.672588466893936e-05,
+      "loss": 0.0006,
+      "step": 13910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6723483939117494e-05,
+      "loss": 0.001,
+      "step": 13911
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6721083209295626e-05,
+      "loss": 0.0007,
+      "step": 13912
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.671868247947376e-05,
+      "loss": 0.7037,
+      "step": 13913
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6716281749651893e-05,
+      "loss": 0.0006,
+      "step": 13914
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.671388101983003e-05,
+      "loss": 0.0006,
+      "step": 13915
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6711480290008163e-05,
+      "loss": 0.0009,
+      "step": 13916
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.67090795601863e-05,
+      "loss": 0.001,
+      "step": 13917
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.670667883036443e-05,
+      "loss": 0.0005,
+      "step": 13918
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9905870518071423,
+      "eval_loss": 0.057951487600803375,
+      "eval_runtime": 265.1418,
+      "eval_samples_per_second": 52.489,
+      "eval_steps_per_second": 1.641,
+      "step": 13918
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6704278100542565e-05,
+      "loss": 0.0006,
+      "step": 13919
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.67018773707207e-05,
+      "loss": 0.0006,
+      "step": 13920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6699476640898836e-05,
+      "loss": 0.0015,
+      "step": 13921
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6697075911076968e-05,
+      "loss": 0.9489,
+      "step": 13922
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6694675181255103e-05,
+      "loss": 0.0007,
+      "step": 13923
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6692274451433238e-05,
+      "loss": 0.0006,
+      "step": 13924
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.668987372161137e-05,
+      "loss": 0.0012,
+      "step": 13925
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6687472991789505e-05,
+      "loss": 0.4077,
+      "step": 13926
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6685072261967637e-05,
+      "loss": 0.0008,
+      "step": 13927
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6682671532145775e-05,
+      "loss": 0.0007,
+      "step": 13928
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6680270802323907e-05,
+      "loss": 0.0006,
+      "step": 13929
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6677870072502042e-05,
+      "loss": 0.001,
+      "step": 13930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6675469342680174e-05,
+      "loss": 0.001,
+      "step": 13931
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.667306861285831e-05,
+      "loss": 0.0008,
+      "step": 13932
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6670667883036444e-05,
+      "loss": 0.0007,
+      "step": 13933
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.666826715321458e-05,
+      "loss": 0.0007,
+      "step": 13934
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.666586642339271e-05,
+      "loss": 0.0007,
+      "step": 13935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6663465693570847e-05,
+      "loss": 0.0006,
+      "step": 13936
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6661064963748982e-05,
+      "loss": 0.0008,
+      "step": 13937
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6658664233927117e-05,
+      "loss": 0.0013,
+      "step": 13938
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.665626350410525e-05,
+      "loss": 0.0007,
+      "step": 13939
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6653862774283384e-05,
+      "loss": 0.0006,
+      "step": 13940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6651462044461516e-05,
+      "loss": 0.0009,
+      "step": 13941
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.664906131463965e-05,
+      "loss": 0.0007,
+      "step": 13942
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6646660584817786e-05,
+      "loss": 0.0009,
+      "step": 13943
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6644259854995918e-05,
+      "loss": 0.0006,
+      "step": 13944
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6641859125174053e-05,
+      "loss": 0.0006,
+      "step": 13945
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6639458395352185e-05,
+      "loss": 0.0006,
+      "step": 13946
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6637057665530323e-05,
+      "loss": 0.0006,
+      "step": 13947
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6634656935708455e-05,
+      "loss": 0.0009,
+      "step": 13948
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.663225620588659e-05,
+      "loss": 0.0009,
+      "step": 13949
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6629855476064722e-05,
+      "loss": 0.0009,
+      "step": 13950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.662745474624286e-05,
+      "loss": 0.0008,
+      "step": 13951
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6625054016420993e-05,
+      "loss": 0.0008,
+      "step": 13952
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6622653286599128e-05,
+      "loss": 0.0006,
+      "step": 13953
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.662025255677726e-05,
+      "loss": 0.0005,
+      "step": 13954
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6617851826955395e-05,
+      "loss": 0.0006,
+      "step": 13955
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.661545109713353e-05,
+      "loss": 0.0006,
+      "step": 13956
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6613050367311665e-05,
+      "loss": 0.001,
+      "step": 13957
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6610649637489797e-05,
+      "loss": 0.0007,
+      "step": 13958
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6608248907667932e-05,
+      "loss": 0.0009,
+      "step": 13959
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6605848177846067e-05,
+      "loss": 0.0007,
+      "step": 13960
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.66034474480242e-05,
+      "loss": 0.0007,
+      "step": 13961
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6601046718202334e-05,
+      "loss": 0.0006,
+      "step": 13962
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.659864598838047e-05,
+      "loss": 0.0007,
+      "step": 13963
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.65962452585586e-05,
+      "loss": 0.0078,
+      "step": 13964
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6593844528736737e-05,
+      "loss": 0.0008,
+      "step": 13965
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6591443798914872e-05,
+      "loss": 0.0007,
+      "step": 13966
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6589043069093004e-05,
+      "loss": 0.0006,
+      "step": 13967
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.658664233927114e-05,
+      "loss": 0.0007,
+      "step": 13968
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6584241609449274e-05,
+      "loss": 0.0007,
+      "step": 13969
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.658184087962741e-05,
+      "loss": 0.0006,
+      "step": 13970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.657944014980554e-05,
+      "loss": 0.0008,
+      "step": 13971
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6577039419983676e-05,
+      "loss": 0.3823,
+      "step": 13972
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6574638690161808e-05,
+      "loss": 0.0015,
+      "step": 13973
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6572237960339946e-05,
+      "loss": 0.0006,
+      "step": 13974
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6569837230518078e-05,
+      "loss": 0.0005,
+      "step": 13975
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6567436500696213e-05,
+      "loss": 0.0007,
+      "step": 13976
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6565035770874345e-05,
+      "loss": 0.0007,
+      "step": 13977
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.656263504105248e-05,
+      "loss": 0.0006,
+      "step": 13978
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6560234311230616e-05,
+      "loss": 0.0006,
+      "step": 13979
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.655783358140875e-05,
+      "loss": 0.0008,
+      "step": 13980
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6555432851586883e-05,
+      "loss": 0.0006,
+      "step": 13981
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6553032121765018e-05,
+      "loss": 0.0006,
+      "step": 13982
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6550631391943153e-05,
+      "loss": 0.0008,
+      "step": 13983
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6548230662121285e-05,
+      "loss": 0.0009,
+      "step": 13984
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.654582993229942e-05,
+      "loss": 0.0006,
+      "step": 13985
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6543429202477552e-05,
+      "loss": 0.0005,
+      "step": 13986
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6541028472655687e-05,
+      "loss": 0.0005,
+      "step": 13987
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6538627742833822e-05,
+      "loss": 0.0006,
+      "step": 13988
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6536227013011957e-05,
+      "loss": 0.0006,
+      "step": 13989
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.653382628319009e-05,
+      "loss": 0.0013,
+      "step": 13990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6531425553368224e-05,
+      "loss": 0.0005,
+      "step": 13991
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.652902482354636e-05,
+      "loss": 0.0006,
+      "step": 13992
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6526624093724495e-05,
+      "loss": 0.0004,
+      "step": 13993
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6524223363902627e-05,
+      "loss": 0.0006,
+      "step": 13994
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6521822634080762e-05,
+      "loss": 0.0006,
+      "step": 13995
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6519421904258894e-05,
+      "loss": 0.0008,
+      "step": 13996
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6517021174437032e-05,
+      "loss": 0.0011,
+      "step": 13997
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6514620444615164e-05,
+      "loss": 0.0007,
+      "step": 13998
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.65122197147933e-05,
+      "loss": 0.0007,
+      "step": 13999
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.650981898497143e-05,
+      "loss": 0.0005,
+      "step": 14000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6507418255149566e-05,
+      "loss": 0.0007,
+      "step": 14001
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.65050175253277e-05,
+      "loss": 0.0006,
+      "step": 14002
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6502616795505833e-05,
+      "loss": 0.0007,
+      "step": 14003
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6500216065683968e-05,
+      "loss": 0.0008,
+      "step": 14004
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6497815335862103e-05,
+      "loss": 0.0006,
+      "step": 14005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.649541460604024e-05,
+      "loss": 0.0105,
+      "step": 14006
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.649301387621837e-05,
+      "loss": 0.0007,
+      "step": 14007
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6490613146396506e-05,
+      "loss": 0.0005,
+      "step": 14008
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6488212416574637e-05,
+      "loss": 0.0007,
+      "step": 14009
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6485811686752773e-05,
+      "loss": 0.0008,
+      "step": 14010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6483410956930908e-05,
+      "loss": 0.0005,
+      "step": 14011
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6481010227109043e-05,
+      "loss": 0.0005,
+      "step": 14012
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6478609497287175e-05,
+      "loss": 0.0005,
+      "step": 14013
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.647620876746531e-05,
+      "loss": 0.0006,
+      "step": 14014
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6473808037643445e-05,
+      "loss": 0.0008,
+      "step": 14015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.647140730782158e-05,
+      "loss": 0.0009,
+      "step": 14016
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6469006577999712e-05,
+      "loss": 0.0009,
+      "step": 14017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6466605848177847e-05,
+      "loss": 0.0005,
+      "step": 14018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.646420511835598e-05,
+      "loss": 0.0008,
+      "step": 14019
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6461804388534118e-05,
+      "loss": 0.0006,
+      "step": 14020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.645940365871225e-05,
+      "loss": 0.0006,
+      "step": 14021
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6457002928890385e-05,
+      "loss": 0.0004,
+      "step": 14022
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6454602199068517e-05,
+      "loss": 0.0005,
+      "step": 14023
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.645220146924665e-05,
+      "loss": 0.0005,
+      "step": 14024
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6449800739424787e-05,
+      "loss": 0.0006,
+      "step": 14025
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.644740000960292e-05,
+      "loss": 0.0008,
+      "step": 14026
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6444999279781054e-05,
+      "loss": 0.0008,
+      "step": 14027
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6442598549959186e-05,
+      "loss": 0.0004,
+      "step": 14028
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6440197820137324e-05,
+      "loss": 0.0006,
+      "step": 14029
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6437797090315456e-05,
+      "loss": 0.0008,
+      "step": 14030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.643539636049359e-05,
+      "loss": 0.0005,
+      "step": 14031
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6432995630671723e-05,
+      "loss": 0.0006,
+      "step": 14032
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.643059490084986e-05,
+      "loss": 0.0004,
+      "step": 14033
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6428194171027993e-05,
+      "loss": 0.0004,
+      "step": 14034
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.642579344120613e-05,
+      "loss": 0.0005,
+      "step": 14035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.642339271138426e-05,
+      "loss": 0.0007,
+      "step": 14036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6420991981562396e-05,
+      "loss": 0.0009,
+      "step": 14037
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.641859125174053e-05,
+      "loss": 0.0006,
+      "step": 14038
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6416190521918666e-05,
+      "loss": 0.0005,
+      "step": 14039
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6413789792096798e-05,
+      "loss": 0.0005,
+      "step": 14040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6411389062274933e-05,
+      "loss": 0.0006,
+      "step": 14041
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6408988332453065e-05,
+      "loss": 0.0005,
+      "step": 14042
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.64065876026312e-05,
+      "loss": 0.0005,
+      "step": 14043
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6404186872809335e-05,
+      "loss": 0.0006,
+      "step": 14044
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.640178614298747e-05,
+      "loss": 0.0005,
+      "step": 14045
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6399385413165602e-05,
+      "loss": 0.0005,
+      "step": 14046
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6396984683343737e-05,
+      "loss": 0.0005,
+      "step": 14047
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6394583953521872e-05,
+      "loss": 0.0004,
+      "step": 14048
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6392183223700004e-05,
+      "loss": 0.0005,
+      "step": 14049
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638978249387814e-05,
+      "loss": 0.6722,
+      "step": 14050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638738176405627e-05,
+      "loss": 0.0005,
+      "step": 14051
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638498103423441e-05,
+      "loss": 0.0006,
+      "step": 14052
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.638258030441254e-05,
+      "loss": 0.0005,
+      "step": 14053
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6380179574590677e-05,
+      "loss": 0.0006,
+      "step": 14054
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.637777884476881e-05,
+      "loss": 0.0007,
+      "step": 14055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6375378114946947e-05,
+      "loss": 0.0004,
+      "step": 14056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.637297738512508e-05,
+      "loss": 0.0006,
+      "step": 14057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6370576655303214e-05,
+      "loss": 0.0007,
+      "step": 14058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6368175925481346e-05,
+      "loss": 0.0005,
+      "step": 14059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.636577519565948e-05,
+      "loss": 0.0007,
+      "step": 14060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6363374465837616e-05,
+      "loss": 0.0007,
+      "step": 14061
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.636097373601575e-05,
+      "loss": 0.0005,
+      "step": 14062
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6358573006193883e-05,
+      "loss": 0.0003,
+      "step": 14063
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.635617227637202e-05,
+      "loss": 0.0011,
+      "step": 14064
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.635377154655015e-05,
+      "loss": 0.0004,
+      "step": 14065
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6351370816728286e-05,
+      "loss": 0.0007,
+      "step": 14066
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.634897008690642e-05,
+      "loss": 0.0006,
+      "step": 14067
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6346569357084553e-05,
+      "loss": 0.0005,
+      "step": 14068
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6344168627262688e-05,
+      "loss": 0.0005,
+      "step": 14069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6341767897440823e-05,
+      "loss": 0.0004,
+      "step": 14070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6339367167618958e-05,
+      "loss": 0.0004,
+      "step": 14071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.633696643779709e-05,
+      "loss": 0.0006,
+      "step": 14072
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6334565707975225e-05,
+      "loss": 0.0007,
+      "step": 14073
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6332164978153357e-05,
+      "loss": 0.0005,
+      "step": 14074
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6329764248331495e-05,
+      "loss": 0.0006,
+      "step": 14075
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6327363518509627e-05,
+      "loss": 0.0073,
+      "step": 14076
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6324962788687762e-05,
+      "loss": 0.0004,
+      "step": 14077
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6322562058865894e-05,
+      "loss": 0.0004,
+      "step": 14078
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6320161329044033e-05,
+      "loss": 0.0005,
+      "step": 14079
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6317760599222165e-05,
+      "loss": 0.0004,
+      "step": 14080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.63153598694003e-05,
+      "loss": 0.0005,
+      "step": 14081
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.631295913957843e-05,
+      "loss": 0.0004,
+      "step": 14082
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6310558409756567e-05,
+      "loss": 0.4054,
+      "step": 14083
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6308157679934702e-05,
+      "loss": 0.0004,
+      "step": 14084
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6305756950112834e-05,
+      "loss": 0.0005,
+      "step": 14085
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.630335622029097e-05,
+      "loss": 0.0004,
+      "step": 14086
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6300955490469104e-05,
+      "loss": 0.0005,
+      "step": 14087
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.629855476064724e-05,
+      "loss": 0.0004,
+      "step": 14088
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.629615403082537e-05,
+      "loss": 0.9582,
+      "step": 14089
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6293753301003506e-05,
+      "loss": 0.0005,
+      "step": 14090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6291352571181638e-05,
+      "loss": 0.786,
+      "step": 14091
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6288951841359773e-05,
+      "loss": 0.0005,
+      "step": 14092
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.628655111153791e-05,
+      "loss": 0.0004,
+      "step": 14093
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6284150381716044e-05,
+      "loss": 0.0005,
+      "step": 14094
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6281749651894176e-05,
+      "loss": 0.0005,
+      "step": 14095
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.627934892207231e-05,
+      "loss": 0.0005,
+      "step": 14096
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6276948192250443e-05,
+      "loss": 0.0004,
+      "step": 14097
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.627454746242858e-05,
+      "loss": 0.0005,
+      "step": 14098
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6272146732606713e-05,
+      "loss": 0.0005,
+      "step": 14099
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6269746002784848e-05,
+      "loss": 0.0004,
+      "step": 14100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.626734527296298e-05,
+      "loss": 0.0004,
+      "step": 14101
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.626494454314112e-05,
+      "loss": 0.0005,
+      "step": 14102
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.626254381331925e-05,
+      "loss": 0.0005,
+      "step": 14103
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6260143083497385e-05,
+      "loss": 0.0005,
+      "step": 14104
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6257742353675517e-05,
+      "loss": 0.0005,
+      "step": 14105
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6255341623853652e-05,
+      "loss": 0.0006,
+      "step": 14106
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6252940894031788e-05,
+      "loss": 0.0007,
+      "step": 14107
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.625054016420992e-05,
+      "loss": 0.0005,
+      "step": 14108
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6248139434388055e-05,
+      "loss": 0.0005,
+      "step": 14109
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6245738704566186e-05,
+      "loss": 0.0006,
+      "step": 14110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6243337974744325e-05,
+      "loss": 0.0005,
+      "step": 14111
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6240937244922457e-05,
+      "loss": 0.0004,
+      "step": 14112
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6238536515100592e-05,
+      "loss": 0.0005,
+      "step": 14113
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6236135785278724e-05,
+      "loss": 0.0005,
+      "step": 14114
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.623373505545686e-05,
+      "loss": 0.001,
+      "step": 14115
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6231334325634994e-05,
+      "loss": 0.0007,
+      "step": 14116
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.622893359581313e-05,
+      "loss": 0.0007,
+      "step": 14117
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.622653286599126e-05,
+      "loss": 0.0004,
+      "step": 14118
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6224132136169396e-05,
+      "loss": 0.3608,
+      "step": 14119
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.622173140634753e-05,
+      "loss": 0.0008,
+      "step": 14120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6219330676525667e-05,
+      "loss": 0.0008,
+      "step": 14121
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.62169299467038e-05,
+      "loss": 0.0005,
+      "step": 14122
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6214529216881934e-05,
+      "loss": 0.0006,
+      "step": 14123
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6212128487060066e-05,
+      "loss": 0.0006,
+      "step": 14124
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.62097277572382e-05,
+      "loss": 0.0064,
+      "step": 14125
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6207327027416336e-05,
+      "loss": 0.0006,
+      "step": 14126
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6204926297594468e-05,
+      "loss": 0.0004,
+      "step": 14127
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6202525567772603e-05,
+      "loss": 0.0004,
+      "step": 14128
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6200124837950738e-05,
+      "loss": 0.0005,
+      "step": 14129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6197724108128873e-05,
+      "loss": 0.0007,
+      "step": 14130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6195323378307005e-05,
+      "loss": 0.0005,
+      "step": 14131
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.619292264848514e-05,
+      "loss": 0.0004,
+      "step": 14132
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6190521918663272e-05,
+      "loss": 0.0007,
+      "step": 14133
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.618812118884141e-05,
+      "loss": 0.3446,
+      "step": 14134
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6185720459019542e-05,
+      "loss": 0.0005,
+      "step": 14135
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6183319729197678e-05,
+      "loss": 0.0005,
+      "step": 14136
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.618091899937581e-05,
+      "loss": 0.0004,
+      "step": 14137
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6178518269553945e-05,
+      "loss": 0.0005,
+      "step": 14138
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.617611753973208e-05,
+      "loss": 0.0011,
+      "step": 14139
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6173716809910215e-05,
+      "loss": 0.0005,
+      "step": 14140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6171316080088347e-05,
+      "loss": 0.0005,
+      "step": 14141
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6168915350266482e-05,
+      "loss": 0.0009,
+      "step": 14142
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6166514620444617e-05,
+      "loss": 0.0005,
+      "step": 14143
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6164113890622752e-05,
+      "loss": 0.0006,
+      "step": 14144
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6161713160800884e-05,
+      "loss": 0.001,
+      "step": 14145
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.615931243097902e-05,
+      "loss": 0.0004,
+      "step": 14146
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.615691170115715e-05,
+      "loss": 0.001,
+      "step": 14147
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6154510971335286e-05,
+      "loss": 0.0008,
+      "step": 14148
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.615211024151342e-05,
+      "loss": 0.0004,
+      "step": 14149
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6149709511691553e-05,
+      "loss": 0.0004,
+      "step": 14150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.614730878186969e-05,
+      "loss": 0.0008,
+      "step": 14151
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6144908052047824e-05,
+      "loss": 0.0004,
+      "step": 14152
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.614250732222596e-05,
+      "loss": 0.0004,
+      "step": 14153
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.614010659240409e-05,
+      "loss": 0.0004,
+      "step": 14154
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6137705862582226e-05,
+      "loss": 0.0005,
+      "step": 14155
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6135305132760358e-05,
+      "loss": 0.0004,
+      "step": 14156
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6132904402938496e-05,
+      "loss": 0.0005,
+      "step": 14157
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6130503673116628e-05,
+      "loss": 0.0004,
+      "step": 14158
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6128102943294763e-05,
+      "loss": 0.0006,
+      "step": 14159
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6125702213472895e-05,
+      "loss": 0.001,
+      "step": 14160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.612330148365103e-05,
+      "loss": 0.0007,
+      "step": 14161
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6120900753829165e-05,
+      "loss": 0.0004,
+      "step": 14162
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.61185000240073e-05,
+      "loss": 0.0005,
+      "step": 14163
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6116099294185432e-05,
+      "loss": 0.0006,
+      "step": 14164
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6113698564363568e-05,
+      "loss": 0.0006,
+      "step": 14165
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6111297834541703e-05,
+      "loss": 0.0005,
+      "step": 14166
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6108897104719835e-05,
+      "loss": 0.0005,
+      "step": 14167
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.610649637489797e-05,
+      "loss": 0.0006,
+      "step": 14168
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.61040956450761e-05,
+      "loss": 0.0004,
+      "step": 14169
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6101694915254237e-05,
+      "loss": 0.0005,
+      "step": 14170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6099294185432372e-05,
+      "loss": 0.0004,
+      "step": 14171
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6096893455610507e-05,
+      "loss": 0.0004,
+      "step": 14172
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.609449272578864e-05,
+      "loss": 0.0005,
+      "step": 14173
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6092091995966774e-05,
+      "loss": 0.0008,
+      "step": 14174
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608969126614491e-05,
+      "loss": 0.0005,
+      "step": 14175
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6087290536323044e-05,
+      "loss": 0.0008,
+      "step": 14176
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6084889806501176e-05,
+      "loss": 0.0007,
+      "step": 14177
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.608248907667931e-05,
+      "loss": 0.0007,
+      "step": 14178
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6080088346857443e-05,
+      "loss": 0.0008,
+      "step": 14179
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6077687617035582e-05,
+      "loss": 0.0004,
+      "step": 14180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6075286887213714e-05,
+      "loss": 0.0004,
+      "step": 14181
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.607288615739185e-05,
+      "loss": 0.0004,
+      "step": 14182
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.607048542756998e-05,
+      "loss": 0.0065,
+      "step": 14183
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.606808469774812e-05,
+      "loss": 0.4364,
+      "step": 14184
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.606568396792625e-05,
+      "loss": 0.0005,
+      "step": 14185
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6063283238104386e-05,
+      "loss": 0.0009,
+      "step": 14186
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6060882508282518e-05,
+      "loss": 0.0006,
+      "step": 14187
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6058481778460653e-05,
+      "loss": 0.0004,
+      "step": 14188
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.605608104863879e-05,
+      "loss": 0.0005,
+      "step": 14189
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.605368031881692e-05,
+      "loss": 0.0004,
+      "step": 14190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6051279588995055e-05,
+      "loss": 0.0005,
+      "step": 14191
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6048878859173187e-05,
+      "loss": 0.0004,
+      "step": 14192
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6046478129351322e-05,
+      "loss": 0.0007,
+      "step": 14193
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6044077399529458e-05,
+      "loss": 0.0004,
+      "step": 14194
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6041676669707593e-05,
+      "loss": 0.0004,
+      "step": 14195
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6039275939885725e-05,
+      "loss": 0.0005,
+      "step": 14196
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.603687521006386e-05,
+      "loss": 0.0005,
+      "step": 14197
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6034474480241995e-05,
+      "loss": 0.0005,
+      "step": 14198
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.603207375042013e-05,
+      "loss": 0.0006,
+      "step": 14199
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6029673020598262e-05,
+      "loss": 0.0004,
+      "step": 14200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6027272290776397e-05,
+      "loss": 0.0005,
+      "step": 14201
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.602487156095453e-05,
+      "loss": 0.0007,
+      "step": 14202
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6022470831132667e-05,
+      "loss": 0.0004,
+      "step": 14203
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.60200701013108e-05,
+      "loss": 0.0004,
+      "step": 14204
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6017669371488934e-05,
+      "loss": 0.0004,
+      "step": 14205
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6015268641667066e-05,
+      "loss": 0.0005,
+      "step": 14206
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.60128679118452e-05,
+      "loss": 0.0004,
+      "step": 14207
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6010467182023337e-05,
+      "loss": 0.0005,
+      "step": 14208
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.600806645220147e-05,
+      "loss": 0.0007,
+      "step": 14209
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6005665722379604e-05,
+      "loss": 0.0061,
+      "step": 14210
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.600326499255774e-05,
+      "loss": 0.0005,
+      "step": 14211
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6000864262735874e-05,
+      "loss": 0.0003,
+      "step": 14212
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5998463532914006e-05,
+      "loss": 0.0009,
+      "step": 14213
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599606280309214e-05,
+      "loss": 0.0007,
+      "step": 14214
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5993662073270273e-05,
+      "loss": 0.001,
+      "step": 14215
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.599126134344841e-05,
+      "loss": 0.0005,
+      "step": 14216
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5988860613626543e-05,
+      "loss": 0.0006,
+      "step": 14217
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598645988380468e-05,
+      "loss": 0.0007,
+      "step": 14218
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.598405915398281e-05,
+      "loss": 0.0004,
+      "step": 14219
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5981658424160945e-05,
+      "loss": 0.0005,
+      "step": 14220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.597925769433908e-05,
+      "loss": 0.7417,
+      "step": 14221
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5976856964517216e-05,
+      "loss": 0.2494,
+      "step": 14222
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5974456234695348e-05,
+      "loss": 0.0006,
+      "step": 14223
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5972055504873483e-05,
+      "loss": 0.0011,
+      "step": 14224
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5969654775051615e-05,
+      "loss": 0.0003,
+      "step": 14225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5967254045229753e-05,
+      "loss": 0.0006,
+      "step": 14226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5964853315407885e-05,
+      "loss": 0.0007,
+      "step": 14227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.596245258558602e-05,
+      "loss": 0.0004,
+      "step": 14228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5960051855764152e-05,
+      "loss": 0.0008,
+      "step": 14229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5957651125942287e-05,
+      "loss": 0.0004,
+      "step": 14230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5955250396120422e-05,
+      "loss": 0.0007,
+      "step": 14231
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5952849666298554e-05,
+      "loss": 0.0004,
+      "step": 14232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.595044893647669e-05,
+      "loss": 0.0008,
+      "step": 14233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.594804820665482e-05,
+      "loss": 0.0004,
+      "step": 14234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.594564747683296e-05,
+      "loss": 0.0003,
+      "step": 14235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.594324674701109e-05,
+      "loss": 0.0004,
+      "step": 14236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5940846017189227e-05,
+      "loss": 0.0004,
+      "step": 14237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.593844528736736e-05,
+      "loss": 0.0004,
+      "step": 14238
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5936044557545497e-05,
+      "loss": 0.0004,
+      "step": 14239
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.593364382772363e-05,
+      "loss": 0.0004,
+      "step": 14240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5931243097901764e-05,
+      "loss": 0.0008,
+      "step": 14241
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5928842368079896e-05,
+      "loss": 0.0004,
+      "step": 14242
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.592644163825803e-05,
+      "loss": 0.0004,
+      "step": 14243
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5924040908436166e-05,
+      "loss": 0.0004,
+      "step": 14244
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.59216401786143e-05,
+      "loss": 0.0005,
+      "step": 14245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5919239448792433e-05,
+      "loss": 0.0004,
+      "step": 14246
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.591683871897057e-05,
+      "loss": 0.0003,
+      "step": 14247
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5914437989148704e-05,
+      "loss": 0.0004,
+      "step": 14248
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5912037259326835e-05,
+      "loss": 0.0294,
+      "step": 14249
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.590963652950497e-05,
+      "loss": 0.0003,
+      "step": 14250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5907235799683102e-05,
+      "loss": 0.0008,
+      "step": 14251
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5904835069861237e-05,
+      "loss": 0.0004,
+      "step": 14252
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5902434340039373e-05,
+      "loss": 0.0004,
+      "step": 14253
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5900033610217508e-05,
+      "loss": 0.0006,
+      "step": 14254
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.589763288039564e-05,
+      "loss": 0.0004,
+      "step": 14255
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5895232150573775e-05,
+      "loss": 0.0102,
+      "step": 14256
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5892831420751907e-05,
+      "loss": 0.0004,
+      "step": 14257
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5890430690930045e-05,
+      "loss": 0.0003,
+      "step": 14258
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5888029961108177e-05,
+      "loss": 0.0005,
+      "step": 14259
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5885629231286312e-05,
+      "loss": 0.0005,
+      "step": 14260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5883228501464444e-05,
+      "loss": 0.0004,
+      "step": 14261
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5880827771642583e-05,
+      "loss": 0.0003,
+      "step": 14262
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5878427041820714e-05,
+      "loss": 0.0003,
+      "step": 14263
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.587602631199885e-05,
+      "loss": 0.0008,
+      "step": 14264
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.587362558217698e-05,
+      "loss": 0.0004,
+      "step": 14265
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5871224852355117e-05,
+      "loss": 0.0004,
+      "step": 14266
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5868824122533252e-05,
+      "loss": 0.0003,
+      "step": 14267
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5866423392711387e-05,
+      "loss": 0.0003,
+      "step": 14268
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.586402266288952e-05,
+      "loss": 0.0004,
+      "step": 14269
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5861621933067654e-05,
+      "loss": 0.0012,
+      "step": 14270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.585922120324579e-05,
+      "loss": 0.0004,
+      "step": 14271
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.585682047342392e-05,
+      "loss": 0.0004,
+      "step": 14272
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5854419743602056e-05,
+      "loss": 0.0004,
+      "step": 14273
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5852019013780188e-05,
+      "loss": 0.0003,
+      "step": 14274
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5849618283958323e-05,
+      "loss": 0.0003,
+      "step": 14275
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5847217554136458e-05,
+      "loss": 0.0005,
+      "step": 14276
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5844816824314593e-05,
+      "loss": 0.0004,
+      "step": 14277
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5842416094492725e-05,
+      "loss": 0.9933,
+      "step": 14278
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.584001536467086e-05,
+      "loss": 0.0004,
+      "step": 14279
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5837614634848996e-05,
+      "loss": 0.0006,
+      "step": 14280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.583521390502713e-05,
+      "loss": 0.0007,
+      "step": 14281
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5832813175205263e-05,
+      "loss": 0.8884,
+      "step": 14282
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5830412445383398e-05,
+      "loss": 0.0006,
+      "step": 14283
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.582801171556153e-05,
+      "loss": 0.0004,
+      "step": 14284
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5825610985739668e-05,
+      "loss": 0.0004,
+      "step": 14285
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.58232102559178e-05,
+      "loss": 0.9742,
+      "step": 14286
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5820809526095935e-05,
+      "loss": 0.0004,
+      "step": 14287
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5818408796274067e-05,
+      "loss": 0.0007,
+      "step": 14288
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5816008066452202e-05,
+      "loss": 0.0004,
+      "step": 14289
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5813607336630337e-05,
+      "loss": 0.0004,
+      "step": 14290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.581120660680847e-05,
+      "loss": 0.0004,
+      "step": 14291
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5808805876986604e-05,
+      "loss": 0.0004,
+      "step": 14292
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5806405147164736e-05,
+      "loss": 0.0005,
+      "step": 14293
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5804004417342875e-05,
+      "loss": 0.0004,
+      "step": 14294
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5801603687521007e-05,
+      "loss": 0.0004,
+      "step": 14295
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5799202957699142e-05,
+      "loss": 0.0004,
+      "step": 14296
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5796802227877274e-05,
+      "loss": 0.0078,
+      "step": 14297
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.579440149805541e-05,
+      "loss": 0.0004,
+      "step": 14298
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5792000768233544e-05,
+      "loss": 0.0004,
+      "step": 14299
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.578960003841168e-05,
+      "loss": 0.0004,
+      "step": 14300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.578719930858981e-05,
+      "loss": 0.8295,
+      "step": 14301
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5784798578767946e-05,
+      "loss": 0.0004,
+      "step": 14302
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.578239784894608e-05,
+      "loss": 0.0005,
+      "step": 14303
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5779997119124216e-05,
+      "loss": 0.0004,
+      "step": 14304
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5777596389302348e-05,
+      "loss": 0.0004,
+      "step": 14305
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5775195659480483e-05,
+      "loss": 0.0004,
+      "step": 14306
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5772794929658615e-05,
+      "loss": 0.0007,
+      "step": 14307
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5770394199836754e-05,
+      "loss": 0.0004,
+      "step": 14308
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5767993470014886e-05,
+      "loss": 0.001,
+      "step": 14309
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.576559274019302e-05,
+      "loss": 0.0007,
+      "step": 14310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5763192010371153e-05,
+      "loss": 0.0005,
+      "step": 14311
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5760791280549288e-05,
+      "loss": 0.0004,
+      "step": 14312
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5758390550727423e-05,
+      "loss": 0.0004,
+      "step": 14313
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5755989820905555e-05,
+      "loss": 0.0004,
+      "step": 14314
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.575358909108369e-05,
+      "loss": 0.0004,
+      "step": 14315
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5751188361261822e-05,
+      "loss": 0.0004,
+      "step": 14316
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.574878763143996e-05,
+      "loss": 0.0005,
+      "step": 14317
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5746386901618092e-05,
+      "loss": 0.0005,
+      "step": 14318
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5743986171796227e-05,
+      "loss": 0.323,
+      "step": 14319
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.574158544197436e-05,
+      "loss": 0.0004,
+      "step": 14320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5739184712152494e-05,
+      "loss": 0.0012,
+      "step": 14321
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.573678398233063e-05,
+      "loss": 0.0004,
+      "step": 14322
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5734383252508765e-05,
+      "loss": 0.0004,
+      "step": 14323
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5731982522686897e-05,
+      "loss": 0.0004,
+      "step": 14324
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5729581792865032e-05,
+      "loss": 0.001,
+      "step": 14325
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5727181063043167e-05,
+      "loss": 0.0003,
+      "step": 14326
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5724780333221302e-05,
+      "loss": 0.0004,
+      "step": 14327
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5722379603399434e-05,
+      "loss": 0.0005,
+      "step": 14328
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.571997887357757e-05,
+      "loss": 0.0006,
+      "step": 14329
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.57175781437557e-05,
+      "loss": 0.0004,
+      "step": 14330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5715177413933836e-05,
+      "loss": 0.0004,
+      "step": 14331
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.571277668411197e-05,
+      "loss": 0.004,
+      "step": 14332
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5710375954290103e-05,
+      "loss": 0.0004,
+      "step": 14333
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5707975224468238e-05,
+      "loss": 0.0009,
+      "step": 14334
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5705574494646373e-05,
+      "loss": 0.0008,
+      "step": 14335
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.570317376482451e-05,
+      "loss": 0.2496,
+      "step": 14336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.570077303500264e-05,
+      "loss": 0.0003,
+      "step": 14337
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5698372305180776e-05,
+      "loss": 0.0052,
+      "step": 14338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5695971575358907e-05,
+      "loss": 0.0004,
+      "step": 14339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5693570845537046e-05,
+      "loss": 0.0003,
+      "step": 14340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5691170115715178e-05,
+      "loss": 0.0007,
+      "step": 14341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5688769385893313e-05,
+      "loss": 0.0007,
+      "step": 14342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5686368656071445e-05,
+      "loss": 0.0007,
+      "step": 14343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5683967926249583e-05,
+      "loss": 0.0015,
+      "step": 14344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5681567196427715e-05,
+      "loss": 0.0004,
+      "step": 14345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.567916646660585e-05,
+      "loss": 0.0004,
+      "step": 14346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5676765736783982e-05,
+      "loss": 0.0004,
+      "step": 14347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5674365006962117e-05,
+      "loss": 0.0007,
+      "step": 14348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5671964277140253e-05,
+      "loss": 0.0003,
+      "step": 14349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5669563547318388e-05,
+      "loss": 0.0003,
+      "step": 14350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.566716281749652e-05,
+      "loss": 0.0004,
+      "step": 14351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5664762087674655e-05,
+      "loss": 0.0004,
+      "step": 14352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5662361357852786e-05,
+      "loss": 0.0004,
+      "step": 14353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.565996062803092e-05,
+      "loss": 0.0003,
+      "step": 14354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5657559898209057e-05,
+      "loss": 0.0003,
+      "step": 14355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.565515916838719e-05,
+      "loss": 0.0004,
+      "step": 14356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5652758438565324e-05,
+      "loss": 0.0005,
+      "step": 14357
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.565035770874346e-05,
+      "loss": 0.0003,
+      "step": 14358
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5647956978921594e-05,
+      "loss": 0.0004,
+      "step": 14359
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5645556249099726e-05,
+      "loss": 0.0004,
+      "step": 14360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.564315551927786e-05,
+      "loss": 0.0007,
+      "step": 14361
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5640754789455993e-05,
+      "loss": 0.0004,
+      "step": 14362
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.563835405963413e-05,
+      "loss": 0.0004,
+      "step": 14363
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5635953329812263e-05,
+      "loss": 0.0007,
+      "step": 14364
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.56335525999904e-05,
+      "loss": 0.0004,
+      "step": 14365
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.563115187016853e-05,
+      "loss": 0.0011,
+      "step": 14366
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.562875114034667e-05,
+      "loss": 0.0004,
+      "step": 14367
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.56263504105248e-05,
+      "loss": 0.0005,
+      "step": 14368
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5623949680702936e-05,
+      "loss": 0.0003,
+      "step": 14369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5621548950881068e-05,
+      "loss": 0.0004,
+      "step": 14370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5619148221059203e-05,
+      "loss": 0.0006,
+      "step": 14371
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5616747491237338e-05,
+      "loss": 0.0005,
+      "step": 14372
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.561434676141547e-05,
+      "loss": 0.0003,
+      "step": 14373
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5611946031593605e-05,
+      "loss": 0.0007,
+      "step": 14374
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5609545301771737e-05,
+      "loss": 0.0003,
+      "step": 14375
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5607144571949875e-05,
+      "loss": 0.0006,
+      "step": 14376
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5604743842128007e-05,
+      "loss": 0.0004,
+      "step": 14377
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5602343112306142e-05,
+      "loss": 0.0007,
+      "step": 14378
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5599942382484274e-05,
+      "loss": 0.0004,
+      "step": 14379
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.559754165266241e-05,
+      "loss": 0.0004,
+      "step": 14380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5595140922840545e-05,
+      "loss": 0.0006,
+      "step": 14381
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.559274019301868e-05,
+      "loss": 0.0003,
+      "step": 14382
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.559033946319681e-05,
+      "loss": 0.0003,
+      "step": 14383
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5587938733374947e-05,
+      "loss": 0.0006,
+      "step": 14384
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.558553800355308e-05,
+      "loss": 0.4685,
+      "step": 14385
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5583137273731217e-05,
+      "loss": 0.0006,
+      "step": 14386
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.558073654390935e-05,
+      "loss": 0.0004,
+      "step": 14387
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5578335814087484e-05,
+      "loss": 0.0005,
+      "step": 14388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5575935084265616e-05,
+      "loss": 0.0008,
+      "step": 14389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.557353435444375e-05,
+      "loss": 0.0003,
+      "step": 14390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5571133624621886e-05,
+      "loss": 0.0005,
+      "step": 14391
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556873289480002e-05,
+      "loss": 0.0004,
+      "step": 14392
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5566332164978153e-05,
+      "loss": 0.0006,
+      "step": 14393
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556393143515629e-05,
+      "loss": 0.0004,
+      "step": 14394
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5561530705334424e-05,
+      "loss": 0.0004,
+      "step": 14395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5559129975512556e-05,
+      "loss": 0.0011,
+      "step": 14396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.555672924569069e-05,
+      "loss": 0.0004,
+      "step": 14397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5554328515868823e-05,
+      "loss": 0.0003,
+      "step": 14398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.555192778604696e-05,
+      "loss": 0.0003,
+      "step": 14399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5549527056225093e-05,
+      "loss": 0.0003,
+      "step": 14400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5547126326403228e-05,
+      "loss": 0.0004,
+      "step": 14401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.554472559658136e-05,
+      "loss": 0.0003,
+      "step": 14402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5542324866759495e-05,
+      "loss": 0.0004,
+      "step": 14403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.553992413693763e-05,
+      "loss": 0.0003,
+      "step": 14404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5537523407115765e-05,
+      "loss": 0.0004,
+      "step": 14405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5535122677293897e-05,
+      "loss": 0.0006,
+      "step": 14406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5532721947472032e-05,
+      "loss": 0.0007,
+      "step": 14407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5530321217650164e-05,
+      "loss": 0.0004,
+      "step": 14408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5527920487828303e-05,
+      "loss": 0.0004,
+      "step": 14409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5525519758006435e-05,
+      "loss": 0.0011,
+      "step": 14410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.552311902818457e-05,
+      "loss": 0.0007,
+      "step": 14411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.55207182983627e-05,
+      "loss": 0.0006,
+      "step": 14412
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5518317568540837e-05,
+      "loss": 0.0003,
+      "step": 14413
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5515916838718972e-05,
+      "loss": 0.0003,
+      "step": 14414
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5513516108897104e-05,
+      "loss": 0.0007,
+      "step": 14415
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.551111537907524e-05,
+      "loss": 0.0008,
+      "step": 14416
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.550871464925337e-05,
+      "loss": 0.0005,
+      "step": 14417
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.550631391943151e-05,
+      "loss": 0.0004,
+      "step": 14418
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.550391318960964e-05,
+      "loss": 0.0353,
+      "step": 14419
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5501512459787776e-05,
+      "loss": 0.0007,
+      "step": 14420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5499111729965908e-05,
+      "loss": 0.0005,
+      "step": 14421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5496711000144047e-05,
+      "loss": 0.0003,
+      "step": 14422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.549431027032218e-05,
+      "loss": 0.0006,
+      "step": 14423
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5491909540500314e-05,
+      "loss": 0.0008,
+      "step": 14424
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5489508810678446e-05,
+      "loss": 0.0011,
+      "step": 14425
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.548710808085658e-05,
+      "loss": 0.0005,
+      "step": 14426
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5484707351034716e-05,
+      "loss": 0.4215,
+      "step": 14427
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.548230662121285e-05,
+      "loss": 0.0003,
+      "step": 14428
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5479905891390983e-05,
+      "loss": 0.0008,
+      "step": 14429
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5477505161569118e-05,
+      "loss": 0.0006,
+      "step": 14430
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5475104431747253e-05,
+      "loss": 0.0003,
+      "step": 14431
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.547270370192539e-05,
+      "loss": 0.0032,
+      "step": 14432
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.547030297210352e-05,
+      "loss": 0.0005,
+      "step": 14433
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5467902242281655e-05,
+      "loss": 0.0003,
+      "step": 14434
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5465501512459787e-05,
+      "loss": 0.0003,
+      "step": 14435
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5463100782637922e-05,
+      "loss": 0.0004,
+      "step": 14436
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5460700052816058e-05,
+      "loss": 0.0005,
+      "step": 14437
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.545829932299419e-05,
+      "loss": 0.0005,
+      "step": 14438
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5455898593172325e-05,
+      "loss": 0.0003,
+      "step": 14439
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5453497863350456e-05,
+      "loss": 0.0003,
+      "step": 14440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5451097133528595e-05,
+      "loss": 0.3685,
+      "step": 14441
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5448696403706727e-05,
+      "loss": 0.0004,
+      "step": 14442
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5446295673884862e-05,
+      "loss": 0.0004,
+      "step": 14443
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5443894944062994e-05,
+      "loss": 0.0003,
+      "step": 14444
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5441494214241132e-05,
+      "loss": 0.0002,
+      "step": 14445
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5439093484419264e-05,
+      "loss": 0.0006,
+      "step": 14446
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.54366927545974e-05,
+      "loss": 0.0006,
+      "step": 14447
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.543429202477553e-05,
+      "loss": 0.0005,
+      "step": 14448
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5431891294953666e-05,
+      "loss": 0.0006,
+      "step": 14449
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.54294905651318e-05,
+      "loss": 0.0003,
+      "step": 14450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5427089835309937e-05,
+      "loss": 0.0005,
+      "step": 14451
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.542468910548807e-05,
+      "loss": 0.0003,
+      "step": 14452
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5422288375666204e-05,
+      "loss": 0.0004,
+      "step": 14453
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.541988764584434e-05,
+      "loss": 0.0004,
+      "step": 14454
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.541748691602247e-05,
+      "loss": 0.0004,
+      "step": 14455
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5415086186200606e-05,
+      "loss": 0.0004,
+      "step": 14456
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5412685456378738e-05,
+      "loss": 0.0004,
+      "step": 14457
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5410284726556873e-05,
+      "loss": 0.0006,
+      "step": 14458
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5407883996735008e-05,
+      "loss": 0.0005,
+      "step": 14459
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5405483266913143e-05,
+      "loss": 0.0007,
+      "step": 14460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5403082537091275e-05,
+      "loss": 0.0004,
+      "step": 14461
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.540068180726941e-05,
+      "loss": 0.0003,
+      "step": 14462
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5398281077447545e-05,
+      "loss": 0.0003,
+      "step": 14463
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.539588034762568e-05,
+      "loss": 0.0004,
+      "step": 14464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5393479617803812e-05,
+      "loss": 0.0003,
+      "step": 14465
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5391078887981948e-05,
+      "loss": 0.0005,
+      "step": 14466
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.538867815816008e-05,
+      "loss": 0.0004,
+      "step": 14467
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5386277428338218e-05,
+      "loss": 0.0044,
+      "step": 14468
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.538387669851635e-05,
+      "loss": 0.0003,
+      "step": 14469
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5381475968694485e-05,
+      "loss": 0.0004,
+      "step": 14470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5379075238872617e-05,
+      "loss": 0.0056,
+      "step": 14471
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5376674509050752e-05,
+      "loss": 0.0003,
+      "step": 14472
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5374273779228887e-05,
+      "loss": 0.0003,
+      "step": 14473
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5371873049407022e-05,
+      "loss": 0.0005,
+      "step": 14474
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5369472319585154e-05,
+      "loss": 0.0004,
+      "step": 14475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.536707158976329e-05,
+      "loss": 0.0003,
+      "step": 14476
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5364670859941424e-05,
+      "loss": 0.0003,
+      "step": 14477
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5362270130119556e-05,
+      "loss": 0.0005,
+      "step": 14478
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.535986940029769e-05,
+      "loss": 0.0003,
+      "step": 14479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5357468670475823e-05,
+      "loss": 0.0003,
+      "step": 14480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.535506794065396e-05,
+      "loss": 0.0006,
+      "step": 14481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5352667210832094e-05,
+      "loss": 0.9868,
+      "step": 14482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.535026648101023e-05,
+      "loss": 0.0005,
+      "step": 14483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.534786575118836e-05,
+      "loss": 0.0003,
+      "step": 14484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5345465021366496e-05,
+      "loss": 0.3531,
+      "step": 14485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.534306429154463e-05,
+      "loss": 0.0003,
+      "step": 14486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5340663561722766e-05,
+      "loss": 0.0005,
+      "step": 14487
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5338262831900898e-05,
+      "loss": 0.0003,
+      "step": 14488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5335862102079033e-05,
+      "loss": 0.0002,
+      "step": 14489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5333461372257165e-05,
+      "loss": 0.0003,
+      "step": 14490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5331060642435304e-05,
+      "loss": 0.0003,
+      "step": 14491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5328659912613435e-05,
+      "loss": 0.0006,
+      "step": 14492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.532625918279157e-05,
+      "loss": 0.0004,
+      "step": 14493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5323858452969702e-05,
+      "loss": 0.0005,
+      "step": 14494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5321457723147838e-05,
+      "loss": 0.0005,
+      "step": 14495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5319056993325973e-05,
+      "loss": 0.0002,
+      "step": 14496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5316656263504105e-05,
+      "loss": 0.0004,
+      "step": 14497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.531425553368224e-05,
+      "loss": 0.0003,
+      "step": 14498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.531185480386037e-05,
+      "loss": 0.0003,
+      "step": 14499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.530945407403851e-05,
+      "loss": 0.0006,
+      "step": 14500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5307053344216642e-05,
+      "loss": 0.0004,
+      "step": 14501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5304652614394777e-05,
+      "loss": 0.0003,
+      "step": 14502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.530225188457291e-05,
+      "loss": 0.0003,
+      "step": 14503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5299851154751044e-05,
+      "loss": 0.0004,
+      "step": 14504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.529745042492918e-05,
+      "loss": 0.0005,
+      "step": 14505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5295049695107314e-05,
+      "loss": 0.0005,
+      "step": 14506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5292648965285446e-05,
+      "loss": 0.0005,
+      "step": 14507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.529024823546358e-05,
+      "loss": 0.0004,
+      "step": 14508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5287847505641717e-05,
+      "loss": 0.0003,
+      "step": 14509
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5285446775819852e-05,
+      "loss": 0.0004,
+      "step": 14510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5283046045997984e-05,
+      "loss": 0.0004,
+      "step": 14511
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.528064531617612e-05,
+      "loss": 0.0004,
+      "step": 14512
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.527824458635425e-05,
+      "loss": 0.0004,
+      "step": 14513
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5275843856532386e-05,
+      "loss": 0.0004,
+      "step": 14514
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.527344312671052e-05,
+      "loss": 0.0003,
+      "step": 14515
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5271042396888656e-05,
+      "loss": 0.0005,
+      "step": 14516
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5268641667066788e-05,
+      "loss": 0.0003,
+      "step": 14517
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5266240937244923e-05,
+      "loss": 0.0005,
+      "step": 14518
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.526384020742306e-05,
+      "loss": 0.0003,
+      "step": 14519
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.526143947760119e-05,
+      "loss": 0.0003,
+      "step": 14520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5259038747779325e-05,
+      "loss": 0.0005,
+      "step": 14521
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5256638017957459e-05,
+      "loss": 0.0004,
+      "step": 14522
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5254237288135596e-05,
+      "loss": 0.0003,
+      "step": 14523
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.525183655831373e-05,
+      "loss": 0.0006,
+      "step": 14524
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5249435828491863e-05,
+      "loss": 0.0004,
+      "step": 14525
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5247035098669996e-05,
+      "loss": 0.0004,
+      "step": 14526
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5244634368848131e-05,
+      "loss": 0.0004,
+      "step": 14527
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5242233639026265e-05,
+      "loss": 0.0005,
+      "step": 14528
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5239832909204398e-05,
+      "loss": 0.0004,
+      "step": 14529
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5237432179382532e-05,
+      "loss": 0.0003,
+      "step": 14530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5235031449560665e-05,
+      "loss": 0.0005,
+      "step": 14531
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5232630719738802e-05,
+      "loss": 0.0004,
+      "step": 14532
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5230229989916936e-05,
+      "loss": 0.0004,
+      "step": 14533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.522782926009507e-05,
+      "loss": 0.0004,
+      "step": 14534
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5225428530273203e-05,
+      "loss": 0.0004,
+      "step": 14535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5223027800451336e-05,
+      "loss": 0.0005,
+      "step": 14536
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5220627070629473e-05,
+      "loss": 0.0003,
+      "step": 14537
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5218226340807607e-05,
+      "loss": 0.0004,
+      "step": 14538
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.521582561098574e-05,
+      "loss": 0.0003,
+      "step": 14539
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5213424881163874e-05,
+      "loss": 0.0004,
+      "step": 14540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.521102415134201e-05,
+      "loss": 0.0003,
+      "step": 14541
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5208623421520144e-05,
+      "loss": 0.0005,
+      "step": 14542
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5206222691698277e-05,
+      "loss": 0.0004,
+      "step": 14543
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5203821961876411e-05,
+      "loss": 0.0002,
+      "step": 14544
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5201421232054544e-05,
+      "loss": 0.0004,
+      "step": 14545
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.519902050223268e-05,
+      "loss": 0.0005,
+      "step": 14546
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5196619772410813e-05,
+      "loss": 0.0003,
+      "step": 14547
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5194219042588947e-05,
+      "loss": 0.0007,
+      "step": 14548
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5191818312767082e-05,
+      "loss": 0.0003,
+      "step": 14549
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5189417582945217e-05,
+      "loss": 0.0192,
+      "step": 14550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.518701685312335e-05,
+      "loss": 0.0004,
+      "step": 14551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5184616123301484e-05,
+      "loss": 0.0003,
+      "step": 14552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5182215393479618e-05,
+      "loss": 0.0003,
+      "step": 14553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5179814663657751e-05,
+      "loss": 0.0003,
+      "step": 14554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5177413933835888e-05,
+      "loss": 0.0114,
+      "step": 14555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5175013204014021e-05,
+      "loss": 0.0004,
+      "step": 14556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5172612474192155e-05,
+      "loss": 0.0003,
+      "step": 14557
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5170211744370288e-05,
+      "loss": 0.0003,
+      "step": 14558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5167811014548425e-05,
+      "loss": 0.0065,
+      "step": 14559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5165410284726559e-05,
+      "loss": 0.0003,
+      "step": 14560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5163009554904692e-05,
+      "loss": 0.0004,
+      "step": 14561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5160608825082826e-05,
+      "loss": 0.0003,
+      "step": 14562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.515820809526096e-05,
+      "loss": 0.3274,
+      "step": 14563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5155807365439096e-05,
+      "loss": 0.0004,
+      "step": 14564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.515340663561723e-05,
+      "loss": 0.0005,
+      "step": 14565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5151005905795363e-05,
+      "loss": 0.0004,
+      "step": 14566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5148605175973497e-05,
+      "loss": 0.0002,
+      "step": 14567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.514620444615163e-05,
+      "loss": 0.0003,
+      "step": 14568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5143803716329765e-05,
+      "loss": 0.0004,
+      "step": 14569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5141402986507899e-05,
+      "loss": 0.0004,
+      "step": 14570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5139002256686032e-05,
+      "loss": 0.0002,
+      "step": 14571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5136601526864166e-05,
+      "loss": 0.8126,
+      "step": 14572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5134200797042303e-05,
+      "loss": 0.0002,
+      "step": 14573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5131800067220436e-05,
+      "loss": 0.0004,
+      "step": 14574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.512939933739857e-05,
+      "loss": 0.0004,
+      "step": 14575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5126998607576703e-05,
+      "loss": 0.0006,
+      "step": 14576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5124597877754837e-05,
+      "loss": 0.0002,
+      "step": 14577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5122197147932973e-05,
+      "loss": 0.0004,
+      "step": 14578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5119796418111107e-05,
+      "loss": 0.0003,
+      "step": 14579
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.511739568828924e-05,
+      "loss": 0.0005,
+      "step": 14580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5114994958467374e-05,
+      "loss": 0.0003,
+      "step": 14581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5112594228645511e-05,
+      "loss": 0.0003,
+      "step": 14582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5110193498823644e-05,
+      "loss": 0.0012,
+      "step": 14583
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5107792769001778e-05,
+      "loss": 0.0009,
+      "step": 14584
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5105392039179911e-05,
+      "loss": 0.0003,
+      "step": 14585
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5102991309358045e-05,
+      "loss": 0.0003,
+      "step": 14586
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.510059057953618e-05,
+      "loss": 0.0008,
+      "step": 14587
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5098189849714314e-05,
+      "loss": 0.0004,
+      "step": 14588
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5095789119892447e-05,
+      "loss": 0.0015,
+      "step": 14589
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5093388390070582e-05,
+      "loss": 0.0003,
+      "step": 14590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5090987660248717e-05,
+      "loss": 0.0004,
+      "step": 14591
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5088586930426851e-05,
+      "loss": 0.0003,
+      "step": 14592
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5086186200604984e-05,
+      "loss": 0.0006,
+      "step": 14593
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5083785470783118e-05,
+      "loss": 0.0003,
+      "step": 14594
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5081384740961251e-05,
+      "loss": 0.0008,
+      "step": 14595
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5078984011139388e-05,
+      "loss": 0.0003,
+      "step": 14596
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5076583281317522e-05,
+      "loss": 0.0004,
+      "step": 14597
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5074182551495655e-05,
+      "loss": 0.0007,
+      "step": 14598
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5071781821673789e-05,
+      "loss": 0.0008,
+      "step": 14599
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5069381091851922e-05,
+      "loss": 0.0003,
+      "step": 14600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5066980362030059e-05,
+      "loss": 0.0004,
+      "step": 14601
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5064579632208193e-05,
+      "loss": 0.0008,
+      "step": 14602
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5062178902386326e-05,
+      "loss": 0.0003,
+      "step": 14603
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.505977817256446e-05,
+      "loss": 0.0003,
+      "step": 14604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5057377442742596e-05,
+      "loss": 0.0008,
+      "step": 14605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.505497671292073e-05,
+      "loss": 0.0006,
+      "step": 14606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5052575983098863e-05,
+      "loss": 0.0101,
+      "step": 14607
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5050175253276997e-05,
+      "loss": 0.0006,
+      "step": 14608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.504777452345513e-05,
+      "loss": 0.0003,
+      "step": 14609
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5045373793633266e-05,
+      "loss": 0.9535,
+      "step": 14610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5042973063811399e-05,
+      "loss": 0.0003,
+      "step": 14611
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5040572333989533e-05,
+      "loss": 0.699,
+      "step": 14612
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5038171604167666e-05,
+      "loss": 0.0003,
+      "step": 14613
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5035770874345803e-05,
+      "loss": 0.0004,
+      "step": 14614
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5033370144523937e-05,
+      "loss": 0.0002,
+      "step": 14615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.503096941470207e-05,
+      "loss": 0.0003,
+      "step": 14616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5028568684880203e-05,
+      "loss": 0.0013,
+      "step": 14617
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5026167955058337e-05,
+      "loss": 0.0004,
+      "step": 14618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5023767225236474e-05,
+      "loss": 0.001,
+      "step": 14619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5021366495414607e-05,
+      "loss": 0.0011,
+      "step": 14620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5018965765592741e-05,
+      "loss": 0.0013,
+      "step": 14621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5016565035770874e-05,
+      "loss": 0.0002,
+      "step": 14622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5014164305949011e-05,
+      "loss": 0.0004,
+      "step": 14623
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5011763576127145e-05,
+      "loss": 0.0012,
+      "step": 14624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5009362846305278e-05,
+      "loss": 0.0004,
+      "step": 14625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5006962116483412e-05,
+      "loss": 0.0008,
+      "step": 14626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5004561386661545e-05,
+      "loss": 0.0004,
+      "step": 14627
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.500216065683968e-05,
+      "loss": 0.9563,
+      "step": 14628
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4999759927017814e-05,
+      "loss": 0.0004,
+      "step": 14629
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4997359197195947e-05,
+      "loss": 0.0017,
+      "step": 14630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4994958467374081e-05,
+      "loss": 0.0003,
+      "step": 14631
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4992557737552216e-05,
+      "loss": 0.0004,
+      "step": 14632
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4990157007730351e-05,
+      "loss": 0.0004,
+      "step": 14633
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4987756277908485e-05,
+      "loss": 0.0016,
+      "step": 14634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4985355548086618e-05,
+      "loss": 0.0004,
+      "step": 14635
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4982954818264752e-05,
+      "loss": 0.0004,
+      "step": 14636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4980554088442889e-05,
+      "loss": 0.0026,
+      "step": 14637
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4978153358621022e-05,
+      "loss": 0.0004,
+      "step": 14638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4975752628799156e-05,
+      "loss": 0.0013,
+      "step": 14639
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4973351898977289e-05,
+      "loss": 0.0013,
+      "step": 14640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4970951169155423e-05,
+      "loss": 0.0003,
+      "step": 14641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.496855043933356e-05,
+      "loss": 0.0003,
+      "step": 14642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4966149709511693e-05,
+      "loss": 0.0003,
+      "step": 14643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4963748979689826e-05,
+      "loss": 0.0011,
+      "step": 14644
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.496134824986796e-05,
+      "loss": 0.0005,
+      "step": 14645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4958947520046095e-05,
+      "loss": 0.0006,
+      "step": 14646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.495654679022423e-05,
+      "loss": 0.0011,
+      "step": 14647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4954146060402364e-05,
+      "loss": 0.0011,
+      "step": 14648
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4951745330580497e-05,
+      "loss": 0.0003,
+      "step": 14649
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.494934460075863e-05,
+      "loss": 0.0004,
+      "step": 14650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4946943870936766e-05,
+      "loss": 0.001,
+      "step": 14651
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.49445431411149e-05,
+      "loss": 0.0081,
+      "step": 14652
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4942142411293033e-05,
+      "loss": 0.0011,
+      "step": 14653
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4939741681471167e-05,
+      "loss": 0.0005,
+      "step": 14654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4937340951649303e-05,
+      "loss": 0.0011,
+      "step": 14655
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4934940221827437e-05,
+      "loss": 0.0005,
+      "step": 14656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.493253949200557e-05,
+      "loss": 0.0014,
+      "step": 14657
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4930138762183704e-05,
+      "loss": 0.0004,
+      "step": 14658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4927738032361837e-05,
+      "loss": 0.0009,
+      "step": 14659
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4925337302539974e-05,
+      "loss": 0.0005,
+      "step": 14660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4922936572718108e-05,
+      "loss": 0.0004,
+      "step": 14661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4920535842896241e-05,
+      "loss": 0.0004,
+      "step": 14662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4918135113074375e-05,
+      "loss": 0.0004,
+      "step": 14663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4915734383252508e-05,
+      "loss": 0.0004,
+      "step": 14664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4913333653430645e-05,
+      "loss": 0.0011,
+      "step": 14665
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4910932923608779e-05,
+      "loss": 0.0003,
+      "step": 14666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4908532193786912e-05,
+      "loss": 0.0003,
+      "step": 14667
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4906131463965046e-05,
+      "loss": 0.0005,
+      "step": 14668
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.490373073414318e-05,
+      "loss": 0.0004,
+      "step": 14669
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4901330004321314e-05,
+      "loss": 0.0012,
+      "step": 14670
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4898929274499448e-05,
+      "loss": 0.0004,
+      "step": 14671
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4896528544677581e-05,
+      "loss": 0.0003,
+      "step": 14672
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4894127814855716e-05,
+      "loss": 0.0004,
+      "step": 14673
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4891727085033852e-05,
+      "loss": 0.0003,
+      "step": 14674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4889326355211985e-05,
+      "loss": 0.0004,
+      "step": 14675
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4886925625390119e-05,
+      "loss": 0.0007,
+      "step": 14676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4884524895568252e-05,
+      "loss": 0.0003,
+      "step": 14677
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4882124165746389e-05,
+      "loss": 0.0008,
+      "step": 14678
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4879723435924522e-05,
+      "loss": 0.0003,
+      "step": 14679
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4877322706102656e-05,
+      "loss": 0.0097,
+      "step": 14680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.487492197628079e-05,
+      "loss": 0.0003,
+      "step": 14681
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4872521246458923e-05,
+      "loss": 0.0105,
+      "step": 14682
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.487012051663706e-05,
+      "loss": 0.0026,
+      "step": 14683
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4867719786815193e-05,
+      "loss": 0.0004,
+      "step": 14684
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4865319056993327e-05,
+      "loss": 0.0004,
+      "step": 14685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.486291832717146e-05,
+      "loss": 0.0004,
+      "step": 14686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4860517597349596e-05,
+      "loss": 0.0004,
+      "step": 14687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.485811686752773e-05,
+      "loss": 0.0004,
+      "step": 14688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4855716137705864e-05,
+      "loss": 0.0017,
+      "step": 14689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4853315407883998e-05,
+      "loss": 0.0003,
+      "step": 14690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4850914678062131e-05,
+      "loss": 0.0003,
+      "step": 14691
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4848513948240266e-05,
+      "loss": 0.0003,
+      "step": 14692
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.48461132184184e-05,
+      "loss": 0.0003,
+      "step": 14693
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4843712488596533e-05,
+      "loss": 0.0004,
+      "step": 14694
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4841311758774667e-05,
+      "loss": 0.0007,
+      "step": 14695
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.48389110289528e-05,
+      "loss": 0.0004,
+      "step": 14696
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4836510299130937e-05,
+      "loss": 0.377,
+      "step": 14697
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.483410956930907e-05,
+      "loss": 0.001,
+      "step": 14698
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4831708839487204e-05,
+      "loss": 0.0009,
+      "step": 14699
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4829308109665338e-05,
+      "loss": 0.0004,
+      "step": 14700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4826907379843475e-05,
+      "loss": 0.0004,
+      "step": 14701
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4824506650021608e-05,
+      "loss": 0.0004,
+      "step": 14702
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4822105920199742e-05,
+      "loss": 0.0004,
+      "step": 14703
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4819705190377875e-05,
+      "loss": 0.0004,
+      "step": 14704
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4817304460556009e-05,
+      "loss": 0.0003,
+      "step": 14705
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4814903730734145e-05,
+      "loss": 0.0003,
+      "step": 14706
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4812503000912279e-05,
+      "loss": 0.0004,
+      "step": 14707
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4810102271090412e-05,
+      "loss": 0.0003,
+      "step": 14708
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4807701541268546e-05,
+      "loss": 0.0008,
+      "step": 14709
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4805300811446681e-05,
+      "loss": 0.0003,
+      "step": 14710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4802900081624815e-05,
+      "loss": 0.0003,
+      "step": 14711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4800499351802948e-05,
+      "loss": 0.0003,
+      "step": 14712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4798098621981082e-05,
+      "loss": 0.0003,
+      "step": 14713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4795697892159217e-05,
+      "loss": 0.0008,
+      "step": 14714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4793297162337352e-05,
+      "loss": 0.0004,
+      "step": 14715
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4790896432515486e-05,
+      "loss": 0.0003,
+      "step": 14716
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4788495702693619e-05,
+      "loss": 0.0002,
+      "step": 14717
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4786094972871752e-05,
+      "loss": 0.0003,
+      "step": 14718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.478369424304989e-05,
+      "loss": 0.0003,
+      "step": 14719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4781293513228023e-05,
+      "loss": 0.0016,
+      "step": 14720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4778892783406156e-05,
+      "loss": 0.0004,
+      "step": 14721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.477649205358429e-05,
+      "loss": 0.0004,
+      "step": 14722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4774091323762423e-05,
+      "loss": 0.0004,
+      "step": 14723
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.477169059394056e-05,
+      "loss": 0.0002,
+      "step": 14724
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4769289864118694e-05,
+      "loss": 0.0003,
+      "step": 14725
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4766889134296827e-05,
+      "loss": 0.0004,
+      "step": 14726
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.476448840447496e-05,
+      "loss": 0.0003,
+      "step": 14727
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4762087674653094e-05,
+      "loss": 0.0004,
+      "step": 14728
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4759686944831231e-05,
+      "loss": 0.0006,
+      "step": 14729
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4757286215009365e-05,
+      "loss": 0.0004,
+      "step": 14730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4754885485187498e-05,
+      "loss": 0.0011,
+      "step": 14731
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4752484755365632e-05,
+      "loss": 0.0007,
+      "step": 14732
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4750084025543767e-05,
+      "loss": 0.0002,
+      "step": 14733
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.47476832957219e-05,
+      "loss": 0.0003,
+      "step": 14734
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4745282565900034e-05,
+      "loss": 0.9136,
+      "step": 14735
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4742881836078167e-05,
+      "loss": 0.0003,
+      "step": 14736
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.47404811062563e-05,
+      "loss": 0.0002,
+      "step": 14737
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4738080376434438e-05,
+      "loss": 0.0012,
+      "step": 14738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4735679646612571e-05,
+      "loss": 0.006,
+      "step": 14739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4733278916790705e-05,
+      "loss": 0.0002,
+      "step": 14740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4730878186968838e-05,
+      "loss": 0.0004,
+      "step": 14741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4728477457146975e-05,
+      "loss": 0.0011,
+      "step": 14742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4726076727325108e-05,
+      "loss": 0.0008,
+      "step": 14743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4723675997503242e-05,
+      "loss": 0.0003,
+      "step": 14744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4721275267681375e-05,
+      "loss": 0.0003,
+      "step": 14745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4718874537859509e-05,
+      "loss": 0.0004,
+      "step": 14746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4716473808037646e-05,
+      "loss": 0.0003,
+      "step": 14747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.471407307821578e-05,
+      "loss": 0.0004,
+      "step": 14748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4711672348393913e-05,
+      "loss": 0.0004,
+      "step": 14749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4709271618572046e-05,
+      "loss": 0.0003,
+      "step": 14750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.470687088875018e-05,
+      "loss": 0.0003,
+      "step": 14751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4704470158928315e-05,
+      "loss": 0.0006,
+      "step": 14752
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4702069429106449e-05,
+      "loss": 0.0004,
+      "step": 14753
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4699668699284582e-05,
+      "loss": 0.0004,
+      "step": 14754
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4697267969462716e-05,
+      "loss": 0.0003,
+      "step": 14755
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4694867239640852e-05,
+      "loss": 0.0004,
+      "step": 14756
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4692466509818986e-05,
+      "loss": 0.0004,
+      "step": 14757
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.469006577999712e-05,
+      "loss": 0.0007,
+      "step": 14758
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4687665050175253e-05,
+      "loss": 0.0003,
+      "step": 14759
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4685264320353386e-05,
+      "loss": 0.0004,
+      "step": 14760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4682863590531523e-05,
+      "loss": 0.0003,
+      "step": 14761
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4680462860709657e-05,
+      "loss": 0.0007,
+      "step": 14762
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.467806213088779e-05,
+      "loss": 0.0003,
+      "step": 14763
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4675661401065924e-05,
+      "loss": 0.0003,
+      "step": 14764
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.467326067124406e-05,
+      "loss": 0.0003,
+      "step": 14765
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4670859941422194e-05,
+      "loss": 0.0006,
+      "step": 14766
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4668459211600328e-05,
+      "loss": 0.0004,
+      "step": 14767
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4666058481778461e-05,
+      "loss": 0.0009,
+      "step": 14768
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4663657751956595e-05,
+      "loss": 0.0005,
+      "step": 14769
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.466125702213473e-05,
+      "loss": 0.0006,
+      "step": 14770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4658856292312865e-05,
+      "loss": 0.0003,
+      "step": 14771
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4656455562490998e-05,
+      "loss": 0.973,
+      "step": 14772
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4654054832669132e-05,
+      "loss": 0.0004,
+      "step": 14773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4651654102847267e-05,
+      "loss": 0.0004,
+      "step": 14774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.46492533730254e-05,
+      "loss": 0.0002,
+      "step": 14775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4646852643203534e-05,
+      "loss": 0.0006,
+      "step": 14776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4644451913381668e-05,
+      "loss": 0.0007,
+      "step": 14777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4642051183559801e-05,
+      "loss": 0.0006,
+      "step": 14778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4639650453737938e-05,
+      "loss": 0.0004,
+      "step": 14779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4637249723916072e-05,
+      "loss": 0.0007,
+      "step": 14780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4634848994094205e-05,
+      "loss": 0.0004,
+      "step": 14781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4632448264272338e-05,
+      "loss": 0.0004,
+      "step": 14782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4630047534450472e-05,
+      "loss": 0.0003,
+      "step": 14783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4627646804628609e-05,
+      "loss": 0.0004,
+      "step": 14784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4625246074806742e-05,
+      "loss": 0.0003,
+      "step": 14785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4622845344984876e-05,
+      "loss": 0.0005,
+      "step": 14786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.462044461516301e-05,
+      "loss": 0.0004,
+      "step": 14787
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4618043885341146e-05,
+      "loss": 0.0008,
+      "step": 14788
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.461564315551928e-05,
+      "loss": 0.0008,
+      "step": 14789
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4613242425697413e-05,
+      "loss": 0.0006,
+      "step": 14790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4610841695875547e-05,
+      "loss": 0.0003,
+      "step": 14791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.460844096605368e-05,
+      "loss": 0.0006,
+      "step": 14792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4606040236231815e-05,
+      "loss": 0.0004,
+      "step": 14793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4603639506409949e-05,
+      "loss": 0.0003,
+      "step": 14794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4601238776588082e-05,
+      "loss": 0.0003,
+      "step": 14795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4598838046766216e-05,
+      "loss": 0.0012,
+      "step": 14796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4596437316944353e-05,
+      "loss": 0.0005,
+      "step": 14797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4594036587122486e-05,
+      "loss": 0.0003,
+      "step": 14798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.459163585730062e-05,
+      "loss": 0.0006,
+      "step": 14799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4589235127478753e-05,
+      "loss": 0.0004,
+      "step": 14800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4586834397656887e-05,
+      "loss": 0.0003,
+      "step": 14801
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4584433667835024e-05,
+      "loss": 0.0004,
+      "step": 14802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4582032938013157e-05,
+      "loss": 0.0004,
+      "step": 14803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.457963220819129e-05,
+      "loss": 0.3967,
+      "step": 14804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4577231478369424e-05,
+      "loss": 0.0007,
+      "step": 14805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4574830748547561e-05,
+      "loss": 0.0004,
+      "step": 14806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4572430018725694e-05,
+      "loss": 0.0003,
+      "step": 14807
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4570029288903828e-05,
+      "loss": 0.0003,
+      "step": 14808
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4567628559081961e-05,
+      "loss": 0.0003,
+      "step": 14809
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4565227829260095e-05,
+      "loss": 0.0004,
+      "step": 14810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.456282709943823e-05,
+      "loss": 0.0005,
+      "step": 14811
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4560426369616365e-05,
+      "loss": 0.0005,
+      "step": 14812
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4558025639794499e-05,
+      "loss": 0.0009,
+      "step": 14813
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4555624909972632e-05,
+      "loss": 0.0004,
+      "step": 14814
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4553224180150766e-05,
+      "loss": 0.0007,
+      "step": 14815
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4550823450328901e-05,
+      "loss": 0.0004,
+      "step": 14816
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4548422720507035e-05,
+      "loss": 0.0006,
+      "step": 14817
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4546021990685168e-05,
+      "loss": 0.0006,
+      "step": 14818
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4543621260863301e-05,
+      "loss": 0.0012,
+      "step": 14819
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4541220531041438e-05,
+      "loss": 0.0003,
+      "step": 14820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4538819801219572e-05,
+      "loss": 0.0004,
+      "step": 14821
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4536419071397705e-05,
+      "loss": 0.0004,
+      "step": 14822
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4534018341575839e-05,
+      "loss": 0.0002,
+      "step": 14823
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4531617611753972e-05,
+      "loss": 0.0007,
+      "step": 14824
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.452921688193211e-05,
+      "loss": 0.0006,
+      "step": 14825
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4526816152110243e-05,
+      "loss": 0.0005,
+      "step": 14826
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4524415422288376e-05,
+      "loss": 0.4385,
+      "step": 14827
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.452201469246651e-05,
+      "loss": 0.0003,
+      "step": 14828
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4519613962644647e-05,
+      "loss": 0.0003,
+      "step": 14829
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.451721323282278e-05,
+      "loss": 0.0006,
+      "step": 14830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4514812503000914e-05,
+      "loss": 0.0003,
+      "step": 14831
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4512411773179047e-05,
+      "loss": 0.9194,
+      "step": 14832
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.451001104335718e-05,
+      "loss": 0.0004,
+      "step": 14833
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4507610313535316e-05,
+      "loss": 0.0002,
+      "step": 14834
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.450520958371345e-05,
+      "loss": 0.0005,
+      "step": 14835
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4502808853891583e-05,
+      "loss": 0.0003,
+      "step": 14836
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4500408124069716e-05,
+      "loss": 0.0004,
+      "step": 14837
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4498007394247853e-05,
+      "loss": 0.0002,
+      "step": 14838
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4495606664425987e-05,
+      "loss": 0.0004,
+      "step": 14839
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.449320593460412e-05,
+      "loss": 0.0003,
+      "step": 14840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4490805204782254e-05,
+      "loss": 0.0004,
+      "step": 14841
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4488404474960387e-05,
+      "loss": 0.0003,
+      "step": 14842
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4486003745138524e-05,
+      "loss": 0.0007,
+      "step": 14843
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4483603015316657e-05,
+      "loss": 0.0009,
+      "step": 14844
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4481202285494791e-05,
+      "loss": 0.0006,
+      "step": 14845
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4478801555672924e-05,
+      "loss": 0.0007,
+      "step": 14846
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4476400825851058e-05,
+      "loss": 0.0003,
+      "step": 14847
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4474000096029195e-05,
+      "loss": 0.0003,
+      "step": 14848
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4471599366207328e-05,
+      "loss": 0.0007,
+      "step": 14849
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4469198636385462e-05,
+      "loss": 0.0004,
+      "step": 14850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4466797906563595e-05,
+      "loss": 0.0003,
+      "step": 14851
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.446439717674173e-05,
+      "loss": 0.001,
+      "step": 14852
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4461996446919866e-05,
+      "loss": 0.0004,
+      "step": 14853
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4459595717098e-05,
+      "loss": 0.0007,
+      "step": 14854
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4457194987276133e-05,
+      "loss": 0.001,
+      "step": 14855
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4454794257454266e-05,
+      "loss": 0.0004,
+      "step": 14856
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4452393527632401e-05,
+      "loss": 0.001,
+      "step": 14857
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4449992797810535e-05,
+      "loss": 0.0004,
+      "step": 14858
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4447592067988668e-05,
+      "loss": 0.0008,
+      "step": 14859
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4445191338166802e-05,
+      "loss": 0.0004,
+      "step": 14860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4442790608344939e-05,
+      "loss": 0.0006,
+      "step": 14861
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4440389878523072e-05,
+      "loss": 0.0005,
+      "step": 14862
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4437989148701206e-05,
+      "loss": 0.0005,
+      "step": 14863
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.443558841887934e-05,
+      "loss": 0.0003,
+      "step": 14864
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4433187689057473e-05,
+      "loss": 0.0004,
+      "step": 14865
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.443078695923561e-05,
+      "loss": 0.0006,
+      "step": 14866
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4428386229413743e-05,
+      "loss": 0.0005,
+      "step": 14867
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4425985499591877e-05,
+      "loss": 0.0007,
+      "step": 14868
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.442358476977001e-05,
+      "loss": 0.0004,
+      "step": 14869
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4421184039948147e-05,
+      "loss": 0.0005,
+      "step": 14870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.441878331012628e-05,
+      "loss": 0.0004,
+      "step": 14871
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4416382580304414e-05,
+      "loss": 0.0007,
+      "step": 14872
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4413981850482547e-05,
+      "loss": 0.0006,
+      "step": 14873
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4411581120660681e-05,
+      "loss": 0.0003,
+      "step": 14874
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4409180390838816e-05,
+      "loss": 0.4823,
+      "step": 14875
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.440677966101695e-05,
+      "loss": 0.0005,
+      "step": 14876
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4404378931195083e-05,
+      "loss": 0.944,
+      "step": 14877
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4401978201373217e-05,
+      "loss": 0.0005,
+      "step": 14878
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.439957747155135e-05,
+      "loss": 0.0003,
+      "step": 14879
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4397176741729487e-05,
+      "loss": 0.0003,
+      "step": 14880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.439477601190762e-05,
+      "loss": 0.0005,
+      "step": 14881
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4392375282085754e-05,
+      "loss": 0.0005,
+      "step": 14882
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4389974552263887e-05,
+      "loss": 0.0003,
+      "step": 14883
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4387573822442024e-05,
+      "loss": 0.0006,
+      "step": 14884
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4385173092620158e-05,
+      "loss": 0.0007,
+      "step": 14885
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4382772362798291e-05,
+      "loss": 0.0003,
+      "step": 14886
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4380371632976425e-05,
+      "loss": 0.0005,
+      "step": 14887
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4377970903154558e-05,
+      "loss": 0.0004,
+      "step": 14888
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4375570173332695e-05,
+      "loss": 0.0005,
+      "step": 14889
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4373169443510829e-05,
+      "loss": 0.0063,
+      "step": 14890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4370768713688962e-05,
+      "loss": 0.0004,
+      "step": 14891
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4368367983867096e-05,
+      "loss": 0.0006,
+      "step": 14892
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4365967254045231e-05,
+      "loss": 0.0004,
+      "step": 14893
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4363566524223364e-05,
+      "loss": 0.0005,
+      "step": 14894
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.43611657944015e-05,
+      "loss": 0.0003,
+      "step": 14895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4358765064579633e-05,
+      "loss": 0.0005,
+      "step": 14896
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4356364334757767e-05,
+      "loss": 0.0004,
+      "step": 14897
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4353963604935902e-05,
+      "loss": 0.0004,
+      "step": 14898
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4351562875114035e-05,
+      "loss": 0.0007,
+      "step": 14899
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4349162145292169e-05,
+      "loss": 0.0005,
+      "step": 14900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4346761415470302e-05,
+      "loss": 0.0005,
+      "step": 14901
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4344360685648439e-05,
+      "loss": 0.0004,
+      "step": 14902
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4341959955826573e-05,
+      "loss": 0.0004,
+      "step": 14903
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4339559226004706e-05,
+      "loss": 0.0007,
+      "step": 14904
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.433715849618284e-05,
+      "loss": 0.0005,
+      "step": 14905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4334757766360973e-05,
+      "loss": 0.0005,
+      "step": 14906
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.433235703653911e-05,
+      "loss": 0.0003,
+      "step": 14907
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4329956306717243e-05,
+      "loss": 0.0005,
+      "step": 14908
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4327555576895377e-05,
+      "loss": 0.3625,
+      "step": 14909
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.432515484707351e-05,
+      "loss": 0.0005,
+      "step": 14910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4322754117251644e-05,
+      "loss": 0.0009,
+      "step": 14911
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4320353387429781e-05,
+      "loss": 0.0005,
+      "step": 14912
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4317952657607914e-05,
+      "loss": 0.0008,
+      "step": 14913
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4315551927786048e-05,
+      "loss": 0.0003,
+      "step": 14914
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4313151197964181e-05,
+      "loss": 0.0002,
+      "step": 14915
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4310750468142317e-05,
+      "loss": 0.0005,
+      "step": 14916
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.430834973832045e-05,
+      "loss": 0.0004,
+      "step": 14917
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4305949008498584e-05,
+      "loss": 0.0006,
+      "step": 14918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4303548278676717e-05,
+      "loss": 0.0005,
+      "step": 14919
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.430114754885485e-05,
+      "loss": 0.0005,
+      "step": 14920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4298746819032987e-05,
+      "loss": 0.0008,
+      "step": 14921
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4296346089211121e-05,
+      "loss": 0.0009,
+      "step": 14922
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4293945359389254e-05,
+      "loss": 0.0005,
+      "step": 14923
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4291544629567388e-05,
+      "loss": 0.0005,
+      "step": 14924
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4289143899745525e-05,
+      "loss": 0.0006,
+      "step": 14925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4286743169923658e-05,
+      "loss": 0.0002,
+      "step": 14926
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4284342440101792e-05,
+      "loss": 0.0006,
+      "step": 14927
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4281941710279925e-05,
+      "loss": 0.0004,
+      "step": 14928
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4279540980458059e-05,
+      "loss": 0.0005,
+      "step": 14929
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4277140250636196e-05,
+      "loss": 0.0003,
+      "step": 14930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4274739520814329e-05,
+      "loss": 0.0004,
+      "step": 14931
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4272338790992463e-05,
+      "loss": 0.0006,
+      "step": 14932
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4269938061170596e-05,
+      "loss": 0.0007,
+      "step": 14933
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4267537331348731e-05,
+      "loss": 0.0002,
+      "step": 14934
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4265136601526865e-05,
+      "loss": 0.0003,
+      "step": 14935
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4262735871705e-05,
+      "loss": 0.0002,
+      "step": 14936
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4260335141883133e-05,
+      "loss": 0.0003,
+      "step": 14937
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4257934412061267e-05,
+      "loss": 0.0009,
+      "step": 14938
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4255533682239402e-05,
+      "loss": 0.0006,
+      "step": 14939
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4253132952417536e-05,
+      "loss": 0.0004,
+      "step": 14940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4250732222595669e-05,
+      "loss": 0.0006,
+      "step": 14941
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4248331492773803e-05,
+      "loss": 0.0006,
+      "step": 14942
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4245930762951936e-05,
+      "loss": 0.0003,
+      "step": 14943
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4243530033130073e-05,
+      "loss": 0.0003,
+      "step": 14944
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4241129303308206e-05,
+      "loss": 0.0003,
+      "step": 14945
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.423872857348634e-05,
+      "loss": 0.0004,
+      "step": 14946
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4236327843664473e-05,
+      "loss": 0.0004,
+      "step": 14947
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.423392711384261e-05,
+      "loss": 0.0005,
+      "step": 14948
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4231526384020744e-05,
+      "loss": 0.0004,
+      "step": 14949
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4229125654198877e-05,
+      "loss": 0.0003,
+      "step": 14950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4226724924377011e-05,
+      "loss": 0.0003,
+      "step": 14951
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4224324194555144e-05,
+      "loss": 0.0003,
+      "step": 14952
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4221923464733281e-05,
+      "loss": 0.0005,
+      "step": 14953
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4219522734911415e-05,
+      "loss": 0.0003,
+      "step": 14954
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4217122005089548e-05,
+      "loss": 0.0004,
+      "step": 14955
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4214721275267682e-05,
+      "loss": 0.0005,
+      "step": 14956
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4212320545445817e-05,
+      "loss": 0.0034,
+      "step": 14957
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.420991981562395e-05,
+      "loss": 0.0003,
+      "step": 14958
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4207519085802084e-05,
+      "loss": 0.0003,
+      "step": 14959
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4205118355980217e-05,
+      "loss": 0.0004,
+      "step": 14960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4202717626158351e-05,
+      "loss": 0.0004,
+      "step": 14961
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4200316896336488e-05,
+      "loss": 0.0004,
+      "step": 14962
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4197916166514621e-05,
+      "loss": 0.0005,
+      "step": 14963
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4195515436692755e-05,
+      "loss": 0.0005,
+      "step": 14964
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4193114706870888e-05,
+      "loss": 0.0104,
+      "step": 14965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4190713977049025e-05,
+      "loss": 0.0003,
+      "step": 14966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4188313247227159e-05,
+      "loss": 0.0005,
+      "step": 14967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4185912517405292e-05,
+      "loss": 0.0003,
+      "step": 14968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4183511787583426e-05,
+      "loss": 0.0004,
+      "step": 14969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4181111057761559e-05,
+      "loss": 0.0005,
+      "step": 14970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4178710327939696e-05,
+      "loss": 0.0003,
+      "step": 14971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.417630959811783e-05,
+      "loss": 0.0003,
+      "step": 14972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4173908868295963e-05,
+      "loss": 0.0002,
+      "step": 14973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4171508138474096e-05,
+      "loss": 0.9482,
+      "step": 14974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.416910740865223e-05,
+      "loss": 0.0002,
+      "step": 14975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4166706678830365e-05,
+      "loss": 0.0004,
+      "step": 14976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.41643059490085e-05,
+      "loss": 0.0009,
+      "step": 14977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4161905219186634e-05,
+      "loss": 0.0004,
+      "step": 14978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4159504489364767e-05,
+      "loss": 0.0004,
+      "step": 14979
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4157103759542903e-05,
+      "loss": 0.0096,
+      "step": 14980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4154703029721036e-05,
+      "loss": 0.0004,
+      "step": 14981
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.415230229989917e-05,
+      "loss": 0.9344,
+      "step": 14982
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4149901570077303e-05,
+      "loss": 0.0004,
+      "step": 14983
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4147500840255436e-05,
+      "loss": 0.0004,
+      "step": 14984
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4145100110433573e-05,
+      "loss": 0.0003,
+      "step": 14985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4142699380611707e-05,
+      "loss": 0.0003,
+      "step": 14986
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.414029865078984e-05,
+      "loss": 0.0006,
+      "step": 14987
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4137897920967974e-05,
+      "loss": 0.7423,
+      "step": 14988
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.413549719114611e-05,
+      "loss": 0.0009,
+      "step": 14989
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4133096461324244e-05,
+      "loss": 0.0005,
+      "step": 14990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4130695731502378e-05,
+      "loss": 0.0008,
+      "step": 14991
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4128295001680511e-05,
+      "loss": 0.0007,
+      "step": 14992
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4125894271858645e-05,
+      "loss": 0.0008,
+      "step": 14993
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4123493542036782e-05,
+      "loss": 0.0005,
+      "step": 14994
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4121092812214915e-05,
+      "loss": 0.0005,
+      "step": 14995
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4118692082393049e-05,
+      "loss": 0.0003,
+      "step": 14996
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4116291352571182e-05,
+      "loss": 0.0006,
+      "step": 14997
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4113890622749317e-05,
+      "loss": 0.0004,
+      "step": 14998
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.411148989292745e-05,
+      "loss": 0.0004,
+      "step": 14999
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4109089163105584e-05,
+      "loss": 0.0006,
+      "step": 15000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4106688433283718e-05,
+      "loss": 0.0005,
+      "step": 15001
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4104287703461851e-05,
+      "loss": 0.0004,
+      "step": 15002
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4101886973639988e-05,
+      "loss": 0.0007,
+      "step": 15003
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4099486243818122e-05,
+      "loss": 0.0003,
+      "step": 15004
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4097085513996255e-05,
+      "loss": 0.0006,
+      "step": 15005
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4094684784174389e-05,
+      "loss": 0.0004,
+      "step": 15006
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4092284054352522e-05,
+      "loss": 0.0005,
+      "step": 15007
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4089883324530659e-05,
+      "loss": 0.0007,
+      "step": 15008
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4087482594708792e-05,
+      "loss": 0.0005,
+      "step": 15009
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4085081864886926e-05,
+      "loss": 0.0004,
+      "step": 15010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.408268113506506e-05,
+      "loss": 0.0009,
+      "step": 15011
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4080280405243196e-05,
+      "loss": 0.0004,
+      "step": 15012
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.407787967542133e-05,
+      "loss": 0.0006,
+      "step": 15013
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4075478945599463e-05,
+      "loss": 0.0003,
+      "step": 15014
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4073078215777597e-05,
+      "loss": 0.0007,
+      "step": 15015
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.407067748595573e-05,
+      "loss": 0.0006,
+      "step": 15016
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4068276756133866e-05,
+      "loss": 0.0011,
+      "step": 15017
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4065876026311999e-05,
+      "loss": 0.0004,
+      "step": 15018
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4063475296490134e-05,
+      "loss": 0.0003,
+      "step": 15019
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4061074566668268e-05,
+      "loss": 0.0005,
+      "step": 15020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4058673836846403e-05,
+      "loss": 0.0007,
+      "step": 15021
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4056273107024536e-05,
+      "loss": 0.0005,
+      "step": 15022
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.405387237720267e-05,
+      "loss": 0.0005,
+      "step": 15023
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4051471647380803e-05,
+      "loss": 0.0009,
+      "step": 15024
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4049070917558937e-05,
+      "loss": 0.0007,
+      "step": 15025
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4046670187737074e-05,
+      "loss": 0.0002,
+      "step": 15026
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4044269457915207e-05,
+      "loss": 0.0073,
+      "step": 15027
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.404186872809334e-05,
+      "loss": 0.0004,
+      "step": 15028
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4039467998271474e-05,
+      "loss": 0.001,
+      "step": 15029
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4037067268449611e-05,
+      "loss": 0.0003,
+      "step": 15030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4034666538627745e-05,
+      "loss": 0.001,
+      "step": 15031
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4032265808805878e-05,
+      "loss": 0.0005,
+      "step": 15032
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4029865078984012e-05,
+      "loss": 0.0005,
+      "step": 15033
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4027464349162145e-05,
+      "loss": 0.0005,
+      "step": 15034
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4025063619340282e-05,
+      "loss": 0.0003,
+      "step": 15035
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4022662889518415e-05,
+      "loss": 0.0007,
+      "step": 15036
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4020262159696549e-05,
+      "loss": 1.1192,
+      "step": 15037
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4017861429874682e-05,
+      "loss": 0.0003,
+      "step": 15038
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4015460700052816e-05,
+      "loss": 0.0004,
+      "step": 15039
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4013059970230951e-05,
+      "loss": 0.0094,
+      "step": 15040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4010659240409085e-05,
+      "loss": 0.0005,
+      "step": 15041
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4008258510587218e-05,
+      "loss": 0.0006,
+      "step": 15042
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4005857780765352e-05,
+      "loss": 0.0005,
+      "step": 15043
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4003457050943489e-05,
+      "loss": 0.0004,
+      "step": 15044
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4001056321121622e-05,
+      "loss": 0.0005,
+      "step": 15045
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3998655591299755e-05,
+      "loss": 0.0005,
+      "step": 15046
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3996254861477889e-05,
+      "loss": 0.0004,
+      "step": 15047
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3993854131656022e-05,
+      "loss": 0.0006,
+      "step": 15048
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.399145340183416e-05,
+      "loss": 0.0004,
+      "step": 15049
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3989052672012293e-05,
+      "loss": 0.0008,
+      "step": 15050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3986651942190426e-05,
+      "loss": 0.0005,
+      "step": 15051
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.398425121236856e-05,
+      "loss": 0.0005,
+      "step": 15052
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3981850482546697e-05,
+      "loss": 0.0005,
+      "step": 15053
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.397944975272483e-05,
+      "loss": 0.0002,
+      "step": 15054
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3977049022902964e-05,
+      "loss": 0.0012,
+      "step": 15055
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3974648293081097e-05,
+      "loss": 0.0005,
+      "step": 15056
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.397224756325923e-05,
+      "loss": 0.0006,
+      "step": 15057
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3969846833437366e-05,
+      "loss": 0.0006,
+      "step": 15058
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.39674461036155e-05,
+      "loss": 0.0014,
+      "step": 15059
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3965045373793635e-05,
+      "loss": 0.0003,
+      "step": 15060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3962644643971768e-05,
+      "loss": 0.0004,
+      "step": 15061
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3960243914149903e-05,
+      "loss": 0.0006,
+      "step": 15062
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3957843184328037e-05,
+      "loss": 0.0005,
+      "step": 15063
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.395544245450617e-05,
+      "loss": 0.0003,
+      "step": 15064
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3953041724684304e-05,
+      "loss": 0.0005,
+      "step": 15065
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3950640994862437e-05,
+      "loss": 0.3949,
+      "step": 15066
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3948240265040574e-05,
+      "loss": 0.0004,
+      "step": 15067
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3945839535218708e-05,
+      "loss": 0.0006,
+      "step": 15068
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3943438805396841e-05,
+      "loss": 0.0007,
+      "step": 15069
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3941038075574975e-05,
+      "loss": 0.0009,
+      "step": 15070
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3938637345753108e-05,
+      "loss": 0.001,
+      "step": 15071
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3936236615931245e-05,
+      "loss": 0.0004,
+      "step": 15072
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3933835886109378e-05,
+      "loss": 0.0005,
+      "step": 15073
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3931435156287512e-05,
+      "loss": 0.0011,
+      "step": 15074
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3929034426465645e-05,
+      "loss": 0.0004,
+      "step": 15075
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3926633696643782e-05,
+      "loss": 0.0008,
+      "step": 15076
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3924232966821916e-05,
+      "loss": 0.0003,
+      "step": 15077
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.392183223700005e-05,
+      "loss": 0.0004,
+      "step": 15078
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3919431507178183e-05,
+      "loss": 0.0007,
+      "step": 15079
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3917030777356316e-05,
+      "loss": 0.0004,
+      "step": 15080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3914630047534452e-05,
+      "loss": 0.0008,
+      "step": 15081
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3912229317712585e-05,
+      "loss": 0.0006,
+      "step": 15082
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3909828587890719e-05,
+      "loss": 0.0004,
+      "step": 15083
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3907427858068852e-05,
+      "loss": 0.0004,
+      "step": 15084
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3905027128246989e-05,
+      "loss": 0.0005,
+      "step": 15085
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3902626398425122e-05,
+      "loss": 0.0003,
+      "step": 15086
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3900225668603256e-05,
+      "loss": 0.0003,
+      "step": 15087
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.389782493878139e-05,
+      "loss": 0.0004,
+      "step": 15088
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3895424208959523e-05,
+      "loss": 0.0004,
+      "step": 15089
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.389302347913766e-05,
+      "loss": 0.0004,
+      "step": 15090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3890622749315793e-05,
+      "loss": 0.001,
+      "step": 15091
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3888222019493927e-05,
+      "loss": 0.0004,
+      "step": 15092
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.388582128967206e-05,
+      "loss": 0.0009,
+      "step": 15093
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3883420559850194e-05,
+      "loss": 0.7253,
+      "step": 15094
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.388101983002833e-05,
+      "loss": 0.0004,
+      "step": 15095
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3878619100206464e-05,
+      "loss": 0.0006,
+      "step": 15096
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3876218370384598e-05,
+      "loss": 0.0004,
+      "step": 15097
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3873817640562731e-05,
+      "loss": 0.0006,
+      "step": 15098
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3871416910740866e-05,
+      "loss": 0.0007,
+      "step": 15099
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3869016180919e-05,
+      "loss": 0.0015,
+      "step": 15100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3866615451097135e-05,
+      "loss": 0.0003,
+      "step": 15101
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3864214721275268e-05,
+      "loss": 0.0006,
+      "step": 15102
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3861813991453402e-05,
+      "loss": 0.0017,
+      "step": 15103
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3859413261631537e-05,
+      "loss": 0.0007,
+      "step": 15104
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.385701253180967e-05,
+      "loss": 0.0011,
+      "step": 15105
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3854611801987804e-05,
+      "loss": 0.0003,
+      "step": 15106
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3852211072165938e-05,
+      "loss": 0.0003,
+      "step": 15107
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3849810342344074e-05,
+      "loss": 0.0008,
+      "step": 15108
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3847409612522208e-05,
+      "loss": 0.0006,
+      "step": 15109
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3845008882700341e-05,
+      "loss": 0.0004,
+      "step": 15110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3842608152878475e-05,
+      "loss": 0.0003,
+      "step": 15111
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3840207423056608e-05,
+      "loss": 0.0004,
+      "step": 15112
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3837806693234745e-05,
+      "loss": 0.0004,
+      "step": 15113
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3835405963412879e-05,
+      "loss": 0.0004,
+      "step": 15114
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3833005233591012e-05,
+      "loss": 0.0004,
+      "step": 15115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3830604503769146e-05,
+      "loss": 0.0004,
+      "step": 15116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3828203773947283e-05,
+      "loss": 0.0003,
+      "step": 15117
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3825803044125416e-05,
+      "loss": 0.0011,
+      "step": 15118
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.382340231430355e-05,
+      "loss": 0.0004,
+      "step": 15119
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3821001584481683e-05,
+      "loss": 0.0002,
+      "step": 15120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3818600854659817e-05,
+      "loss": 0.0004,
+      "step": 15121
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3816200124837952e-05,
+      "loss": 0.0009,
+      "step": 15122
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3813799395016085e-05,
+      "loss": 0.0014,
+      "step": 15123
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3811398665194219e-05,
+      "loss": 0.0005,
+      "step": 15124
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3808997935372352e-05,
+      "loss": 0.001,
+      "step": 15125
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3806597205550486e-05,
+      "loss": 0.0005,
+      "step": 15126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3804196475728623e-05,
+      "loss": 0.0004,
+      "step": 15127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3801795745906756e-05,
+      "loss": 0.0004,
+      "step": 15128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.379939501608489e-05,
+      "loss": 0.0005,
+      "step": 15129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3796994286263023e-05,
+      "loss": 0.9701,
+      "step": 15130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.379459355644116e-05,
+      "loss": 0.0005,
+      "step": 15131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3792192826619294e-05,
+      "loss": 0.0004,
+      "step": 15132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3789792096797427e-05,
+      "loss": 0.0013,
+      "step": 15133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.378739136697556e-05,
+      "loss": 0.0011,
+      "step": 15134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3784990637153694e-05,
+      "loss": 0.0004,
+      "step": 15135
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3782589907331831e-05,
+      "loss": 0.0002,
+      "step": 15136
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3780189177509964e-05,
+      "loss": 0.0009,
+      "step": 15137
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3777788447688098e-05,
+      "loss": 0.0005,
+      "step": 15138
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3775387717866231e-05,
+      "loss": 0.0005,
+      "step": 15139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3772986988044367e-05,
+      "loss": 0.0004,
+      "step": 15140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.37705862582225e-05,
+      "loss": 0.0008,
+      "step": 15141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3768185528400634e-05,
+      "loss": 0.0008,
+      "step": 15142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3765784798578769e-05,
+      "loss": 0.001,
+      "step": 15143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3763384068756902e-05,
+      "loss": 0.9574,
+      "step": 15144
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3760983338935038e-05,
+      "loss": 0.0003,
+      "step": 15145
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3758582609113171e-05,
+      "loss": 0.0006,
+      "step": 15146
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3756181879291304e-05,
+      "loss": 0.0004,
+      "step": 15147
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3753781149469438e-05,
+      "loss": 0.0005,
+      "step": 15148
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3751380419647575e-05,
+      "loss": 0.0003,
+      "step": 15149
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3748979689825708e-05,
+      "loss": 0.0004,
+      "step": 15150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3746578960003842e-05,
+      "loss": 0.0004,
+      "step": 15151
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3744178230181975e-05,
+      "loss": 0.0012,
+      "step": 15152
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3741777500360109e-05,
+      "loss": 0.0004,
+      "step": 15153
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3739376770538246e-05,
+      "loss": 0.0005,
+      "step": 15154
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.373697604071638e-05,
+      "loss": 0.0006,
+      "step": 15155
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3734575310894513e-05,
+      "loss": 0.0006,
+      "step": 15156
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3732174581072646e-05,
+      "loss": 0.0006,
+      "step": 15157
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.372977385125078e-05,
+      "loss": 0.0003,
+      "step": 15158
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3727373121428917e-05,
+      "loss": 0.0004,
+      "step": 15159
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.372497239160705e-05,
+      "loss": 0.001,
+      "step": 15160
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3722571661785184e-05,
+      "loss": 0.0006,
+      "step": 15161
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3720170931963317e-05,
+      "loss": 0.0005,
+      "step": 15162
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3717770202141452e-05,
+      "loss": 0.0005,
+      "step": 15163
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3715369472319586e-05,
+      "loss": 0.0007,
+      "step": 15164
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.371296874249772e-05,
+      "loss": 0.001,
+      "step": 15165
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3710568012675853e-05,
+      "loss": 0.0005,
+      "step": 15166
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3708167282853986e-05,
+      "loss": 0.0004,
+      "step": 15167
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3705766553032123e-05,
+      "loss": 0.0015,
+      "step": 15168
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3703365823210257e-05,
+      "loss": 0.0004,
+      "step": 15169
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.370096509338839e-05,
+      "loss": 0.0004,
+      "step": 15170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3698564363566524e-05,
+      "loss": 0.0006,
+      "step": 15171
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.369616363374466e-05,
+      "loss": 0.0004,
+      "step": 15172
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3693762903922794e-05,
+      "loss": 0.0011,
+      "step": 15173
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3691362174100927e-05,
+      "loss": 0.0003,
+      "step": 15174
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3688961444279061e-05,
+      "loss": 0.0004,
+      "step": 15175
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3686560714457194e-05,
+      "loss": 0.0005,
+      "step": 15176
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3684159984635331e-05,
+      "loss": 0.0004,
+      "step": 15177
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3681759254813465e-05,
+      "loss": 0.0003,
+      "step": 15178
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3679358524991598e-05,
+      "loss": 0.0004,
+      "step": 15179
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3676957795169732e-05,
+      "loss": 0.0005,
+      "step": 15180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3674557065347867e-05,
+      "loss": 0.0004,
+      "step": 15181
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3672156335526e-05,
+      "loss": 0.0003,
+      "step": 15182
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3669755605704134e-05,
+      "loss": 0.0004,
+      "step": 15183
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.366735487588227e-05,
+      "loss": 0.0003,
+      "step": 15184
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3664954146060403e-05,
+      "loss": 0.0004,
+      "step": 15185
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3662553416238538e-05,
+      "loss": 0.9494,
+      "step": 15186
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3660152686416671e-05,
+      "loss": 0.0004,
+      "step": 15187
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3657751956594805e-05,
+      "loss": 0.0003,
+      "step": 15188
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3655351226772938e-05,
+      "loss": 0.0008,
+      "step": 15189
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3652950496951072e-05,
+      "loss": 0.0048,
+      "step": 15190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3650549767129209e-05,
+      "loss": 0.0008,
+      "step": 15191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3648149037307342e-05,
+      "loss": 0.0004,
+      "step": 15192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3645748307485476e-05,
+      "loss": 0.0006,
+      "step": 15193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.364334757766361e-05,
+      "loss": 0.0004,
+      "step": 15194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3640946847841746e-05,
+      "loss": 0.0004,
+      "step": 15195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.363854611801988e-05,
+      "loss": 0.001,
+      "step": 15196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3636145388198013e-05,
+      "loss": 0.0004,
+      "step": 15197
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3633744658376147e-05,
+      "loss": 0.0003,
+      "step": 15198
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.363134392855428e-05,
+      "loss": 0.0004,
+      "step": 15199
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3628943198732417e-05,
+      "loss": 0.0003,
+      "step": 15200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.362654246891055e-05,
+      "loss": 0.0008,
+      "step": 15201
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3624141739088684e-05,
+      "loss": 0.0005,
+      "step": 15202
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3621741009266817e-05,
+      "loss": 0.3367,
+      "step": 15203
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3619340279444953e-05,
+      "loss": 0.0005,
+      "step": 15204
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3616939549623086e-05,
+      "loss": 0.0005,
+      "step": 15205
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.361453881980122e-05,
+      "loss": 0.0005,
+      "step": 15206
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3612138089979353e-05,
+      "loss": 0.0014,
+      "step": 15207
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3609737360157487e-05,
+      "loss": 0.0013,
+      "step": 15208
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3607336630335624e-05,
+      "loss": 0.0006,
+      "step": 15209
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3604935900513757e-05,
+      "loss": 0.0006,
+      "step": 15210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.360253517069189e-05,
+      "loss": 0.0008,
+      "step": 15211
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3600134440870024e-05,
+      "loss": 0.0004,
+      "step": 15212
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3597733711048161e-05,
+      "loss": 0.0005,
+      "step": 15213
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3595332981226294e-05,
+      "loss": 0.0004,
+      "step": 15214
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3592932251404428e-05,
+      "loss": 0.0004,
+      "step": 15215
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3590531521582561e-05,
+      "loss": 0.0004,
+      "step": 15216
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3588130791760695e-05,
+      "loss": 0.0005,
+      "step": 15217
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3585730061938832e-05,
+      "loss": 0.0006,
+      "step": 15218
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3583329332116965e-05,
+      "loss": 0.0004,
+      "step": 15219
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3580928602295099e-05,
+      "loss": 0.0006,
+      "step": 15220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3578527872473232e-05,
+      "loss": 0.0014,
+      "step": 15221
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3576127142651366e-05,
+      "loss": 0.0006,
+      "step": 15222
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3573726412829501e-05,
+      "loss": 0.0005,
+      "step": 15223
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3571325683007634e-05,
+      "loss": 0.0004,
+      "step": 15224
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.356892495318577e-05,
+      "loss": 0.0002,
+      "step": 15225
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3566524223363903e-05,
+      "loss": 0.0004,
+      "step": 15226
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3564123493542038e-05,
+      "loss": 0.0005,
+      "step": 15227
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3561722763720172e-05,
+      "loss": 0.0008,
+      "step": 15228
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 0.0007,
+      "step": 15229
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3556921304076439e-05,
+      "loss": 0.0006,
+      "step": 15230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3554520574254572e-05,
+      "loss": 0.0003,
+      "step": 15231
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3552119844432709e-05,
+      "loss": 0.0009,
+      "step": 15232
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3549719114610843e-05,
+      "loss": 0.0004,
+      "step": 15233
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3547318384788976e-05,
+      "loss": 0.0013,
+      "step": 15234
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.354491765496711e-05,
+      "loss": 0.0004,
+      "step": 15235
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3542516925145246e-05,
+      "loss": 0.0009,
+      "step": 15236
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.354011619532338e-05,
+      "loss": 0.0005,
+      "step": 15237
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3537715465501513e-05,
+      "loss": 0.0004,
+      "step": 15238
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3535314735679647e-05,
+      "loss": 0.0005,
+      "step": 15239
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.353291400585778e-05,
+      "loss": 0.001,
+      "step": 15240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3530513276035917e-05,
+      "loss": 0.0004,
+      "step": 15241
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.352811254621405e-05,
+      "loss": 0.0003,
+      "step": 15242
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3525711816392184e-05,
+      "loss": 0.0007,
+      "step": 15243
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3523311086570318e-05,
+      "loss": 0.0005,
+      "step": 15244
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3520910356748453e-05,
+      "loss": 0.0005,
+      "step": 15245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3518509626926587e-05,
+      "loss": 0.0007,
+      "step": 15246
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.351610889710472e-05,
+      "loss": 0.0004,
+      "step": 15247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3513708167282853e-05,
+      "loss": 0.0002,
+      "step": 15248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3511307437460987e-05,
+      "loss": 0.0004,
+      "step": 15249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3508906707639124e-05,
+      "loss": 0.0003,
+      "step": 15250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3506505977817257e-05,
+      "loss": 0.0006,
+      "step": 15251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3504105247995391e-05,
+      "loss": 0.001,
+      "step": 15252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3501704518173524e-05,
+      "loss": 0.0003,
+      "step": 15253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3499303788351658e-05,
+      "loss": 0.0004,
+      "step": 15254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3496903058529795e-05,
+      "loss": 0.0004,
+      "step": 15255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3494502328707928e-05,
+      "loss": 0.0003,
+      "step": 15256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3492101598886062e-05,
+      "loss": 0.0006,
+      "step": 15257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3489700869064195e-05,
+      "loss": 0.0008,
+      "step": 15258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3487300139242332e-05,
+      "loss": 0.001,
+      "step": 15259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3484899409420466e-05,
+      "loss": 0.0004,
+      "step": 15260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3482498679598599e-05,
+      "loss": 0.0008,
+      "step": 15261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3480097949776733e-05,
+      "loss": 0.0007,
+      "step": 15262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3477697219954866e-05,
+      "loss": 0.0008,
+      "step": 15263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3475296490133001e-05,
+      "loss": 0.0007,
+      "step": 15264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3472895760311135e-05,
+      "loss": 0.0004,
+      "step": 15265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3470495030489268e-05,
+      "loss": 0.0006,
+      "step": 15266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3468094300667403e-05,
+      "loss": 0.0008,
+      "step": 15267
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3465693570845539e-05,
+      "loss": 0.0014,
+      "step": 15268
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3463292841023672e-05,
+      "loss": 0.0004,
+      "step": 15269
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3460892111201806e-05,
+      "loss": 0.0005,
+      "step": 15270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3458491381379939e-05,
+      "loss": 0.0004,
+      "step": 15271
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3456090651558073e-05,
+      "loss": 0.0002,
+      "step": 15272
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.345368992173621e-05,
+      "loss": 0.0007,
+      "step": 15273
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3451289191914343e-05,
+      "loss": 0.0007,
+      "step": 15274
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3448888462092476e-05,
+      "loss": 0.381,
+      "step": 15275
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.344648773227061e-05,
+      "loss": 0.0004,
+      "step": 15276
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3444087002448747e-05,
+      "loss": 0.0003,
+      "step": 15277
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.344168627262688e-05,
+      "loss": 0.0005,
+      "step": 15278
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3439285542805014e-05,
+      "loss": 0.0009,
+      "step": 15279
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3436884812983147e-05,
+      "loss": 0.0003,
+      "step": 15280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.343448408316128e-05,
+      "loss": 0.0007,
+      "step": 15281
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3432083353339418e-05,
+      "loss": 0.0009,
+      "step": 15282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3429682623517551e-05,
+      "loss": 0.0004,
+      "step": 15283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3427281893695685e-05,
+      "loss": 0.0005,
+      "step": 15284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3424881163873818e-05,
+      "loss": 0.0006,
+      "step": 15285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3422480434051952e-05,
+      "loss": 0.0004,
+      "step": 15286
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3420079704230087e-05,
+      "loss": 0.0004,
+      "step": 15287
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.341767897440822e-05,
+      "loss": 0.0004,
+      "step": 15288
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3415278244586354e-05,
+      "loss": 0.0006,
+      "step": 15289
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3412877514764487e-05,
+      "loss": 0.0006,
+      "step": 15290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3410476784942624e-05,
+      "loss": 0.0004,
+      "step": 15291
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3408076055120758e-05,
+      "loss": 0.0002,
+      "step": 15292
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3405675325298891e-05,
+      "loss": 0.0004,
+      "step": 15293
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3403274595477025e-05,
+      "loss": 0.0005,
+      "step": 15294
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3400873865655158e-05,
+      "loss": 0.0006,
+      "step": 15295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3398473135833295e-05,
+      "loss": 0.0004,
+      "step": 15296
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3396072406011429e-05,
+      "loss": 0.0006,
+      "step": 15297
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3393671676189562e-05,
+      "loss": 0.8017,
+      "step": 15298
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3391270946367696e-05,
+      "loss": 0.0004,
+      "step": 15299
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3388870216545832e-05,
+      "loss": 0.0006,
+      "step": 15300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3386469486723966e-05,
+      "loss": 0.0006,
+      "step": 15301
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.33840687569021e-05,
+      "loss": 0.0007,
+      "step": 15302
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3381668027080233e-05,
+      "loss": 0.0007,
+      "step": 15303
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3379267297258366e-05,
+      "loss": 0.0006,
+      "step": 15304
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3376866567436502e-05,
+      "loss": 0.0003,
+      "step": 15305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3374465837614635e-05,
+      "loss": 0.0004,
+      "step": 15306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3372065107792769e-05,
+      "loss": 0.0005,
+      "step": 15307
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3369664377970904e-05,
+      "loss": 0.0006,
+      "step": 15308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3367263648149039e-05,
+      "loss": 0.0006,
+      "step": 15309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3364862918327173e-05,
+      "loss": 0.0003,
+      "step": 15310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3362462188505306e-05,
+      "loss": 0.0007,
+      "step": 15311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.336006145868344e-05,
+      "loss": 0.0005,
+      "step": 15312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3357660728861573e-05,
+      "loss": 0.0004,
+      "step": 15313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.335525999903971e-05,
+      "loss": 0.0005,
+      "step": 15314
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3352859269217843e-05,
+      "loss": 0.0004,
+      "step": 15315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3350458539395977e-05,
+      "loss": 0.0003,
+      "step": 15316
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.334805780957411e-05,
+      "loss": 0.0005,
+      "step": 15317
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3345657079752244e-05,
+      "loss": 0.0005,
+      "step": 15318
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.334325634993038e-05,
+      "loss": 0.0003,
+      "step": 15319
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3340855620108514e-05,
+      "loss": 0.0003,
+      "step": 15320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3338454890286648e-05,
+      "loss": 0.0004,
+      "step": 15321
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3336054160464781e-05,
+      "loss": 0.0004,
+      "step": 15322
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3333653430642918e-05,
+      "loss": 0.0006,
+      "step": 15323
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3331252700821052e-05,
+      "loss": 0.0002,
+      "step": 15324
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3328851970999185e-05,
+      "loss": 0.007,
+      "step": 15325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3326451241177319e-05,
+      "loss": 0.0008,
+      "step": 15326
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3324050511355452e-05,
+      "loss": 0.0002,
+      "step": 15327
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3321649781533587e-05,
+      "loss": 0.0005,
+      "step": 15328
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.331924905171172e-05,
+      "loss": 0.0006,
+      "step": 15329
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3316848321889854e-05,
+      "loss": 0.0007,
+      "step": 15330
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3314447592067988e-05,
+      "loss": 0.0003,
+      "step": 15331
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3312046862246125e-05,
+      "loss": 0.0006,
+      "step": 15332
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3309646132424258e-05,
+      "loss": 0.0003,
+      "step": 15333
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3307245402602392e-05,
+      "loss": 0.0006,
+      "step": 15334
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3304844672780525e-05,
+      "loss": 0.7089,
+      "step": 15335
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3302443942958659e-05,
+      "loss": 0.0005,
+      "step": 15336
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3300043213136795e-05,
+      "loss": 0.0004,
+      "step": 15337
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3297642483314929e-05,
+      "loss": 0.0004,
+      "step": 15338
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3295241753493062e-05,
+      "loss": 0.0007,
+      "step": 15339
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3292841023671196e-05,
+      "loss": 0.0007,
+      "step": 15340
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3290440293849333e-05,
+      "loss": 0.0003,
+      "step": 15341
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3288039564027466e-05,
+      "loss": 0.0003,
+      "step": 15342
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.32856388342056e-05,
+      "loss": 0.0004,
+      "step": 15343
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3283238104383733e-05,
+      "loss": 0.0003,
+      "step": 15344
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3280837374561867e-05,
+      "loss": 0.0004,
+      "step": 15345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3278436644740002e-05,
+      "loss": 0.0004,
+      "step": 15346
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3276035914918136e-05,
+      "loss": 0.0004,
+      "step": 15347
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3273635185096269e-05,
+      "loss": 0.0002,
+      "step": 15348
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3271234455274402e-05,
+      "loss": 0.0005,
+      "step": 15349
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3268833725452538e-05,
+      "loss": 0.0015,
+      "step": 15350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3266432995630673e-05,
+      "loss": 0.0005,
+      "step": 15351
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3264032265808806e-05,
+      "loss": 0.0003,
+      "step": 15352
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.326163153598694e-05,
+      "loss": 0.0003,
+      "step": 15353
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3259230806165073e-05,
+      "loss": 0.0015,
+      "step": 15354
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.325683007634321e-05,
+      "loss": 0.0013,
+      "step": 15355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3254429346521344e-05,
+      "loss": 0.0004,
+      "step": 15356
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3252028616699477e-05,
+      "loss": 0.0003,
+      "step": 15357
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.324962788687761e-05,
+      "loss": 0.0002,
+      "step": 15358
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3247227157055744e-05,
+      "loss": 0.0004,
+      "step": 15359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3244826427233881e-05,
+      "loss": 0.0003,
+      "step": 15360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3242425697412015e-05,
+      "loss": 0.0009,
+      "step": 15361
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3240024967590148e-05,
+      "loss": 0.0004,
+      "step": 15362
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3237624237768282e-05,
+      "loss": 0.0017,
+      "step": 15363
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3235223507946418e-05,
+      "loss": 0.0003,
+      "step": 15364
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3232822778124552e-05,
+      "loss": 0.0012,
+      "step": 15365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3230422048302685e-05,
+      "loss": 0.0004,
+      "step": 15366
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3228021318480819e-05,
+      "loss": 0.0003,
+      "step": 15367
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3225620588658952e-05,
+      "loss": 0.0004,
+      "step": 15368
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3223219858837088e-05,
+      "loss": 0.0012,
+      "step": 15369
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3220819129015221e-05,
+      "loss": 0.0005,
+      "step": 15370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3218418399193355e-05,
+      "loss": 0.0004,
+      "step": 15371
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3216017669371488e-05,
+      "loss": 0.0004,
+      "step": 15372
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3213616939549625e-05,
+      "loss": 0.0003,
+      "step": 15373
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3211216209727758e-05,
+      "loss": 0.0003,
+      "step": 15374
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3208815479905892e-05,
+      "loss": 0.0007,
+      "step": 15375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3206414750084025e-05,
+      "loss": 0.9163,
+      "step": 15376
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3204014020262159e-05,
+      "loss": 0.0004,
+      "step": 15377
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3201613290440296e-05,
+      "loss": 0.0003,
+      "step": 15378
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.319921256061843e-05,
+      "loss": 0.0011,
+      "step": 15379
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3196811830796563e-05,
+      "loss": 0.0005,
+      "step": 15380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3194411100974696e-05,
+      "loss": 0.0004,
+      "step": 15381
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.319201037115283e-05,
+      "loss": 0.0009,
+      "step": 15382
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3189609641330967e-05,
+      "loss": 0.0004,
+      "step": 15383
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.31872089115091e-05,
+      "loss": 0.9969,
+      "step": 15384
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3184808181687234e-05,
+      "loss": 0.0003,
+      "step": 15385
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3182407451865367e-05,
+      "loss": 0.0004,
+      "step": 15386
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3180006722043502e-05,
+      "loss": 0.0013,
+      "step": 15387
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3177605992221636e-05,
+      "loss": 0.0004,
+      "step": 15388
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.317520526239977e-05,
+      "loss": 0.0004,
+      "step": 15389
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3172804532577903e-05,
+      "loss": 0.0003,
+      "step": 15390
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3170403802756038e-05,
+      "loss": 0.0005,
+      "step": 15391
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3168003072934173e-05,
+      "loss": 0.001,
+      "step": 15392
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3165602343112307e-05,
+      "loss": 0.0002,
+      "step": 15393
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.316320161329044e-05,
+      "loss": 0.0005,
+      "step": 15394
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3160800883468574e-05,
+      "loss": 0.0004,
+      "step": 15395
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.315840015364671e-05,
+      "loss": 0.0003,
+      "step": 15396
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3155999423824844e-05,
+      "loss": 0.0003,
+      "step": 15397
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3153598694002978e-05,
+      "loss": 0.0002,
+      "step": 15398
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3151197964181111e-05,
+      "loss": 0.0003,
+      "step": 15399
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3148797234359245e-05,
+      "loss": 0.0008,
+      "step": 15400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3146396504537381e-05,
+      "loss": 0.0009,
+      "step": 15401
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3143995774715515e-05,
+      "loss": 0.0004,
+      "step": 15402
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3141595044893648e-05,
+      "loss": 0.0002,
+      "step": 15403
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3139194315071782e-05,
+      "loss": 0.0004,
+      "step": 15404
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3136793585249917e-05,
+      "loss": 0.0005,
+      "step": 15405
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3134392855428052e-05,
+      "loss": 0.0004,
+      "step": 15406
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3131992125606186e-05,
+      "loss": 0.0004,
+      "step": 15407
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.312959139578432e-05,
+      "loss": 0.0003,
+      "step": 15408
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3127190665962453e-05,
+      "loss": 0.0004,
+      "step": 15409
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3124789936140588e-05,
+      "loss": 0.0017,
+      "step": 15410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3122389206318722e-05,
+      "loss": 0.0003,
+      "step": 15411
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3119988476496855e-05,
+      "loss": 0.0003,
+      "step": 15412
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3117587746674988e-05,
+      "loss": 0.0005,
+      "step": 15413
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3115187016853122e-05,
+      "loss": 0.0007,
+      "step": 15414
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3112786287031259e-05,
+      "loss": 0.0007,
+      "step": 15415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3110385557209392e-05,
+      "loss": 0.0009,
+      "step": 15416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3107984827387526e-05,
+      "loss": 0.0003,
+      "step": 15417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.310558409756566e-05,
+      "loss": 0.0003,
+      "step": 15418
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3103183367743796e-05,
+      "loss": 0.0006,
+      "step": 15419
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.310078263792193e-05,
+      "loss": 0.0003,
+      "step": 15420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3098381908100063e-05,
+      "loss": 0.0003,
+      "step": 15421
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3095981178278197e-05,
+      "loss": 0.0004,
+      "step": 15422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.309358044845633e-05,
+      "loss": 0.0003,
+      "step": 15423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3091179718634467e-05,
+      "loss": 0.0003,
+      "step": 15424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.30887789888126e-05,
+      "loss": 0.0005,
+      "step": 15425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3086378258990734e-05,
+      "loss": 0.0003,
+      "step": 15426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3083977529168868e-05,
+      "loss": 0.0013,
+      "step": 15427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3081576799347003e-05,
+      "loss": 0.0004,
+      "step": 15428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3079176069525136e-05,
+      "loss": 0.0004,
+      "step": 15429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.307677533970327e-05,
+      "loss": 0.0004,
+      "step": 15430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3074374609881403e-05,
+      "loss": 0.0005,
+      "step": 15431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3071973880059538e-05,
+      "loss": 0.0002,
+      "step": 15432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3069573150237674e-05,
+      "loss": 0.0009,
+      "step": 15433
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3067172420415807e-05,
+      "loss": 0.0004,
+      "step": 15434
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.306477169059394e-05,
+      "loss": 0.0004,
+      "step": 15435
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3062370960772074e-05,
+      "loss": 0.0003,
+      "step": 15436
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3059970230950208e-05,
+      "loss": 0.0008,
+      "step": 15437
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3057569501128344e-05,
+      "loss": 0.0003,
+      "step": 15438
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3055168771306478e-05,
+      "loss": 0.0003,
+      "step": 15439
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3052768041484611e-05,
+      "loss": 0.0004,
+      "step": 15440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3050367311662745e-05,
+      "loss": 0.0002,
+      "step": 15441
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3047966581840882e-05,
+      "loss": 0.0009,
+      "step": 15442
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3045565852019015e-05,
+      "loss": 0.0003,
+      "step": 15443
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3043165122197149e-05,
+      "loss": 0.0003,
+      "step": 15444
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3040764392375282e-05,
+      "loss": 0.0003,
+      "step": 15445
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3038363662553416e-05,
+      "loss": 0.0004,
+      "step": 15446
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3035962932731553e-05,
+      "loss": 0.0003,
+      "step": 15447
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3033562202909686e-05,
+      "loss": 0.0003,
+      "step": 15448
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.303116147308782e-05,
+      "loss": 0.0013,
+      "step": 15449
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3028760743265953e-05,
+      "loss": 0.0001,
+      "step": 15450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3026360013444088e-05,
+      "loss": 0.0003,
+      "step": 15451
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3023959283622222e-05,
+      "loss": 0.0004,
+      "step": 15452
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3021558553800355e-05,
+      "loss": 0.0006,
+      "step": 15453
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3019157823978489e-05,
+      "loss": 0.0004,
+      "step": 15454
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3016757094156622e-05,
+      "loss": 0.0008,
+      "step": 15455
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.301435636433476e-05,
+      "loss": 0.0003,
+      "step": 15456
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3011955634512893e-05,
+      "loss": 0.0003,
+      "step": 15457
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3009554904691026e-05,
+      "loss": 0.001,
+      "step": 15458
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.300715417486916e-05,
+      "loss": 0.0004,
+      "step": 15459
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3004753445047297e-05,
+      "loss": 0.0007,
+      "step": 15460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.300235271522543e-05,
+      "loss": 0.0004,
+      "step": 15461
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2999951985403564e-05,
+      "loss": 0.0005,
+      "step": 15462
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2997551255581697e-05,
+      "loss": 0.0003,
+      "step": 15463
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.299515052575983e-05,
+      "loss": 0.0002,
+      "step": 15464
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2992749795937967e-05,
+      "loss": 0.0004,
+      "step": 15465
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2990349066116101e-05,
+      "loss": 0.0003,
+      "step": 15466
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2987948336294234e-05,
+      "loss": 0.0005,
+      "step": 15467
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2985547606472368e-05,
+      "loss": 0.0004,
+      "step": 15468
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2983146876650501e-05,
+      "loss": 0.0003,
+      "step": 15469
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2980746146828637e-05,
+      "loss": 0.0002,
+      "step": 15470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.297834541700677e-05,
+      "loss": 0.0003,
+      "step": 15471
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2975944687184904e-05,
+      "loss": 1.8892,
+      "step": 15472
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2973543957363037e-05,
+      "loss": 0.0006,
+      "step": 15473
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2971143227541174e-05,
+      "loss": 0.0007,
+      "step": 15474
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2968742497719307e-05,
+      "loss": 0.0007,
+      "step": 15475
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2966341767897441e-05,
+      "loss": 0.001,
+      "step": 15476
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2963941038075574e-05,
+      "loss": 0.0003,
+      "step": 15477
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2961540308253708e-05,
+      "loss": 0.0004,
+      "step": 15478
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2959139578431845e-05,
+      "loss": 0.0008,
+      "step": 15479
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2956738848609978e-05,
+      "loss": 0.0009,
+      "step": 15480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2954338118788112e-05,
+      "loss": 0.0005,
+      "step": 15481
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2951937388966245e-05,
+      "loss": 0.0003,
+      "step": 15482
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2949536659144382e-05,
+      "loss": 0.001,
+      "step": 15483
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2947135929322516e-05,
+      "loss": 0.0007,
+      "step": 15484
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.294473519950065e-05,
+      "loss": 0.0003,
+      "step": 15485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2942334469678783e-05,
+      "loss": 0.0002,
+      "step": 15486
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2939933739856916e-05,
+      "loss": 0.0004,
+      "step": 15487
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2937533010035053e-05,
+      "loss": 0.0008,
+      "step": 15488
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2935132280213187e-05,
+      "loss": 0.0011,
+      "step": 15489
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.293273155039132e-05,
+      "loss": 0.0004,
+      "step": 15490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2930330820569454e-05,
+      "loss": 0.0009,
+      "step": 15491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2927930090747589e-05,
+      "loss": 0.0003,
+      "step": 15492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2925529360925722e-05,
+      "loss": 0.0004,
+      "step": 15493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2923128631103856e-05,
+      "loss": 0.0004,
+      "step": 15494
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.292072790128199e-05,
+      "loss": 0.0007,
+      "step": 15495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2918327171460123e-05,
+      "loss": 0.0008,
+      "step": 15496
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.291592644163826e-05,
+      "loss": 0.0004,
+      "step": 15497
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2913525711816393e-05,
+      "loss": 0.0003,
+      "step": 15498
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2911124981994527e-05,
+      "loss": 0.0004,
+      "step": 15499
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.290872425217266e-05,
+      "loss": 0.0006,
+      "step": 15500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2906323522350794e-05,
+      "loss": 0.0003,
+      "step": 15501
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.290392279252893e-05,
+      "loss": 0.0006,
+      "step": 15502
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2901522062707064e-05,
+      "loss": 0.0006,
+      "step": 15503
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2899121332885197e-05,
+      "loss": 0.0003,
+      "step": 15504
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2896720603063331e-05,
+      "loss": 0.0004,
+      "step": 15505
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2894319873241468e-05,
+      "loss": 0.0004,
+      "step": 15506
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2891919143419601e-05,
+      "loss": 0.0005,
+      "step": 15507
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2889518413597735e-05,
+      "loss": 0.0002,
+      "step": 15508
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2887117683775868e-05,
+      "loss": 0.0003,
+      "step": 15509
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2884716953954002e-05,
+      "loss": 0.0003,
+      "step": 15510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2882316224132137e-05,
+      "loss": 0.0003,
+      "step": 15511
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.287991549431027e-05,
+      "loss": 0.0004,
+      "step": 15512
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2877514764488404e-05,
+      "loss": 0.0004,
+      "step": 15513
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2875114034666537e-05,
+      "loss": 0.0003,
+      "step": 15514
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2872713304844674e-05,
+      "loss": 0.0003,
+      "step": 15515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2870312575022808e-05,
+      "loss": 0.0003,
+      "step": 15516
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2867911845200941e-05,
+      "loss": 0.0003,
+      "step": 15517
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2865511115379075e-05,
+      "loss": 0.0003,
+      "step": 15518
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2863110385557208e-05,
+      "loss": 0.0008,
+      "step": 15519
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2860709655735345e-05,
+      "loss": 0.0003,
+      "step": 15520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2858308925913479e-05,
+      "loss": 0.0004,
+      "step": 15521
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2855908196091612e-05,
+      "loss": 0.0003,
+      "step": 15522
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2853507466269746e-05,
+      "loss": 0.0007,
+      "step": 15523
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2851106736447883e-05,
+      "loss": 0.0003,
+      "step": 15524
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2848706006626016e-05,
+      "loss": 0.0003,
+      "step": 15525
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.284630527680415e-05,
+      "loss": 0.0003,
+      "step": 15526
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2843904546982283e-05,
+      "loss": 0.0009,
+      "step": 15527
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2841503817160417e-05,
+      "loss": 0.0003,
+      "step": 15528
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2839103087338552e-05,
+      "loss": 0.0012,
+      "step": 15529
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2836702357516687e-05,
+      "loss": 0.0003,
+      "step": 15530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.283430162769482e-05,
+      "loss": 0.0004,
+      "step": 15531
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2831900897872954e-05,
+      "loss": 0.0006,
+      "step": 15532
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2829500168051087e-05,
+      "loss": 0.0003,
+      "step": 15533
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2827099438229223e-05,
+      "loss": 0.0004,
+      "step": 15534
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2824698708407356e-05,
+      "loss": 0.0002,
+      "step": 15535
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.282229797858549e-05,
+      "loss": 0.0004,
+      "step": 15536
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2819897248763623e-05,
+      "loss": 0.0006,
+      "step": 15537
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.281749651894176e-05,
+      "loss": 0.0004,
+      "step": 15538
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2815095789119893e-05,
+      "loss": 0.0002,
+      "step": 15539
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2812695059298027e-05,
+      "loss": 0.0004,
+      "step": 15540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.281029432947616e-05,
+      "loss": 0.0002,
+      "step": 15541
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2807893599654294e-05,
+      "loss": 0.0004,
+      "step": 15542
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2805492869832431e-05,
+      "loss": 0.0003,
+      "step": 15543
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2803092140010564e-05,
+      "loss": 0.0003,
+      "step": 15544
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2800691410188698e-05,
+      "loss": 0.0005,
+      "step": 15545
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2798290680366831e-05,
+      "loss": 0.0006,
+      "step": 15546
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2795889950544968e-05,
+      "loss": 0.0007,
+      "step": 15547
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2793489220723102e-05,
+      "loss": 0.0007,
+      "step": 15548
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2791088490901235e-05,
+      "loss": 0.0009,
+      "step": 15549
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2788687761079369e-05,
+      "loss": 0.0002,
+      "step": 15550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2786287031257502e-05,
+      "loss": 0.0002,
+      "step": 15551
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2783886301435637e-05,
+      "loss": 0.0006,
+      "step": 15552
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2781485571613771e-05,
+      "loss": 0.0003,
+      "step": 15553
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2779084841791904e-05,
+      "loss": 0.0002,
+      "step": 15554
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2776684111970038e-05,
+      "loss": 0.0006,
+      "step": 15555
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2774283382148175e-05,
+      "loss": 0.0007,
+      "step": 15556
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2771882652326308e-05,
+      "loss": 0.0004,
+      "step": 15557
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2769481922504442e-05,
+      "loss": 0.0004,
+      "step": 15558
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2767081192682575e-05,
+      "loss": 0.0001,
+      "step": 15559
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2764680462860709e-05,
+      "loss": 0.0004,
+      "step": 15560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2762279733038846e-05,
+      "loss": 0.0003,
+      "step": 15561
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2759879003216979e-05,
+      "loss": 0.0007,
+      "step": 15562
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2757478273395113e-05,
+      "loss": 0.4724,
+      "step": 15563
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2755077543573246e-05,
+      "loss": 0.0004,
+      "step": 15564
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.275267681375138e-05,
+      "loss": 0.001,
+      "step": 15565
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2750276083929516e-05,
+      "loss": 0.0002,
+      "step": 15566
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.274787535410765e-05,
+      "loss": 0.0004,
+      "step": 15567
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2745474624285783e-05,
+      "loss": 0.0003,
+      "step": 15568
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2743073894463917e-05,
+      "loss": 0.0002,
+      "step": 15569
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2740673164642052e-05,
+      "loss": 0.0004,
+      "step": 15570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2738272434820187e-05,
+      "loss": 0.0006,
+      "step": 15571
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.273587170499832e-05,
+      "loss": 0.0003,
+      "step": 15572
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2733470975176454e-05,
+      "loss": 0.0003,
+      "step": 15573
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2731070245354588e-05,
+      "loss": 0.0007,
+      "step": 15574
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2728669515532723e-05,
+      "loss": 0.0009,
+      "step": 15575
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2726268785710856e-05,
+      "loss": 0.0008,
+      "step": 15576
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.272386805588899e-05,
+      "loss": 0.0003,
+      "step": 15577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2721467326067123e-05,
+      "loss": 0.0003,
+      "step": 15578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.271906659624526e-05,
+      "loss": 0.0002,
+      "step": 15579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2716665866423394e-05,
+      "loss": 1.1448,
+      "step": 15580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2714265136601527e-05,
+      "loss": 0.0003,
+      "step": 15581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2711864406779661e-05,
+      "loss": 0.0003,
+      "step": 15582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2709463676957794e-05,
+      "loss": 0.0007,
+      "step": 15583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2707062947135931e-05,
+      "loss": 0.0002,
+      "step": 15584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2704662217314065e-05,
+      "loss": 0.0005,
+      "step": 15585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2702261487492198e-05,
+      "loss": 0.0003,
+      "step": 15586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2699860757670332e-05,
+      "loss": 0.0004,
+      "step": 15587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2697460027848469e-05,
+      "loss": 0.0005,
+      "step": 15588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2695059298026602e-05,
+      "loss": 0.0005,
+      "step": 15589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2692658568204736e-05,
+      "loss": 0.0005,
+      "step": 15590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2690257838382869e-05,
+      "loss": 0.0003,
+      "step": 15591
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2687857108561003e-05,
+      "loss": 0.0003,
+      "step": 15592
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2685456378739138e-05,
+      "loss": 0.0003,
+      "step": 15593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2683055648917271e-05,
+      "loss": 0.0006,
+      "step": 15594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2680654919095405e-05,
+      "loss": 0.0004,
+      "step": 15595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2678254189273538e-05,
+      "loss": 0.0002,
+      "step": 15596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2675853459451672e-05,
+      "loss": 0.0007,
+      "step": 15597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2673452729629809e-05,
+      "loss": 0.0007,
+      "step": 15598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2671051999807942e-05,
+      "loss": 0.0003,
+      "step": 15599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2668651269986076e-05,
+      "loss": 0.7539,
+      "step": 15600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2666250540164209e-05,
+      "loss": 0.3463,
+      "step": 15601
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2663849810342346e-05,
+      "loss": 0.0002,
+      "step": 15602
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.266144908052048e-05,
+      "loss": 1.4704,
+      "step": 15603
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2659048350698613e-05,
+      "loss": 0.0004,
+      "step": 15604
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2656647620876746e-05,
+      "loss": 0.0003,
+      "step": 15605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.265424689105488e-05,
+      "loss": 0.0003,
+      "step": 15606
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2651846161233017e-05,
+      "loss": 0.0003,
+      "step": 15607
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.264944543141115e-05,
+      "loss": 0.0002,
+      "step": 15608
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2647044701589284e-05,
+      "loss": 0.0004,
+      "step": 15609
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2644643971767417e-05,
+      "loss": 0.0008,
+      "step": 15610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2642243241945553e-05,
+      "loss": 0.0005,
+      "step": 15611
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2639842512123688e-05,
+      "loss": 0.0003,
+      "step": 15612
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2637441782301821e-05,
+      "loss": 0.0003,
+      "step": 15613
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2635041052479955e-05,
+      "loss": 0.0003,
+      "step": 15614
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2632640322658088e-05,
+      "loss": 0.0004,
+      "step": 15615
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2630239592836223e-05,
+      "loss": 0.0004,
+      "step": 15616
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2627838863014357e-05,
+      "loss": 0.0004,
+      "step": 15617
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.262543813319249e-05,
+      "loss": 0.0004,
+      "step": 15618
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2623037403370624e-05,
+      "loss": 0.0004,
+      "step": 15619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.262063667354876e-05,
+      "loss": 0.0004,
+      "step": 15620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2618235943726894e-05,
+      "loss": 0.0004,
+      "step": 15621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2615835213905028e-05,
+      "loss": 0.0015,
+      "step": 15622
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2613434484083161e-05,
+      "loss": 0.0004,
+      "step": 15623
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2611033754261295e-05,
+      "loss": 0.0288,
+      "step": 15624
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2608633024439432e-05,
+      "loss": 0.0003,
+      "step": 15625
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2606232294617565e-05,
+      "loss": 0.0003,
+      "step": 15626
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2603831564795699e-05,
+      "loss": 0.652,
+      "step": 15627
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2601430834973832e-05,
+      "loss": 0.0003,
+      "step": 15628
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2599030105151966e-05,
+      "loss": 0.0004,
+      "step": 15629
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2596629375330102e-05,
+      "loss": 0.0004,
+      "step": 15630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2594228645508236e-05,
+      "loss": 0.0022,
+      "step": 15631
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.259182791568637e-05,
+      "loss": 0.0004,
+      "step": 15632
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2589427185864503e-05,
+      "loss": 0.0013,
+      "step": 15633
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2587026456042638e-05,
+      "loss": 0.0003,
+      "step": 15634
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2584625726220772e-05,
+      "loss": 0.0014,
+      "step": 15635
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2582224996398905e-05,
+      "loss": 0.0004,
+      "step": 15636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2579824266577039e-05,
+      "loss": 0.0003,
+      "step": 15637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2577423536755172e-05,
+      "loss": 0.0013,
+      "step": 15638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2575022806933309e-05,
+      "loss": 0.0014,
+      "step": 15639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2572622077111442e-05,
+      "loss": 0.002,
+      "step": 15640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2570221347289576e-05,
+      "loss": 0.0004,
+      "step": 15641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.256782061746771e-05,
+      "loss": 0.0004,
+      "step": 15642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2565419887645846e-05,
+      "loss": 0.0006,
+      "step": 15643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.256301915782398e-05,
+      "loss": 0.0019,
+      "step": 15644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2560618428002113e-05,
+      "loss": 0.0003,
+      "step": 15645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2558217698180247e-05,
+      "loss": 0.0003,
+      "step": 15646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.255581696835838e-05,
+      "loss": 0.0013,
+      "step": 15647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2553416238536517e-05,
+      "loss": 0.0003,
+      "step": 15648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.255101550871465e-05,
+      "loss": 0.0004,
+      "step": 15649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2548614778892784e-05,
+      "loss": 0.0004,
+      "step": 15650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2546214049070918e-05,
+      "loss": 0.0004,
+      "step": 15651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2543813319249053e-05,
+      "loss": 0.0003,
+      "step": 15652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2541412589427186e-05,
+      "loss": 0.0019,
+      "step": 15653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2539011859605322e-05,
+      "loss": 0.0014,
+      "step": 15654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2536611129783455e-05,
+      "loss": 0.0003,
+      "step": 15655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2534210399961589e-05,
+      "loss": 0.0003,
+      "step": 15656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2531809670139724e-05,
+      "loss": 0.0013,
+      "step": 15657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2529408940317857e-05,
+      "loss": 0.0003,
+      "step": 15658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.252700821049599e-05,
+      "loss": 0.0026,
+      "step": 15659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2524607480674124e-05,
+      "loss": 0.0003,
+      "step": 15660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2522206750852258e-05,
+      "loss": 0.0003,
+      "step": 15661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2519806021030395e-05,
+      "loss": 0.0004,
+      "step": 15662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2517405291208528e-05,
+      "loss": 0.0004,
+      "step": 15663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2515004561386662e-05,
+      "loss": 0.0023,
+      "step": 15664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2512603831564795e-05,
+      "loss": 0.0009,
+      "step": 15665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2510203101742932e-05,
+      "loss": 0.0003,
+      "step": 15666
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2507802371921065e-05,
+      "loss": 0.0011,
+      "step": 15667
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2505401642099199e-05,
+      "loss": 0.0009,
+      "step": 15668
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2503000912277332e-05,
+      "loss": 0.0003,
+      "step": 15669
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2500600182455466e-05,
+      "loss": 0.0003,
+      "step": 15670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2498199452633601e-05,
+      "loss": 0.0004,
+      "step": 15671
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2495798722811736e-05,
+      "loss": 0.0003,
+      "step": 15672
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.249339799298987e-05,
+      "loss": 0.0004,
+      "step": 15673
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2490997263168005e-05,
+      "loss": 0.0015,
+      "step": 15674
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2488596533346139e-05,
+      "loss": 0.0005,
+      "step": 15675
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2486195803524272e-05,
+      "loss": 0.0003,
+      "step": 15676
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2483795073702405e-05,
+      "loss": 0.0003,
+      "step": 15677
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2481394343880539e-05,
+      "loss": 0.0002,
+      "step": 15678
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2478993614058674e-05,
+      "loss": 0.0025,
+      "step": 15679
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2476592884236808e-05,
+      "loss": 0.0012,
+      "step": 15680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2474192154414943e-05,
+      "loss": 0.0003,
+      "step": 15681
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2471791424593076e-05,
+      "loss": 0.0003,
+      "step": 15682
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2469390694771212e-05,
+      "loss": 0.0004,
+      "step": 15683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2466989964949345e-05,
+      "loss": 0.0003,
+      "step": 15684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2464589235127479e-05,
+      "loss": 0.0017,
+      "step": 15685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2462188505305614e-05,
+      "loss": 0.0003,
+      "step": 15686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2459787775483747e-05,
+      "loss": 0.0004,
+      "step": 15687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2457387045661882e-05,
+      "loss": 0.0003,
+      "step": 15688
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2454986315840016e-05,
+      "loss": 0.0003,
+      "step": 15689
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2452585586018151e-05,
+      "loss": 0.9451,
+      "step": 15690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2450184856196285e-05,
+      "loss": 0.0005,
+      "step": 15691
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.244778412637442e-05,
+      "loss": 0.0004,
+      "step": 15692
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2445383396552553e-05,
+      "loss": 0.0003,
+      "step": 15693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2442982666730687e-05,
+      "loss": 0.0003,
+      "step": 15694
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2440581936908822e-05,
+      "loss": 0.0003,
+      "step": 15695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2438181207086955e-05,
+      "loss": 0.0003,
+      "step": 15696
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2435780477265089e-05,
+      "loss": 0.0003,
+      "step": 15697
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2433379747443222e-05,
+      "loss": 0.0004,
+      "step": 15698
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2430979017621358e-05,
+      "loss": 0.0012,
+      "step": 15699
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2428578287799491e-05,
+      "loss": 0.0017,
+      "step": 15700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2426177557977625e-05,
+      "loss": 0.0003,
+      "step": 15701
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.242377682815576e-05,
+      "loss": 0.0004,
+      "step": 15702
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2421376098333893e-05,
+      "loss": 0.0004,
+      "step": 15703
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2418975368512028e-05,
+      "loss": 0.9172,
+      "step": 15704
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2416574638690162e-05,
+      "loss": 0.0004,
+      "step": 15705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2414173908868297e-05,
+      "loss": 0.0003,
+      "step": 15706
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.241177317904643e-05,
+      "loss": 0.0004,
+      "step": 15707
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2409372449224566e-05,
+      "loss": 0.0004,
+      "step": 15708
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.24069717194027e-05,
+      "loss": 0.0003,
+      "step": 15709
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2404570989580833e-05,
+      "loss": 0.0121,
+      "step": 15710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2402170259758968e-05,
+      "loss": 0.0003,
+      "step": 15711
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2399769529937102e-05,
+      "loss": 0.0006,
+      "step": 15712
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2397368800115237e-05,
+      "loss": 0.0009,
+      "step": 15713
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.239496807029337e-05,
+      "loss": 0.0012,
+      "step": 15714
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2392567340471504e-05,
+      "loss": 0.0011,
+      "step": 15715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2390166610649639e-05,
+      "loss": 0.0004,
+      "step": 15716
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2387765880827772e-05,
+      "loss": 0.0004,
+      "step": 15717
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2385365151005906e-05,
+      "loss": 0.0005,
+      "step": 15718
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.238296442118404e-05,
+      "loss": 0.0004,
+      "step": 15719
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2380563691362175e-05,
+      "loss": 0.0004,
+      "step": 15720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2378162961540308e-05,
+      "loss": 0.0004,
+      "step": 15721
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2375762231718443e-05,
+      "loss": 0.0004,
+      "step": 15722
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2373361501896577e-05,
+      "loss": 0.0003,
+      "step": 15723
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2370960772074712e-05,
+      "loss": 0.0006,
+      "step": 15724
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2368560042252845e-05,
+      "loss": 0.0003,
+      "step": 15725
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2366159312430979e-05,
+      "loss": 0.0004,
+      "step": 15726
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2363758582609114e-05,
+      "loss": 0.0011,
+      "step": 15727
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2361357852787248e-05,
+      "loss": 0.9579,
+      "step": 15728
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2358957122965383e-05,
+      "loss": 0.0018,
+      "step": 15729
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2356556393143516e-05,
+      "loss": 0.0003,
+      "step": 15730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2354155663321651e-05,
+      "loss": 0.0004,
+      "step": 15731
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2351754933499785e-05,
+      "loss": 0.0009,
+      "step": 15732
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2349354203677918e-05,
+      "loss": 0.0003,
+      "step": 15733
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2346953473856054e-05,
+      "loss": 0.0005,
+      "step": 15734
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2344552744034187e-05,
+      "loss": 0.0004,
+      "step": 15735
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2342152014212322e-05,
+      "loss": 0.0005,
+      "step": 15736
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2339751284390456e-05,
+      "loss": 0.0004,
+      "step": 15737
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.233735055456859e-05,
+      "loss": 0.0003,
+      "step": 15738
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2334949824746723e-05,
+      "loss": 0.9156,
+      "step": 15739
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2332549094924858e-05,
+      "loss": 0.0004,
+      "step": 15740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2330148365102991e-05,
+      "loss": 0.0009,
+      "step": 15741
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2327747635281125e-05,
+      "loss": 0.0003,
+      "step": 15742
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.232534690545926e-05,
+      "loss": 0.0017,
+      "step": 15743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2322946175637394e-05,
+      "loss": 0.0021,
+      "step": 15744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2320545445815529e-05,
+      "loss": 0.0004,
+      "step": 15745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2318144715993662e-05,
+      "loss": 0.0003,
+      "step": 15746
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2315743986171798e-05,
+      "loss": 0.0005,
+      "step": 15747
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2313343256349931e-05,
+      "loss": 0.0005,
+      "step": 15748
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2310942526528065e-05,
+      "loss": 0.0004,
+      "step": 15749
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.23085417967062e-05,
+      "loss": 0.0003,
+      "step": 15750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2306141066884333e-05,
+      "loss": 0.4099,
+      "step": 15751
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2303740337062468e-05,
+      "loss": 0.0015,
+      "step": 15752
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2301339607240602e-05,
+      "loss": 0.001,
+      "step": 15753
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2298938877418737e-05,
+      "loss": 0.0003,
+      "step": 15754
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.229653814759687e-05,
+      "loss": 0.0007,
+      "step": 15755
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2294137417775004e-05,
+      "loss": 0.0004,
+      "step": 15756
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.229173668795314e-05,
+      "loss": 0.001,
+      "step": 15757
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2289335958131273e-05,
+      "loss": 0.0003,
+      "step": 15758
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2286935228309406e-05,
+      "loss": 0.0009,
+      "step": 15759
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.228453449848754e-05,
+      "loss": 0.0005,
+      "step": 15760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2282133768665675e-05,
+      "loss": 0.0006,
+      "step": 15761
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2279733038843808e-05,
+      "loss": 0.0006,
+      "step": 15762
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2277332309021944e-05,
+      "loss": 0.0009,
+      "step": 15763
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2274931579200077e-05,
+      "loss": 0.0004,
+      "step": 15764
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.227253084937821e-05,
+      "loss": 0.0004,
+      "step": 15765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2270130119556346e-05,
+      "loss": 0.0014,
+      "step": 15766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.226772938973448e-05,
+      "loss": 0.0003,
+      "step": 15767
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2265328659912614e-05,
+      "loss": 0.0009,
+      "step": 15768
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2262927930090748e-05,
+      "loss": 0.0006,
+      "step": 15769
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2260527200268883e-05,
+      "loss": 0.0005,
+      "step": 15770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2258126470447017e-05,
+      "loss": 0.0009,
+      "step": 15771
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.225572574062515e-05,
+      "loss": 0.0008,
+      "step": 15772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2253325010803285e-05,
+      "loss": 0.0005,
+      "step": 15773
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2250924280981419e-05,
+      "loss": 0.0004,
+      "step": 15774
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2248523551159554e-05,
+      "loss": 0.0004,
+      "step": 15775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2246122821337688e-05,
+      "loss": 0.0017,
+      "step": 15776
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2243722091515821e-05,
+      "loss": 0.0006,
+      "step": 15777
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2241321361693956e-05,
+      "loss": 0.001,
+      "step": 15778
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.223892063187209e-05,
+      "loss": 0.0003,
+      "step": 15779
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2236519902050223e-05,
+      "loss": 0.0003,
+      "step": 15780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2234119172228357e-05,
+      "loss": 0.0004,
+      "step": 15781
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2231718442406492e-05,
+      "loss": 0.0008,
+      "step": 15782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2229317712584625e-05,
+      "loss": 0.0004,
+      "step": 15783
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.222691698276276e-05,
+      "loss": 0.0004,
+      "step": 15784
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2224516252940894e-05,
+      "loss": 0.0015,
+      "step": 15785
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.222211552311903e-05,
+      "loss": 0.0013,
+      "step": 15786
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2219714793297163e-05,
+      "loss": 0.0004,
+      "step": 15787
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2217314063475296e-05,
+      "loss": 0.0004,
+      "step": 15788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2214913333653431e-05,
+      "loss": 0.0011,
+      "step": 15789
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2212512603831565e-05,
+      "loss": 0.0066,
+      "step": 15790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.22101118740097e-05,
+      "loss": 0.0003,
+      "step": 15791
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2207711144187834e-05,
+      "loss": 0.0011,
+      "step": 15792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2205310414365969e-05,
+      "loss": 0.0004,
+      "step": 15793
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2202909684544102e-05,
+      "loss": 0.0008,
+      "step": 15794
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2200508954722237e-05,
+      "loss": 0.8595,
+      "step": 15795
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2198108224900371e-05,
+      "loss": 0.0005,
+      "step": 15796
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2195707495078504e-05,
+      "loss": 0.0009,
+      "step": 15797
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.219330676525664e-05,
+      "loss": 0.0004,
+      "step": 15798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2190906035434773e-05,
+      "loss": 0.0003,
+      "step": 15799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2188505305612907e-05,
+      "loss": 0.0004,
+      "step": 15800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.218610457579104e-05,
+      "loss": 0.0004,
+      "step": 15801
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2183703845969175e-05,
+      "loss": 0.0004,
+      "step": 15802
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2181303116147309e-05,
+      "loss": 0.0009,
+      "step": 15803
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2178902386325442e-05,
+      "loss": 0.0004,
+      "step": 15804
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2176501656503577e-05,
+      "loss": 0.0011,
+      "step": 15805
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2174100926681711e-05,
+      "loss": 0.001,
+      "step": 15806
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2171700196859846e-05,
+      "loss": 0.0008,
+      "step": 15807
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.216929946703798e-05,
+      "loss": 0.0008,
+      "step": 15808
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2166898737216115e-05,
+      "loss": 0.0002,
+      "step": 15809
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2164498007394248e-05,
+      "loss": 0.0004,
+      "step": 15810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2162097277572384e-05,
+      "loss": 0.0003,
+      "step": 15811
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2159696547750517e-05,
+      "loss": 0.0005,
+      "step": 15812
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.215729581792865e-05,
+      "loss": 0.0011,
+      "step": 15813
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2154895088106786e-05,
+      "loss": 0.0009,
+      "step": 15814
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.215249435828492e-05,
+      "loss": 0.0004,
+      "step": 15815
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2150093628463054e-05,
+      "loss": 0.0012,
+      "step": 15816
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2147692898641188e-05,
+      "loss": 0.0008,
+      "step": 15817
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2145292168819321e-05,
+      "loss": 0.0003,
+      "step": 15818
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2142891438997457e-05,
+      "loss": 0.0005,
+      "step": 15819
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.214049070917559e-05,
+      "loss": 0.0007,
+      "step": 15820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2138089979353724e-05,
+      "loss": 0.0005,
+      "step": 15821
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2135689249531857e-05,
+      "loss": 0.0008,
+      "step": 15822
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2133288519709992e-05,
+      "loss": 0.0002,
+      "step": 15823
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2130887789888126e-05,
+      "loss": 0.0011,
+      "step": 15824
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2128487060066261e-05,
+      "loss": 0.0007,
+      "step": 15825
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2126086330244394e-05,
+      "loss": 0.0002,
+      "step": 15826
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.212368560042253e-05,
+      "loss": 0.6712,
+      "step": 15827
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2121284870600663e-05,
+      "loss": 0.0005,
+      "step": 15828
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2118884140778797e-05,
+      "loss": 0.0009,
+      "step": 15829
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2116483410956932e-05,
+      "loss": 0.0004,
+      "step": 15830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2114082681135065e-05,
+      "loss": 0.0004,
+      "step": 15831
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.21116819513132e-05,
+      "loss": 0.001,
+      "step": 15832
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2109281221491334e-05,
+      "loss": 0.0003,
+      "step": 15833
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2106880491669469e-05,
+      "loss": 0.0003,
+      "step": 15834
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2104479761847603e-05,
+      "loss": 0.0004,
+      "step": 15835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2102079032025736e-05,
+      "loss": 0.0004,
+      "step": 15836
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2099678302203871e-05,
+      "loss": 0.0004,
+      "step": 15837
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2097277572382005e-05,
+      "loss": 0.4413,
+      "step": 15838
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2094876842560138e-05,
+      "loss": 0.0009,
+      "step": 15839
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2092476112738274e-05,
+      "loss": 0.0004,
+      "step": 15840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2090075382916407e-05,
+      "loss": 0.0003,
+      "step": 15841
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.208767465309454e-05,
+      "loss": 0.001,
+      "step": 15842
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2085273923272676e-05,
+      "loss": 0.0005,
+      "step": 15843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.208287319345081e-05,
+      "loss": 0.0002,
+      "step": 15844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2080472463628943e-05,
+      "loss": 0.0004,
+      "step": 15845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2078071733807078e-05,
+      "loss": 0.001,
+      "step": 15846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2075671003985211e-05,
+      "loss": 0.0004,
+      "step": 15847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2073270274163347e-05,
+      "loss": 0.0005,
+      "step": 15848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.207086954434148e-05,
+      "loss": 0.2327,
+      "step": 15849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2068468814519615e-05,
+      "loss": 0.0003,
+      "step": 15850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2066068084697749e-05,
+      "loss": 0.0005,
+      "step": 15851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2063667354875882e-05,
+      "loss": 0.1883,
+      "step": 15852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2061266625054017e-05,
+      "loss": 0.0004,
+      "step": 15853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2058865895232151e-05,
+      "loss": 0.0004,
+      "step": 15854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2056465165410286e-05,
+      "loss": 0.0003,
+      "step": 15855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.205406443558842e-05,
+      "loss": 0.0003,
+      "step": 15856
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2051663705766555e-05,
+      "loss": 0.0004,
+      "step": 15857
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2049262975944688e-05,
+      "loss": 0.0004,
+      "step": 15858
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2046862246122822e-05,
+      "loss": 0.6442,
+      "step": 15859
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2044461516300955e-05,
+      "loss": 0.001,
+      "step": 15860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.204206078647909e-05,
+      "loss": 0.0003,
+      "step": 15861
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2039660056657224e-05,
+      "loss": 0.0003,
+      "step": 15862
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2037259326835357e-05,
+      "loss": 0.0003,
+      "step": 15863
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2034858597013493e-05,
+      "loss": 0.0039,
+      "step": 15864
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2032457867191626e-05,
+      "loss": 0.6409,
+      "step": 15865
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2030057137369761e-05,
+      "loss": 0.0002,
+      "step": 15866
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2027656407547895e-05,
+      "loss": 0.0013,
+      "step": 15867
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2025255677726028e-05,
+      "loss": 0.0006,
+      "step": 15868
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2022854947904163e-05,
+      "loss": 0.0004,
+      "step": 15869
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2020454218082297e-05,
+      "loss": 0.6319,
+      "step": 15870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2018053488260432e-05,
+      "loss": 0.0004,
+      "step": 15871
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2015652758438566e-05,
+      "loss": 0.0003,
+      "step": 15872
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.20132520286167e-05,
+      "loss": 0.0015,
+      "step": 15873
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2010851298794834e-05,
+      "loss": 0.0018,
+      "step": 15874
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.200845056897297e-05,
+      "loss": 0.0714,
+      "step": 15875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2006049839151103e-05,
+      "loss": 0.0005,
+      "step": 15876
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2003649109329237e-05,
+      "loss": 0.0042,
+      "step": 15877
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2001248379507372e-05,
+      "loss": 0.003,
+      "step": 15878
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1998847649685505e-05,
+      "loss": 0.0004,
+      "step": 15879
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1996446919863639e-05,
+      "loss": 0.0004,
+      "step": 15880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1994046190041774e-05,
+      "loss": 0.0004,
+      "step": 15881
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1991645460219907e-05,
+      "loss": 0.0046,
+      "step": 15882
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1989244730398041e-05,
+      "loss": 0.0043,
+      "step": 15883
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1986844000576174e-05,
+      "loss": 0.0004,
+      "step": 15884
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.198444327075431e-05,
+      "loss": 0.0022,
+      "step": 15885
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1982042540932443e-05,
+      "loss": 0.0026,
+      "step": 15886
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1979641811110578e-05,
+      "loss": 0.0019,
+      "step": 15887
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1977241081288712e-05,
+      "loss": 0.0004,
+      "step": 15888
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1974840351466847e-05,
+      "loss": 0.0046,
+      "step": 15889
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.197243962164498e-05,
+      "loss": 0.0002,
+      "step": 15890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1970038891823116e-05,
+      "loss": 0.0003,
+      "step": 15891
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1967638162001249e-05,
+      "loss": 0.0004,
+      "step": 15892
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1965237432179383e-05,
+      "loss": 0.2521,
+      "step": 15893
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1962836702357518e-05,
+      "loss": 0.0004,
+      "step": 15894
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1960435972535651e-05,
+      "loss": 0.0046,
+      "step": 15895
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1958035242713786e-05,
+      "loss": 0.0003,
+      "step": 15896
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.195563451289192e-05,
+      "loss": 0.0003,
+      "step": 15897
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1953233783070055e-05,
+      "loss": 0.0003,
+      "step": 15898
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1950833053248189e-05,
+      "loss": 0.0004,
+      "step": 15899
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1948432323426322e-05,
+      "loss": 0.0018,
+      "step": 15900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1946031593604456e-05,
+      "loss": 0.0002,
+      "step": 15901
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.194363086378259e-05,
+      "loss": 0.0005,
+      "step": 15902
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1941230133960724e-05,
+      "loss": 0.0019,
+      "step": 15903
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1938829404138858e-05,
+      "loss": 0.0019,
+      "step": 15904
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1936428674316993e-05,
+      "loss": 0.0003,
+      "step": 15905
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1934027944495126e-05,
+      "loss": 0.0003,
+      "step": 15906
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1931627214673262e-05,
+      "loss": 0.0004,
+      "step": 15907
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1929226484851395e-05,
+      "loss": 0.0004,
+      "step": 15908
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1926825755029529e-05,
+      "loss": 0.0015,
+      "step": 15909
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1924425025207664e-05,
+      "loss": 0.0004,
+      "step": 15910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1922024295385797e-05,
+      "loss": 0.0004,
+      "step": 15911
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1919623565563933e-05,
+      "loss": 0.0003,
+      "step": 15912
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1917222835742066e-05,
+      "loss": 0.0004,
+      "step": 15913
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1914822105920201e-05,
+      "loss": 0.0004,
+      "step": 15914
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1912421376098335e-05,
+      "loss": 0.0003,
+      "step": 15915
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1910020646276468e-05,
+      "loss": 0.0016,
+      "step": 15916
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1907619916454603e-05,
+      "loss": 0.0003,
+      "step": 15917
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1905219186632737e-05,
+      "loss": 0.0004,
+      "step": 15918
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1902818456810872e-05,
+      "loss": 0.0004,
+      "step": 15919
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1900417726989006e-05,
+      "loss": 0.0003,
+      "step": 15920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1898016997167139e-05,
+      "loss": 0.6059,
+      "step": 15921
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1895616267345273e-05,
+      "loss": 0.0003,
+      "step": 15922
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1893215537523408e-05,
+      "loss": 0.0003,
+      "step": 15923
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1890814807701541e-05,
+      "loss": 0.0003,
+      "step": 15924
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1888414077879675e-05,
+      "loss": 0.0018,
+      "step": 15925
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.188601334805781e-05,
+      "loss": 0.0003,
+      "step": 15926
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1883612618235943e-05,
+      "loss": 0.0003,
+      "step": 15927
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1881211888414079e-05,
+      "loss": 0.0003,
+      "step": 15928
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1878811158592212e-05,
+      "loss": 0.0004,
+      "step": 15929
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1876410428770347e-05,
+      "loss": 0.0004,
+      "step": 15930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.187400969894848e-05,
+      "loss": 0.0003,
+      "step": 15931
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1871608969126614e-05,
+      "loss": 0.0003,
+      "step": 15932
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.186920823930475e-05,
+      "loss": 0.0002,
+      "step": 15933
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1866807509482883e-05,
+      "loss": 0.0002,
+      "step": 15934
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1864406779661018e-05,
+      "loss": 0.0004,
+      "step": 15935
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1862006049839152e-05,
+      "loss": 0.0003,
+      "step": 15936
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1859605320017287e-05,
+      "loss": 0.0003,
+      "step": 15937
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185720459019542e-05,
+      "loss": 0.0003,
+      "step": 15938
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854803860373556e-05,
+      "loss": 0.0003,
+      "step": 15939
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1852403130551689e-05,
+      "loss": 0.5326,
+      "step": 15940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1850002400729823e-05,
+      "loss": 0.0004,
+      "step": 15941
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1847601670907956e-05,
+      "loss": 0.0048,
+      "step": 15942
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1845200941086091e-05,
+      "loss": 0.0038,
+      "step": 15943
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1842800211264225e-05,
+      "loss": 0.002,
+      "step": 15944
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840399481442358e-05,
+      "loss": 0.0005,
+      "step": 15945
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1837998751620493e-05,
+      "loss": 0.0022,
+      "step": 15946
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1835598021798627e-05,
+      "loss": 0.0005,
+      "step": 15947
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183319729197676e-05,
+      "loss": 0.0003,
+      "step": 15948
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1830796562154896e-05,
+      "loss": 0.0004,
+      "step": 15949
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1828395832333029e-05,
+      "loss": 0.0004,
+      "step": 15950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1825995102511164e-05,
+      "loss": 0.0003,
+      "step": 15951
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1823594372689298e-05,
+      "loss": 0.0003,
+      "step": 15952
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1821193642867433e-05,
+      "loss": 0.0002,
+      "step": 15953
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1818792913045566e-05,
+      "loss": 0.0004,
+      "step": 15954
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1816392183223702e-05,
+      "loss": 0.0003,
+      "step": 15955
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1813991453401835e-05,
+      "loss": 0.0072,
+      "step": 15956
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1811590723579969e-05,
+      "loss": 0.0003,
+      "step": 15957
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1809189993758104e-05,
+      "loss": 0.0003,
+      "step": 15958
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1806789263936237e-05,
+      "loss": 0.0003,
+      "step": 15959
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1804388534114372e-05,
+      "loss": 0.0042,
+      "step": 15960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1801987804292506e-05,
+      "loss": 0.0019,
+      "step": 15961
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.179958707447064e-05,
+      "loss": 0.0002,
+      "step": 15962
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1797186344648773e-05,
+      "loss": 0.0003,
+      "step": 15963
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1794785614826908e-05,
+      "loss": 0.0002,
+      "step": 15964
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1792384885005042e-05,
+      "loss": 0.0029,
+      "step": 15965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1789984155183175e-05,
+      "loss": 0.0002,
+      "step": 15966
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.178758342536131e-05,
+      "loss": 0.0004,
+      "step": 15967
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1785182695539444e-05,
+      "loss": 0.0003,
+      "step": 15968
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1782781965717579e-05,
+      "loss": 0.0002,
+      "step": 15969
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1780381235895712e-05,
+      "loss": 0.0003,
+      "step": 15970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1777980506073848e-05,
+      "loss": 0.0002,
+      "step": 15971
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1775579776251981e-05,
+      "loss": 0.0029,
+      "step": 15972
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1773179046430115e-05,
+      "loss": 0.0024,
+      "step": 15973
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177077831660825e-05,
+      "loss": 0.0004,
+      "step": 15974
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1768377586786383e-05,
+      "loss": 0.0003,
+      "step": 15975
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1765976856964519e-05,
+      "loss": 0.0025,
+      "step": 15976
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1763576127142652e-05,
+      "loss": 0.0004,
+      "step": 15977
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1761175397320787e-05,
+      "loss": 0.0004,
+      "step": 15978
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175877466749892e-05,
+      "loss": 0.0003,
+      "step": 15979
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756373937677054e-05,
+      "loss": 0.0011,
+      "step": 15980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175397320785519e-05,
+      "loss": 0.0003,
+      "step": 15981
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1751572478033323e-05,
+      "loss": 0.0004,
+      "step": 15982
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1749171748211456e-05,
+      "loss": 0.0003,
+      "step": 15983
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.174677101838959e-05,
+      "loss": 0.0003,
+      "step": 15984
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1744370288567725e-05,
+      "loss": 0.003,
+      "step": 15985
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1741969558745859e-05,
+      "loss": 0.0004,
+      "step": 15986
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1739568828923994e-05,
+      "loss": 0.0003,
+      "step": 15987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1737168099102127e-05,
+      "loss": 0.0002,
+      "step": 15988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.173476736928026e-05,
+      "loss": 0.0003,
+      "step": 15989
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1732366639458396e-05,
+      "loss": 1.1508,
+      "step": 15990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.172996590963653e-05,
+      "loss": 0.0004,
+      "step": 15991
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1727565179814665e-05,
+      "loss": 0.0027,
+      "step": 15992
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1725164449992798e-05,
+      "loss": 0.0002,
+      "step": 15993
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1722763720170933e-05,
+      "loss": 0.9266,
+      "step": 15994
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1720362990349067e-05,
+      "loss": 0.0002,
+      "step": 15995
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17179622605272e-05,
+      "loss": 0.0003,
+      "step": 15996
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1715561530705335e-05,
+      "loss": 0.0004,
+      "step": 15997
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1713160800883469e-05,
+      "loss": 0.002,
+      "step": 15998
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1710760071061604e-05,
+      "loss": 0.0004,
+      "step": 15999
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1708359341239738e-05,
+      "loss": 0.0004,
+      "step": 16000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1705958611417873e-05,
+      "loss": 0.0022,
+      "step": 16001
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1703557881596006e-05,
+      "loss": 0.0003,
+      "step": 16002
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.170115715177414e-05,
+      "loss": 0.0037,
+      "step": 16003
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1698756421952273e-05,
+      "loss": 0.0003,
+      "step": 16004
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1696355692130408e-05,
+      "loss": 0.0015,
+      "step": 16005
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1693954962308542e-05,
+      "loss": 0.0005,
+      "step": 16006
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1691554232486675e-05,
+      "loss": 0.0003,
+      "step": 16007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.168915350266481e-05,
+      "loss": 0.0004,
+      "step": 16008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1686752772842944e-05,
+      "loss": 0.0005,
+      "step": 16009
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.168435204302108e-05,
+      "loss": 0.0003,
+      "step": 16010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1681951313199213e-05,
+      "loss": 0.0006,
+      "step": 16011
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1679550583377346e-05,
+      "loss": 0.0004,
+      "step": 16012
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1677149853555482e-05,
+      "loss": 0.0017,
+      "step": 16013
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1674749123733615e-05,
+      "loss": 0.0003,
+      "step": 16014
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.167234839391175e-05,
+      "loss": 0.0388,
+      "step": 16015
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1669947664089884e-05,
+      "loss": 0.0014,
+      "step": 16016
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1667546934268019e-05,
+      "loss": 0.001,
+      "step": 16017
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1665146204446152e-05,
+      "loss": 0.1633,
+      "step": 16018
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1662745474624288e-05,
+      "loss": 0.0014,
+      "step": 16019
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1660344744802421e-05,
+      "loss": 0.0004,
+      "step": 16020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1657944014980555e-05,
+      "loss": 0.0017,
+      "step": 16021
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.165554328515869e-05,
+      "loss": 0.0012,
+      "step": 16022
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1653142555336823e-05,
+      "loss": 0.0004,
+      "step": 16023
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1650741825514957e-05,
+      "loss": 0.0003,
+      "step": 16024
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.164834109569309e-05,
+      "loss": 0.0004,
+      "step": 16025
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1645940365871225e-05,
+      "loss": 0.0034,
+      "step": 16026
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1643539636049359e-05,
+      "loss": 0.0015,
+      "step": 16027
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1641138906227492e-05,
+      "loss": 1.0396,
+      "step": 16028
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1638738176405628e-05,
+      "loss": 0.0004,
+      "step": 16029
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1636337446583761e-05,
+      "loss": 0.0015,
+      "step": 16030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1633936716761896e-05,
+      "loss": 0.0004,
+      "step": 16031
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.163153598694003e-05,
+      "loss": 0.0004,
+      "step": 16032
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1629135257118165e-05,
+      "loss": 0.0004,
+      "step": 16033
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1626734527296298e-05,
+      "loss": 0.0004,
+      "step": 16034
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1624333797474434e-05,
+      "loss": 0.0005,
+      "step": 16035
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1621933067652567e-05,
+      "loss": 0.0003,
+      "step": 16036
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.16195323378307e-05,
+      "loss": 0.0016,
+      "step": 16037
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1617131608008836e-05,
+      "loss": 0.0004,
+      "step": 16038
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.161473087818697e-05,
+      "loss": 0.0003,
+      "step": 16039
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1612330148365105e-05,
+      "loss": 0.0004,
+      "step": 16040
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1609929418543238e-05,
+      "loss": 0.0005,
+      "step": 16041
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1607528688721373e-05,
+      "loss": 0.0018,
+      "step": 16042
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1605127958899507e-05,
+      "loss": 0.0005,
+      "step": 16043
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.160272722907764e-05,
+      "loss": 0.0004,
+      "step": 16044
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1600326499255774e-05,
+      "loss": 0.0004,
+      "step": 16045
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1597925769433907e-05,
+      "loss": 0.6603,
+      "step": 16046
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1595525039612042e-05,
+      "loss": 0.0005,
+      "step": 16047
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1593124309790176e-05,
+      "loss": 0.002,
+      "step": 16048
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1590723579968311e-05,
+      "loss": 0.0004,
+      "step": 16049
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1588322850146445e-05,
+      "loss": 0.0004,
+      "step": 16050
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.158592212032458e-05,
+      "loss": 0.0004,
+      "step": 16051
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1583521390502713e-05,
+      "loss": 0.0004,
+      "step": 16052
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1581120660680847e-05,
+      "loss": 0.0016,
+      "step": 16053
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1578719930858982e-05,
+      "loss": 0.0012,
+      "step": 16054
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1576319201037115e-05,
+      "loss": 0.0013,
+      "step": 16055
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.157391847121525e-05,
+      "loss": 0.0004,
+      "step": 16056
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1571517741393384e-05,
+      "loss": 0.0005,
+      "step": 16057
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.156911701157152e-05,
+      "loss": 0.0004,
+      "step": 16058
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1566716281749653e-05,
+      "loss": 0.0004,
+      "step": 16059
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1564315551927786e-05,
+      "loss": 0.0004,
+      "step": 16060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1561914822105921e-05,
+      "loss": 0.0013,
+      "step": 16061
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1559514092284055e-05,
+      "loss": 0.0004,
+      "step": 16062
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.155711336246219e-05,
+      "loss": 0.0004,
+      "step": 16063
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1554712632640324e-05,
+      "loss": 0.0021,
+      "step": 16064
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1552311902818457e-05,
+      "loss": 0.0005,
+      "step": 16065
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.154991117299659e-05,
+      "loss": 0.0005,
+      "step": 16066
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1547510443174726e-05,
+      "loss": 0.0003,
+      "step": 16067
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.154510971335286e-05,
+      "loss": 0.0003,
+      "step": 16068
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1542708983530993e-05,
+      "loss": 0.0005,
+      "step": 16069
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1540308253709128e-05,
+      "loss": 0.0004,
+      "step": 16070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1537907523887261e-05,
+      "loss": 0.0003,
+      "step": 16071
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1535506794065397e-05,
+      "loss": 0.0004,
+      "step": 16072
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.153310606424353e-05,
+      "loss": 0.0016,
+      "step": 16073
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1530705334421665e-05,
+      "loss": 0.0004,
+      "step": 16074
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1528304604599799e-05,
+      "loss": 0.0013,
+      "step": 16075
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1525903874777932e-05,
+      "loss": 0.0242,
+      "step": 16076
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1523503144956068e-05,
+      "loss": 0.0003,
+      "step": 16077
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1521102415134201e-05,
+      "loss": 0.0003,
+      "step": 16078
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1518701685312336e-05,
+      "loss": 0.0004,
+      "step": 16079
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.151630095549047e-05,
+      "loss": 0.0004,
+      "step": 16080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1513900225668605e-05,
+      "loss": 0.0004,
+      "step": 16081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1511499495846738e-05,
+      "loss": 0.0032,
+      "step": 16082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1509098766024874e-05,
+      "loss": 0.0004,
+      "step": 16083
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1506698036203007e-05,
+      "loss": 0.0004,
+      "step": 16084
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.150429730638114e-05,
+      "loss": 0.0003,
+      "step": 16085
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1501896576559274e-05,
+      "loss": 0.0003,
+      "step": 16086
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1499495846737408e-05,
+      "loss": 0.0005,
+      "step": 16087
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1497095116915543e-05,
+      "loss": 0.0003,
+      "step": 16088
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1494694387093676e-05,
+      "loss": 0.0004,
+      "step": 16089
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1492293657271811e-05,
+      "loss": 0.0004,
+      "step": 16090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1489892927449945e-05,
+      "loss": 0.0015,
+      "step": 16091
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1487492197628078e-05,
+      "loss": 0.0003,
+      "step": 16092
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1485091467806214e-05,
+      "loss": 0.0005,
+      "step": 16093
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1482690737984347e-05,
+      "loss": 0.0004,
+      "step": 16094
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1480290008162482e-05,
+      "loss": 0.0022,
+      "step": 16095
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1477889278340616e-05,
+      "loss": 0.0017,
+      "step": 16096
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1475488548518751e-05,
+      "loss": 0.0018,
+      "step": 16097
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1473087818696884e-05,
+      "loss": 0.0004,
+      "step": 16098
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1470687088875018e-05,
+      "loss": 0.0003,
+      "step": 16099
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1468286359053153e-05,
+      "loss": 0.0004,
+      "step": 16100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1465885629231287e-05,
+      "loss": 0.0014,
+      "step": 16101
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1463484899409422e-05,
+      "loss": 0.0005,
+      "step": 16102
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1461084169587555e-05,
+      "loss": 0.0003,
+      "step": 16103
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.145868343976569e-05,
+      "loss": 0.0003,
+      "step": 16104
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1456282709943824e-05,
+      "loss": 0.0003,
+      "step": 16105
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1453881980121957e-05,
+      "loss": 0.0029,
+      "step": 16106
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1451481250300091e-05,
+      "loss": 0.0003,
+      "step": 16107
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1449080520478224e-05,
+      "loss": 0.0004,
+      "step": 16108
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.144667979065636e-05,
+      "loss": 0.0012,
+      "step": 16109
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1444279060834493e-05,
+      "loss": 0.0004,
+      "step": 16110
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1441878331012628e-05,
+      "loss": 0.0003,
+      "step": 16111
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1439477601190762e-05,
+      "loss": 0.0013,
+      "step": 16112
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1437076871368897e-05,
+      "loss": 0.0004,
+      "step": 16113
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.143467614154703e-05,
+      "loss": 0.0028,
+      "step": 16114
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1432275411725164e-05,
+      "loss": 0.0005,
+      "step": 16115
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.14298746819033e-05,
+      "loss": 0.0003,
+      "step": 16116
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1427473952081433e-05,
+      "loss": 0.0003,
+      "step": 16117
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1425073222259568e-05,
+      "loss": 0.0002,
+      "step": 16118
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1422672492437701e-05,
+      "loss": 0.0003,
+      "step": 16119
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1420271762615837e-05,
+      "loss": 0.0004,
+      "step": 16120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.141787103279397e-05,
+      "loss": 1.1737,
+      "step": 16121
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1415470302972105e-05,
+      "loss": 0.0003,
+      "step": 16122
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1413069573150239e-05,
+      "loss": 0.0031,
+      "step": 16123
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1410668843328372e-05,
+      "loss": 0.0003,
+      "step": 16124
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1408268113506507e-05,
+      "loss": 0.0015,
+      "step": 16125
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1405867383684641e-05,
+      "loss": 0.0003,
+      "step": 16126
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1403466653862774e-05,
+      "loss": 0.0013,
+      "step": 16127
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1401065924040908e-05,
+      "loss": 0.0007,
+      "step": 16128
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1398665194219043e-05,
+      "loss": 0.0003,
+      "step": 16129
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1396264464397177e-05,
+      "loss": 0.0004,
+      "step": 16130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.139386373457531e-05,
+      "loss": 0.0019,
+      "step": 16131
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1391463004753445e-05,
+      "loss": 0.0003,
+      "step": 16132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1389062274931579e-05,
+      "loss": 0.0003,
+      "step": 16133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1386661545109714e-05,
+      "loss": 0.0012,
+      "step": 16134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1384260815287847e-05,
+      "loss": 0.0017,
+      "step": 16135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1381860085465983e-05,
+      "loss": 0.0004,
+      "step": 16136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1379459355644116e-05,
+      "loss": 0.0003,
+      "step": 16137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1377058625822251e-05,
+      "loss": 0.0029,
+      "step": 16138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1374657896000385e-05,
+      "loss": 0.0003,
+      "step": 16139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1372257166178518e-05,
+      "loss": 0.0003,
+      "step": 16140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1369856436356654e-05,
+      "loss": 0.0004,
+      "step": 16141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1367455706534787e-05,
+      "loss": 0.0002,
+      "step": 16142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1365054976712922e-05,
+      "loss": 0.0003,
+      "step": 16143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1362654246891056e-05,
+      "loss": 0.0018,
+      "step": 16144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1360253517069191e-05,
+      "loss": 0.003,
+      "step": 16145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1357852787247324e-05,
+      "loss": 0.5583,
+      "step": 16146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1355452057425458e-05,
+      "loss": 0.0003,
+      "step": 16147
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1353051327603591e-05,
+      "loss": 0.0003,
+      "step": 16148
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1350650597781725e-05,
+      "loss": 0.0003,
+      "step": 16149
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.134824986795986e-05,
+      "loss": 0.0003,
+      "step": 16150
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1345849138137994e-05,
+      "loss": 0.0003,
+      "step": 16151
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1343448408316129e-05,
+      "loss": 0.0004,
+      "step": 16152
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1341047678494262e-05,
+      "loss": 0.0003,
+      "step": 16153
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1338646948672397e-05,
+      "loss": 0.001,
+      "step": 16154
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1336246218850531e-05,
+      "loss": 0.0002,
+      "step": 16155
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1333845489028664e-05,
+      "loss": 0.0003,
+      "step": 16156
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.13314447592068e-05,
+      "loss": 0.0003,
+      "step": 16157
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1329044029384933e-05,
+      "loss": 0.0003,
+      "step": 16158
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1326643299563068e-05,
+      "loss": 0.0019,
+      "step": 16159
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1324242569741202e-05,
+      "loss": 0.0003,
+      "step": 16160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1321841839919337e-05,
+      "loss": 0.0019,
+      "step": 16161
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.131944111009747e-05,
+      "loss": 0.0002,
+      "step": 16162
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1317040380275604e-05,
+      "loss": 0.0027,
+      "step": 16163
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1314639650453739e-05,
+      "loss": 0.0004,
+      "step": 16164
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1312238920631873e-05,
+      "loss": 0.001,
+      "step": 16165
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1309838190810008e-05,
+      "loss": 0.0003,
+      "step": 16166
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1307437460988141e-05,
+      "loss": 0.0004,
+      "step": 16167
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1305036731166275e-05,
+      "loss": 0.0017,
+      "step": 16168
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1302636001344408e-05,
+      "loss": 0.0011,
+      "step": 16169
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1300235271522543e-05,
+      "loss": 0.0002,
+      "step": 16170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1297834541700677e-05,
+      "loss": 0.0003,
+      "step": 16171
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.129543381187881e-05,
+      "loss": 0.639,
+      "step": 16172
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1293033082056946e-05,
+      "loss": 0.0003,
+      "step": 16173
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.129063235223508e-05,
+      "loss": 0.0003,
+      "step": 16174
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1288231622413214e-05,
+      "loss": 0.0003,
+      "step": 16175
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1285830892591348e-05,
+      "loss": 0.0003,
+      "step": 16176
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1283430162769483e-05,
+      "loss": 0.0003,
+      "step": 16177
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1281029432947617e-05,
+      "loss": 0.0002,
+      "step": 16178
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.127862870312575e-05,
+      "loss": 0.0004,
+      "step": 16179
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1276227973303885e-05,
+      "loss": 0.0013,
+      "step": 16180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1273827243482019e-05,
+      "loss": 0.0017,
+      "step": 16181
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1271426513660154e-05,
+      "loss": 0.0003,
+      "step": 16182
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1269025783838287e-05,
+      "loss": 0.0023,
+      "step": 16183
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1266625054016423e-05,
+      "loss": 0.0003,
+      "step": 16184
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1264224324194556e-05,
+      "loss": 0.0011,
+      "step": 16185
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1261823594372691e-05,
+      "loss": 0.0018,
+      "step": 16186
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1259422864550825e-05,
+      "loss": 0.0015,
+      "step": 16187
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1257022134728958e-05,
+      "loss": 0.0003,
+      "step": 16188
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1254621404907092e-05,
+      "loss": 0.0002,
+      "step": 16189
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1252220675085225e-05,
+      "loss": 0.0003,
+      "step": 16190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.124981994526336e-05,
+      "loss": 0.0002,
+      "step": 16191
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1247419215441494e-05,
+      "loss": 0.0012,
+      "step": 16192
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1245018485619629e-05,
+      "loss": 0.0004,
+      "step": 16193
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1242617755797763e-05,
+      "loss": 0.0003,
+      "step": 16194
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1240217025975896e-05,
+      "loss": 0.0003,
+      "step": 16195
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1237816296154031e-05,
+      "loss": 0.0023,
+      "step": 16196
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1235415566332165e-05,
+      "loss": 0.0003,
+      "step": 16197
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.12330148365103e-05,
+      "loss": 0.002,
+      "step": 16198
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1230614106688433e-05,
+      "loss": 0.0019,
+      "step": 16199
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1228213376866569e-05,
+      "loss": 0.8505,
+      "step": 16200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1225812647044702e-05,
+      "loss": 0.001,
+      "step": 16201
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1223411917222837e-05,
+      "loss": 0.0004,
+      "step": 16202
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.122101118740097e-05,
+      "loss": 0.0003,
+      "step": 16203
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1218610457579104e-05,
+      "loss": 0.0003,
+      "step": 16204
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.121620972775724e-05,
+      "loss": 0.0003,
+      "step": 16205
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1213808997935373e-05,
+      "loss": 0.0007,
+      "step": 16206
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1211408268113508e-05,
+      "loss": 0.0005,
+      "step": 16207
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1209007538291642e-05,
+      "loss": 0.0002,
+      "step": 16208
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1206606808469775e-05,
+      "loss": 0.0003,
+      "step": 16209
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1204206078647909e-05,
+      "loss": 0.0003,
+      "step": 16210
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1201805348826042e-05,
+      "loss": 0.0003,
+      "step": 16211
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1199404619004177e-05,
+      "loss": 0.0019,
+      "step": 16212
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1197003889182311e-05,
+      "loss": 0.0013,
+      "step": 16213
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1194603159360446e-05,
+      "loss": 0.0002,
+      "step": 16214
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.119220242953858e-05,
+      "loss": 0.0013,
+      "step": 16215
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1189801699716715e-05,
+      "loss": 0.0002,
+      "step": 16216
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1187400969894848e-05,
+      "loss": 0.0004,
+      "step": 16217
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1185000240072983e-05,
+      "loss": 0.6283,
+      "step": 16218
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1182599510251117e-05,
+      "loss": 0.0009,
+      "step": 16219
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.118019878042925e-05,
+      "loss": 0.001,
+      "step": 16220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1177798050607386e-05,
+      "loss": 0.0003,
+      "step": 16221
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1175397320785519e-05,
+      "loss": 0.0003,
+      "step": 16222
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1172996590963654e-05,
+      "loss": 0.3221,
+      "step": 16223
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1170595861141788e-05,
+      "loss": 0.0014,
+      "step": 16224
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1168195131319923e-05,
+      "loss": 0.0003,
+      "step": 16225
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1165794401498056e-05,
+      "loss": 0.0003,
+      "step": 16226
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.116339367167619e-05,
+      "loss": 0.0003,
+      "step": 16227
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1160992941854325e-05,
+      "loss": 0.0002,
+      "step": 16228
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1158592212032459e-05,
+      "loss": 0.0015,
+      "step": 16229
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1156191482210592e-05,
+      "loss": 0.0017,
+      "step": 16230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1153790752388726e-05,
+      "loss": 0.0009,
+      "step": 16231
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.115139002256686e-05,
+      "loss": 0.0003,
+      "step": 16232
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1148989292744994e-05,
+      "loss": 0.0003,
+      "step": 16233
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.114658856292313e-05,
+      "loss": 0.0002,
+      "step": 16234
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1144187833101263e-05,
+      "loss": 0.0003,
+      "step": 16235
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1141787103279396e-05,
+      "loss": 0.0003,
+      "step": 16236
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1139386373457532e-05,
+      "loss": 0.0003,
+      "step": 16237
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1136985643635665e-05,
+      "loss": 0.0013,
+      "step": 16238
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.11345849138138e-05,
+      "loss": 0.0003,
+      "step": 16239
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1132184183991934e-05,
+      "loss": 0.0003,
+      "step": 16240
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1129783454170069e-05,
+      "loss": 0.0002,
+      "step": 16241
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1127382724348203e-05,
+      "loss": 0.0003,
+      "step": 16242
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1124981994526336e-05,
+      "loss": 0.0003,
+      "step": 16243
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1122581264704471e-05,
+      "loss": 0.0002,
+      "step": 16244
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1120180534882605e-05,
+      "loss": 0.0003,
+      "step": 16245
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.111777980506074e-05,
+      "loss": 0.0003,
+      "step": 16246
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1115379075238873e-05,
+      "loss": 0.5835,
+      "step": 16247
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1112978345417009e-05,
+      "loss": 0.0003,
+      "step": 16248
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1110577615595142e-05,
+      "loss": 0.0015,
+      "step": 16249
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1108176885773276e-05,
+      "loss": 0.0003,
+      "step": 16250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1105776155951409e-05,
+      "loss": 0.0003,
+      "step": 16251
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1103375426129543e-05,
+      "loss": 0.0003,
+      "step": 16252
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1100974696307678e-05,
+      "loss": 0.0003,
+      "step": 16253
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1098573966485811e-05,
+      "loss": 0.0015,
+      "step": 16254
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1096173236663946e-05,
+      "loss": 0.0003,
+      "step": 16255
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.109377250684208e-05,
+      "loss": 0.0002,
+      "step": 16256
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1091371777020215e-05,
+      "loss": 0.0041,
+      "step": 16257
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1088971047198349e-05,
+      "loss": 0.0002,
+      "step": 16258
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1086570317376482e-05,
+      "loss": 0.0009,
+      "step": 16259
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1084169587554617e-05,
+      "loss": 0.0118,
+      "step": 16260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.108176885773275e-05,
+      "loss": 0.0002,
+      "step": 16261
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1079368127910886e-05,
+      "loss": 0.0022,
+      "step": 16262
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.107696739808902e-05,
+      "loss": 0.0002,
+      "step": 16263
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1074566668267155e-05,
+      "loss": 0.0002,
+      "step": 16264
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1072165938445288e-05,
+      "loss": 0.0003,
+      "step": 16265
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1069765208623423e-05,
+      "loss": 0.0003,
+      "step": 16266
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1067364478801557e-05,
+      "loss": 0.0003,
+      "step": 16267
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.106496374897969e-05,
+      "loss": 0.0003,
+      "step": 16268
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1062563019157826e-05,
+      "loss": 0.0011,
+      "step": 16269
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1060162289335959e-05,
+      "loss": 0.0015,
+      "step": 16270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1057761559514092e-05,
+      "loss": 0.0011,
+      "step": 16271
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1055360829692226e-05,
+      "loss": 0.0011,
+      "step": 16272
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1052960099870361e-05,
+      "loss": 0.0041,
+      "step": 16273
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1050559370048495e-05,
+      "loss": 0.001,
+      "step": 16274
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1048158640226628e-05,
+      "loss": 0.0002,
+      "step": 16275
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1045757910404763e-05,
+      "loss": 0.5714,
+      "step": 16276
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1043357180582897e-05,
+      "loss": 0.0022,
+      "step": 16277
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1040956450761032e-05,
+      "loss": 0.0002,
+      "step": 16278
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1038555720939166e-05,
+      "loss": 0.0003,
+      "step": 16279
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.10361549911173e-05,
+      "loss": 0.0002,
+      "step": 16280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1033754261295434e-05,
+      "loss": 0.0021,
+      "step": 16281
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.103135353147357e-05,
+      "loss": 0.0003,
+      "step": 16282
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1028952801651703e-05,
+      "loss": 0.0002,
+      "step": 16283
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1026552071829836e-05,
+      "loss": 0.0011,
+      "step": 16284
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1024151342007972e-05,
+      "loss": 0.0002,
+      "step": 16285
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1021750612186105e-05,
+      "loss": 0.0002,
+      "step": 16286
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.101934988236424e-05,
+      "loss": 0.0003,
+      "step": 16287
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1016949152542374e-05,
+      "loss": 0.0002,
+      "step": 16288
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1014548422720509e-05,
+      "loss": 0.0008,
+      "step": 16289
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1012147692898642e-05,
+      "loss": 0.001,
+      "step": 16290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1009746963076776e-05,
+      "loss": 0.0003,
+      "step": 16291
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.100734623325491e-05,
+      "loss": 0.0009,
+      "step": 16292
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1004945503433043e-05,
+      "loss": 0.0003,
+      "step": 16293
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1002544773611178e-05,
+      "loss": 0.0003,
+      "step": 16294
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1000144043789312e-05,
+      "loss": 0.0014,
+      "step": 16295
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0997743313967447e-05,
+      "loss": 0.0003,
+      "step": 16296
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.099534258414558e-05,
+      "loss": 0.0006,
+      "step": 16297
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0992941854323715e-05,
+      "loss": 0.0003,
+      "step": 16298
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0990541124501849e-05,
+      "loss": 0.0002,
+      "step": 16299
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0988140394679982e-05,
+      "loss": 0.0002,
+      "step": 16300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0985739664858118e-05,
+      "loss": 0.0002,
+      "step": 16301
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0983338935036251e-05,
+      "loss": 0.0003,
+      "step": 16302
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0980938205214386e-05,
+      "loss": 0.0003,
+      "step": 16303
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.097853747539252e-05,
+      "loss": 0.0009,
+      "step": 16304
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0976136745570655e-05,
+      "loss": 0.001,
+      "step": 16305
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0973736015748789e-05,
+      "loss": 0.0009,
+      "step": 16306
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0971335285926922e-05,
+      "loss": 0.0008,
+      "step": 16307
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0968934556105057e-05,
+      "loss": 0.0003,
+      "step": 16308
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.096653382628319e-05,
+      "loss": 0.6387,
+      "step": 16309
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0964133096461326e-05,
+      "loss": 0.0016,
+      "step": 16310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.096173236663946e-05,
+      "loss": 0.0003,
+      "step": 16311
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0959331636817593e-05,
+      "loss": 0.0002,
+      "step": 16312
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0956930906995726e-05,
+      "loss": 0.0003,
+      "step": 16313
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0954530177173862e-05,
+      "loss": 0.0002,
+      "step": 16314
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0952129447351995e-05,
+      "loss": 0.0012,
+      "step": 16315
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0949728717530129e-05,
+      "loss": 0.0002,
+      "step": 16316
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0947327987708264e-05,
+      "loss": 0.0003,
+      "step": 16317
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0944927257886397e-05,
+      "loss": 0.0002,
+      "step": 16318
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0942526528064532e-05,
+      "loss": 0.0003,
+      "step": 16319
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0940125798242666e-05,
+      "loss": 0.0002,
+      "step": 16320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0937725068420801e-05,
+      "loss": 0.0003,
+      "step": 16321
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0935324338598935e-05,
+      "loss": 0.5536,
+      "step": 16322
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0932923608777068e-05,
+      "loss": 0.0003,
+      "step": 16323
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0930522878955203e-05,
+      "loss": 0.0003,
+      "step": 16324
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0928122149133337e-05,
+      "loss": 0.0003,
+      "step": 16325
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0925721419311472e-05,
+      "loss": 0.0002,
+      "step": 16326
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0923320689489605e-05,
+      "loss": 0.0003,
+      "step": 16327
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.092091995966774e-05,
+      "loss": 0.0003,
+      "step": 16328
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0918519229845874e-05,
+      "loss": 0.0002,
+      "step": 16329
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.091611850002401e-05,
+      "loss": 0.0003,
+      "step": 16330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0913717770202143e-05,
+      "loss": 0.0004,
+      "step": 16331
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0911317040380276e-05,
+      "loss": 0.0003,
+      "step": 16332
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.090891631055841e-05,
+      "loss": 0.0002,
+      "step": 16333
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0906515580736543e-05,
+      "loss": 0.0002,
+      "step": 16334
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0904114850914678e-05,
+      "loss": 0.0002,
+      "step": 16335
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0901714121092812e-05,
+      "loss": 0.0003,
+      "step": 16336
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0899313391270947e-05,
+      "loss": 0.0003,
+      "step": 16337
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.089691266144908e-05,
+      "loss": 0.0003,
+      "step": 16338
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0894511931627214e-05,
+      "loss": 0.0003,
+      "step": 16339
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.089211120180535e-05,
+      "loss": 0.0002,
+      "step": 16340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0889710471983483e-05,
+      "loss": 0.0002,
+      "step": 16341
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0887309742161618e-05,
+      "loss": 0.0017,
+      "step": 16342
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0884909012339752e-05,
+      "loss": 0.0006,
+      "step": 16343
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0882508282517887e-05,
+      "loss": 0.0002,
+      "step": 16344
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.088010755269602e-05,
+      "loss": 0.0002,
+      "step": 16345
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0877706822874155e-05,
+      "loss": 0.0002,
+      "step": 16346
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0875306093052289e-05,
+      "loss": 0.0002,
+      "step": 16347
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0872905363230422e-05,
+      "loss": 0.0003,
+      "step": 16348
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0870504633408558e-05,
+      "loss": 0.0003,
+      "step": 16349
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0868103903586691e-05,
+      "loss": 0.0002,
+      "step": 16350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0865703173764826e-05,
+      "loss": 0.0016,
+      "step": 16351
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.086330244394296e-05,
+      "loss": 0.0004,
+      "step": 16352
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0860901714121093e-05,
+      "loss": 0.0003,
+      "step": 16353
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0858500984299227e-05,
+      "loss": 0.0002,
+      "step": 16354
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.085610025447736e-05,
+      "loss": 0.0003,
+      "step": 16355
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0853699524655495e-05,
+      "loss": 0.0003,
+      "step": 16356
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0851298794833629e-05,
+      "loss": 0.0002,
+      "step": 16357
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0848898065011764e-05,
+      "loss": 0.0003,
+      "step": 16358
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0846497335189898e-05,
+      "loss": 0.0018,
+      "step": 16359
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0844096605368033e-05,
+      "loss": 0.0002,
+      "step": 16360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0841695875546166e-05,
+      "loss": 0.0028,
+      "step": 16361
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0839295145724301e-05,
+      "loss": 0.0003,
+      "step": 16362
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0836894415902435e-05,
+      "loss": 0.5326,
+      "step": 16363
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0834493686080568e-05,
+      "loss": 0.0002,
+      "step": 16364
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0832092956258704e-05,
+      "loss": 0.0002,
+      "step": 16365
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0829692226436837e-05,
+      "loss": 0.0028,
+      "step": 16366
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0827291496614972e-05,
+      "loss": 0.0022,
+      "step": 16367
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0824890766793106e-05,
+      "loss": 0.0024,
+      "step": 16368
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0822490036971241e-05,
+      "loss": 0.0028,
+      "step": 16369
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0820089307149375e-05,
+      "loss": 0.0003,
+      "step": 16370
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0817688577327508e-05,
+      "loss": 0.0027,
+      "step": 16371
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0815287847505643e-05,
+      "loss": 0.0002,
+      "step": 16372
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0812887117683777e-05,
+      "loss": 0.0002,
+      "step": 16373
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.081048638786191e-05,
+      "loss": 0.0023,
+      "step": 16374
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0808085658040044e-05,
+      "loss": 0.0002,
+      "step": 16375
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0805684928218179e-05,
+      "loss": 0.0003,
+      "step": 16376
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0803284198396312e-05,
+      "loss": 0.5553,
+      "step": 16377
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0800883468574448e-05,
+      "loss": 0.0003,
+      "step": 16378
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0798482738752581e-05,
+      "loss": 0.0058,
+      "step": 16379
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0796082008930715e-05,
+      "loss": 0.0031,
+      "step": 16380
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.079368127910885e-05,
+      "loss": 0.0003,
+      "step": 16381
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0791280549286983e-05,
+      "loss": 0.9274,
+      "step": 16382
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0788879819465118e-05,
+      "loss": 0.0025,
+      "step": 16383
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0786479089643252e-05,
+      "loss": 0.0019,
+      "step": 16384
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0784078359821387e-05,
+      "loss": 0.0002,
+      "step": 16385
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.078167762999952e-05,
+      "loss": 0.0002,
+      "step": 16386
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0779276900177654e-05,
+      "loss": 0.0003,
+      "step": 16387
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.077687617035579e-05,
+      "loss": 0.0002,
+      "step": 16388
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0774475440533923e-05,
+      "loss": 0.0002,
+      "step": 16389
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0772074710712058e-05,
+      "loss": 0.0003,
+      "step": 16390
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0769673980890191e-05,
+      "loss": 0.0023,
+      "step": 16391
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0767273251068327e-05,
+      "loss": 0.0003,
+      "step": 16392
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.076487252124646e-05,
+      "loss": 0.0002,
+      "step": 16393
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0762471791424594e-05,
+      "loss": 0.0003,
+      "step": 16394
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0760071061602727e-05,
+      "loss": 0.0003,
+      "step": 16395
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.075767033178086e-05,
+      "loss": 0.0021,
+      "step": 16396
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0755269601958996e-05,
+      "loss": 0.0003,
+      "step": 16397
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.075286887213713e-05,
+      "loss": 0.0024,
+      "step": 16398
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0750468142315264e-05,
+      "loss": 0.0002,
+      "step": 16399
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0748067412493398e-05,
+      "loss": 0.0003,
+      "step": 16400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0745666682671533e-05,
+      "loss": 0.0023,
+      "step": 16401
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0743265952849667e-05,
+      "loss": 0.0003,
+      "step": 16402
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.07408652230278e-05,
+      "loss": 0.0003,
+      "step": 16403
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0738464493205935e-05,
+      "loss": 0.0004,
+      "step": 16404
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0736063763384069e-05,
+      "loss": 0.0003,
+      "step": 16405
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0733663033562204e-05,
+      "loss": 0.0002,
+      "step": 16406
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0731262303740338e-05,
+      "loss": 0.0017,
+      "step": 16407
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0728861573918473e-05,
+      "loss": 0.0003,
+      "step": 16408
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0726460844096606e-05,
+      "loss": 0.0003,
+      "step": 16409
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0724060114274741e-05,
+      "loss": 0.0015,
+      "step": 16410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0721659384452875e-05,
+      "loss": 0.0002,
+      "step": 16411
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0719258654631008e-05,
+      "loss": 0.0003,
+      "step": 16412
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0716857924809144e-05,
+      "loss": 0.0004,
+      "step": 16413
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0714457194987277e-05,
+      "loss": 0.0014,
+      "step": 16414
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.071205646516541e-05,
+      "loss": 0.0038,
+      "step": 16415
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0709655735343544e-05,
+      "loss": 0.0003,
+      "step": 16416
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.070725500552168e-05,
+      "loss": 0.0003,
+      "step": 16417
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0704854275699813e-05,
+      "loss": 0.0018,
+      "step": 16418
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0702453545877946e-05,
+      "loss": 0.0022,
+      "step": 16419
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0700052816056081e-05,
+      "loss": 0.0003,
+      "step": 16420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0697652086234215e-05,
+      "loss": 0.0015,
+      "step": 16421
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.069525135641235e-05,
+      "loss": 0.0002,
+      "step": 16422
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0692850626590484e-05,
+      "loss": 0.0003,
+      "step": 16423
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0690449896768619e-05,
+      "loss": 0.0003,
+      "step": 16424
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0688049166946752e-05,
+      "loss": 0.0023,
+      "step": 16425
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0685648437124887e-05,
+      "loss": 0.0012,
+      "step": 16426
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0683247707303021e-05,
+      "loss": 0.0003,
+      "step": 16427
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0680846977481154e-05,
+      "loss": 0.0002,
+      "step": 16428
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.067844624765929e-05,
+      "loss": 0.0003,
+      "step": 16429
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0676045517837423e-05,
+      "loss": 0.0013,
+      "step": 16430
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0673644788015558e-05,
+      "loss": 0.0003,
+      "step": 16431
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0671244058193692e-05,
+      "loss": 0.0013,
+      "step": 16432
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0668843328371825e-05,
+      "loss": 0.0002,
+      "step": 16433
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.066644259854996e-05,
+      "loss": 0.0003,
+      "step": 16434
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0664041868728094e-05,
+      "loss": 0.003,
+      "step": 16435
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0661641138906227e-05,
+      "loss": 0.0005,
+      "step": 16436
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0659240409084361e-05,
+      "loss": 0.0002,
+      "step": 16437
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0656839679262496e-05,
+      "loss": 0.0003,
+      "step": 16438
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.065443894944063e-05,
+      "loss": 0.0019,
+      "step": 16439
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0652038219618765e-05,
+      "loss": 0.0003,
+      "step": 16440
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0649637489796898e-05,
+      "loss": 0.0013,
+      "step": 16441
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0647236759975032e-05,
+      "loss": 0.0003,
+      "step": 16442
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0644836030153167e-05,
+      "loss": 0.0003,
+      "step": 16443
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.06424353003313e-05,
+      "loss": 0.0033,
+      "step": 16444
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0640034570509436e-05,
+      "loss": 0.0003,
+      "step": 16445
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.063763384068757e-05,
+      "loss": 0.0002,
+      "step": 16446
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0635233110865704e-05,
+      "loss": 0.0002,
+      "step": 16447
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0632832381043838e-05,
+      "loss": 0.0015,
+      "step": 16448
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0630431651221973e-05,
+      "loss": 0.0003,
+      "step": 16449
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0628030921400107e-05,
+      "loss": 0.0002,
+      "step": 16450
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.062563019157824e-05,
+      "loss": 0.0014,
+      "step": 16451
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0623229461756375e-05,
+      "loss": 0.0012,
+      "step": 16452
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0620828731934509e-05,
+      "loss": 0.0003,
+      "step": 16453
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0618428002112644e-05,
+      "loss": 0.0011,
+      "step": 16454
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0616027272290777e-05,
+      "loss": 0.0023,
+      "step": 16455
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0613626542468911e-05,
+      "loss": 0.0003,
+      "step": 16456
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0611225812647044e-05,
+      "loss": 0.0002,
+      "step": 16457
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0608825082825178e-05,
+      "loss": 0.0003,
+      "step": 16458
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0606424353003313e-05,
+      "loss": 0.0025,
+      "step": 16459
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0604023623181447e-05,
+      "loss": 0.0003,
+      "step": 16460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0601622893359582e-05,
+      "loss": 0.0002,
+      "step": 16461
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0599222163537715e-05,
+      "loss": 0.0011,
+      "step": 16462
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.059682143371585e-05,
+      "loss": 0.0003,
+      "step": 16463
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0594420703893984e-05,
+      "loss": 0.0002,
+      "step": 16464
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0592019974072119e-05,
+      "loss": 0.0013,
+      "step": 16465
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0589619244250253e-05,
+      "loss": 0.0002,
+      "step": 16466
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0587218514428386e-05,
+      "loss": 0.0003,
+      "step": 16467
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0584817784606521e-05,
+      "loss": 0.0002,
+      "step": 16468
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0582417054784655e-05,
+      "loss": 0.0012,
+      "step": 16469
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.058001632496279e-05,
+      "loss": 0.0012,
+      "step": 16470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0577615595140924e-05,
+      "loss": 0.0003,
+      "step": 16471
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0575214865319059e-05,
+      "loss": 0.0012,
+      "step": 16472
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0572814135497192e-05,
+      "loss": 0.0003,
+      "step": 16473
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0570413405675326e-05,
+      "loss": 0.001,
+      "step": 16474
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0568012675853461e-05,
+      "loss": 0.0003,
+      "step": 16475
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0565611946031594e-05,
+      "loss": 0.002,
+      "step": 16476
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0563211216209728e-05,
+      "loss": 0.0003,
+      "step": 16477
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0560810486387861e-05,
+      "loss": 0.0003,
+      "step": 16478
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0558409756565997e-05,
+      "loss": 0.0002,
+      "step": 16479
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.055600902674413e-05,
+      "loss": 0.0003,
+      "step": 16480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0553608296922265e-05,
+      "loss": 0.0003,
+      "step": 16481
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0551207567100399e-05,
+      "loss": 0.001,
+      "step": 16482
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0548806837278532e-05,
+      "loss": 0.0002,
+      "step": 16483
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0546406107456667e-05,
+      "loss": 0.0003,
+      "step": 16484
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0544005377634801e-05,
+      "loss": 0.0012,
+      "step": 16485
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0541604647812936e-05,
+      "loss": 0.0008,
+      "step": 16486
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.053920391799107e-05,
+      "loss": 0.0002,
+      "step": 16487
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0536803188169205e-05,
+      "loss": 0.0002,
+      "step": 16488
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0534402458347338e-05,
+      "loss": 0.0003,
+      "step": 16489
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0532001728525472e-05,
+      "loss": 0.0003,
+      "step": 16490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0529600998703607e-05,
+      "loss": 0.0009,
+      "step": 16491
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.052720026888174e-05,
+      "loss": 0.0003,
+      "step": 16492
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0524799539059876e-05,
+      "loss": 0.0002,
+      "step": 16493
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0522398809238009e-05,
+      "loss": 0.0003,
+      "step": 16494
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0519998079416143e-05,
+      "loss": 0.0003,
+      "step": 16495
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0517597349594278e-05,
+      "loss": 0.0003,
+      "step": 16496
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0515196619772411e-05,
+      "loss": 0.0002,
+      "step": 16497
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0512795889950545e-05,
+      "loss": 0.001,
+      "step": 16498
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0510395160128678e-05,
+      "loss": 0.0003,
+      "step": 16499
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0507994430306813e-05,
+      "loss": 0.0003,
+      "step": 16500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0505593700484947e-05,
+      "loss": 0.0016,
+      "step": 16501
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0503192970663082e-05,
+      "loss": 0.0002,
+      "step": 16502
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0500792240841216e-05,
+      "loss": 0.0009,
+      "step": 16503
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.049839151101935e-05,
+      "loss": 0.0002,
+      "step": 16504
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0495990781197484e-05,
+      "loss": 0.0009,
+      "step": 16505
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0493590051375618e-05,
+      "loss": 0.0004,
+      "step": 16506
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0491189321553753e-05,
+      "loss": 0.0009,
+      "step": 16507
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0488788591731887e-05,
+      "loss": 0.0007,
+      "step": 16508
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0486387861910022e-05,
+      "loss": 0.0002,
+      "step": 16509
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0483987132088155e-05,
+      "loss": 0.0002,
+      "step": 16510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.048158640226629e-05,
+      "loss": 0.0002,
+      "step": 16511
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0479185672444424e-05,
+      "loss": 0.0003,
+      "step": 16512
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0476784942622559e-05,
+      "loss": 0.0003,
+      "step": 16513
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0474384212800693e-05,
+      "loss": 0.0009,
+      "step": 16514
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0471983482978826e-05,
+      "loss": 0.8233,
+      "step": 16515
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0469582753156961e-05,
+      "loss": 0.0002,
+      "step": 16516
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0467182023335095e-05,
+      "loss": 0.0008,
+      "step": 16517
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0464781293513228e-05,
+      "loss": 0.0013,
+      "step": 16518
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0462380563691362e-05,
+      "loss": 0.0002,
+      "step": 16519
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0459979833869497e-05,
+      "loss": 0.0008,
+      "step": 16520
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.045757910404763e-05,
+      "loss": 0.826,
+      "step": 16521
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0455178374225764e-05,
+      "loss": 0.0002,
+      "step": 16522
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0452777644403899e-05,
+      "loss": 0.0002,
+      "step": 16523
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0450376914582033e-05,
+      "loss": 0.0008,
+      "step": 16524
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0447976184760168e-05,
+      "loss": 0.0002,
+      "step": 16525
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0445575454938301e-05,
+      "loss": 0.001,
+      "step": 16526
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0443174725116436e-05,
+      "loss": 0.0002,
+      "step": 16527
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.044077399529457e-05,
+      "loss": 0.0002,
+      "step": 16528
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0438373265472705e-05,
+      "loss": 0.0003,
+      "step": 16529
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0435972535650839e-05,
+      "loss": 0.0012,
+      "step": 16530
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0433571805828972e-05,
+      "loss": 0.0002,
+      "step": 16531
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0431171076007107e-05,
+      "loss": 0.0002,
+      "step": 16532
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.042877034618524e-05,
+      "loss": 0.0002,
+      "step": 16533
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0426369616363376e-05,
+      "loss": 0.0002,
+      "step": 16534
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.042396888654151e-05,
+      "loss": 0.0007,
+      "step": 16535
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0421568156719643e-05,
+      "loss": 0.0003,
+      "step": 16536
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0419167426897778e-05,
+      "loss": 0.0003,
+      "step": 16537
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0416766697075912e-05,
+      "loss": 0.5574,
+      "step": 16538
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0414365967254045e-05,
+      "loss": 0.0002,
+      "step": 16539
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0411965237432179e-05,
+      "loss": 0.0002,
+      "step": 16540
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0409564507610314e-05,
+      "loss": 0.0026,
+      "step": 16541
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0407163777788447e-05,
+      "loss": 0.0049,
+      "step": 16542
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0404763047966583e-05,
+      "loss": 0.0002,
+      "step": 16543
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0402362318144716e-05,
+      "loss": 0.0005,
+      "step": 16544
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0399961588322851e-05,
+      "loss": 0.0003,
+      "step": 16545
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0397560858500985e-05,
+      "loss": 0.0003,
+      "step": 16546
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0395160128679118e-05,
+      "loss": 0.0002,
+      "step": 16547
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0392759398857253e-05,
+      "loss": 0.0023,
+      "step": 16548
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0390358669035387e-05,
+      "loss": 0.0013,
+      "step": 16549
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0387957939213522e-05,
+      "loss": 0.0003,
+      "step": 16550
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0385557209391656e-05,
+      "loss": 0.0007,
+      "step": 16551
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.038315647956979e-05,
+      "loss": 0.0002,
+      "step": 16552
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0380755749747924e-05,
+      "loss": 0.0022,
+      "step": 16553
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0378355019926058e-05,
+      "loss": 0.0012,
+      "step": 16554
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0375954290104193e-05,
+      "loss": 0.0007,
+      "step": 16555
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0373553560282326e-05,
+      "loss": 0.0015,
+      "step": 16556
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.037115283046046e-05,
+      "loss": 0.0006,
+      "step": 16557
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0368752100638595e-05,
+      "loss": 0.0002,
+      "step": 16558
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0366351370816729e-05,
+      "loss": 0.0008,
+      "step": 16559
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0363950640994862e-05,
+      "loss": 0.0011,
+      "step": 16560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0361549911172997e-05,
+      "loss": 0.6407,
+      "step": 16561
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.035914918135113e-05,
+      "loss": 0.0002,
+      "step": 16562
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0356748451529264e-05,
+      "loss": 0.0002,
+      "step": 16563
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.03543477217074e-05,
+      "loss": 0.0003,
+      "step": 16564
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0351946991885533e-05,
+      "loss": 0.0003,
+      "step": 16565
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0349546262063668e-05,
+      "loss": 0.0001,
+      "step": 16566
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0347145532241802e-05,
+      "loss": 0.001,
+      "step": 16567
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0344744802419937e-05,
+      "loss": 0.0002,
+      "step": 16568
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.034234407259807e-05,
+      "loss": 0.001,
+      "step": 16569
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0339943342776204e-05,
+      "loss": 0.0002,
+      "step": 16570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0337542612954339e-05,
+      "loss": 0.0002,
+      "step": 16571
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0335141883132473e-05,
+      "loss": 0.0002,
+      "step": 16572
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0332741153310608e-05,
+      "loss": 0.0008,
+      "step": 16573
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0330340423488741e-05,
+      "loss": 0.001,
+      "step": 16574
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0327939693666876e-05,
+      "loss": 0.0002,
+      "step": 16575
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.032553896384501e-05,
+      "loss": 0.0009,
+      "step": 16576
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0323138234023143e-05,
+      "loss": 0.0002,
+      "step": 16577
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0320737504201279e-05,
+      "loss": 0.001,
+      "step": 16578
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0318336774379412e-05,
+      "loss": 0.0002,
+      "step": 16579
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0315936044557546e-05,
+      "loss": 0.0002,
+      "step": 16580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0313535314735679e-05,
+      "loss": 0.0002,
+      "step": 16581
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0311134584913814e-05,
+      "loss": 0.0003,
+      "step": 16582
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0308733855091948e-05,
+      "loss": 0.0002,
+      "step": 16583
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0306333125270083e-05,
+      "loss": 0.0002,
+      "step": 16584
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0303932395448216e-05,
+      "loss": 0.0011,
+      "step": 16585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.030153166562635e-05,
+      "loss": 0.0002,
+      "step": 16586
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0299130935804485e-05,
+      "loss": 0.0002,
+      "step": 16587
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0296730205982619e-05,
+      "loss": 0.0003,
+      "step": 16588
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0294329476160754e-05,
+      "loss": 0.0011,
+      "step": 16589
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0291928746338887e-05,
+      "loss": 0.0011,
+      "step": 16590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0289528016517022e-05,
+      "loss": 0.001,
+      "step": 16591
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0287127286695156e-05,
+      "loss": 0.0003,
+      "step": 16592
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0284726556873291e-05,
+      "loss": 0.003,
+      "step": 16593
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0282325827051425e-05,
+      "loss": 0.0002,
+      "step": 16594
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0279925097229558e-05,
+      "loss": 0.0008,
+      "step": 16595
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0277524367407693e-05,
+      "loss": 0.0002,
+      "step": 16596
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0275123637585827e-05,
+      "loss": 0.0002,
+      "step": 16597
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.027272290776396e-05,
+      "loss": 0.0009,
+      "step": 16598
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0270322177942095e-05,
+      "loss": 0.0014,
+      "step": 16599
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0267921448120229e-05,
+      "loss": 0.0001,
+      "step": 16600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0265520718298362e-05,
+      "loss": 0.0002,
+      "step": 16601
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0263119988476496e-05,
+      "loss": 0.0001,
+      "step": 16602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0260719258654631e-05,
+      "loss": 0.0009,
+      "step": 16603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0258318528832765e-05,
+      "loss": 0.0003,
+      "step": 16604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.02559177990109e-05,
+      "loss": 0.0002,
+      "step": 16605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0253517069189033e-05,
+      "loss": 0.0003,
+      "step": 16606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0251116339367169e-05,
+      "loss": 0.0002,
+      "step": 16607
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0248715609545302e-05,
+      "loss": 0.0002,
+      "step": 16608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0246314879723437e-05,
+      "loss": 0.0002,
+      "step": 16609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.024391414990157e-05,
+      "loss": 0.0003,
+      "step": 16610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0241513420079704e-05,
+      "loss": 0.0003,
+      "step": 16611
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.023911269025784e-05,
+      "loss": 0.0002,
+      "step": 16612
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0236711960435973e-05,
+      "loss": 0.0003,
+      "step": 16613
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0234311230614108e-05,
+      "loss": 0.0002,
+      "step": 16614
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0231910500792242e-05,
+      "loss": 0.0003,
+      "step": 16615
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0229509770970377e-05,
+      "loss": 0.0002,
+      "step": 16616
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.022710904114851e-05,
+      "loss": 0.0003,
+      "step": 16617
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0224708311326644e-05,
+      "loss": 0.0002,
+      "step": 16618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0222307581504777e-05,
+      "loss": 0.0002,
+      "step": 16619
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0219906851682912e-05,
+      "loss": 0.0008,
+      "step": 16620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0217506121861046e-05,
+      "loss": 0.0003,
+      "step": 16621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.021510539203918e-05,
+      "loss": 0.0003,
+      "step": 16622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0212704662217315e-05,
+      "loss": 0.0008,
+      "step": 16623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0210303932395448e-05,
+      "loss": 0.0003,
+      "step": 16624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0207903202573583e-05,
+      "loss": 0.0002,
+      "step": 16625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0205502472751717e-05,
+      "loss": 0.0002,
+      "step": 16626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.020310174292985e-05,
+      "loss": 0.0014,
+      "step": 16627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0200701013107985e-05,
+      "loss": 0.0003,
+      "step": 16628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0198300283286119e-05,
+      "loss": 0.0009,
+      "step": 16629
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0195899553464254e-05,
+      "loss": 0.0002,
+      "step": 16630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0193498823642388e-05,
+      "loss": 0.0003,
+      "step": 16631
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0191098093820523e-05,
+      "loss": 0.001,
+      "step": 16632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0188697363998656e-05,
+      "loss": 0.0002,
+      "step": 16633
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.018629663417679e-05,
+      "loss": 0.0002,
+      "step": 16634
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0183895904354925e-05,
+      "loss": 0.0002,
+      "step": 16635
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0181495174533058e-05,
+      "loss": 0.001,
+      "step": 16636
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0179094444711194e-05,
+      "loss": 0.0002,
+      "step": 16637
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0176693714889327e-05,
+      "loss": 0.0008,
+      "step": 16638
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.017429298506746e-05,
+      "loss": 0.001,
+      "step": 16639
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0171892255245596e-05,
+      "loss": 0.001,
+      "step": 16640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.016949152542373e-05,
+      "loss": 0.0002,
+      "step": 16641
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0167090795601863e-05,
+      "loss": 0.0007,
+      "step": 16642
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0164690065779996e-05,
+      "loss": 0.6164,
+      "step": 16643
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0162289335958132e-05,
+      "loss": 0.0002,
+      "step": 16644
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0159888606136265e-05,
+      "loss": 0.0002,
+      "step": 16645
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.01574878763144e-05,
+      "loss": 0.0002,
+      "step": 16646
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0155087146492534e-05,
+      "loss": 0.0002,
+      "step": 16647
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0152686416670669e-05,
+      "loss": 0.0002,
+      "step": 16648
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0150285686848802e-05,
+      "loss": 0.0002,
+      "step": 16649
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0147884957026936e-05,
+      "loss": 0.0002,
+      "step": 16650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0145484227205071e-05,
+      "loss": 0.0011,
+      "step": 16651
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0143083497383205e-05,
+      "loss": 0.0002,
+      "step": 16652
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.014068276756134e-05,
+      "loss": 0.0002,
+      "step": 16653
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0138282037739473e-05,
+      "loss": 0.0002,
+      "step": 16654
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0135881307917608e-05,
+      "loss": 0.0002,
+      "step": 16655
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0133480578095742e-05,
+      "loss": 0.0011,
+      "step": 16656
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0131079848273877e-05,
+      "loss": 0.0026,
+      "step": 16657
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.012867911845201e-05,
+      "loss": 0.0002,
+      "step": 16658
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0126278388630144e-05,
+      "loss": 0.0012,
+      "step": 16659
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0123877658808278e-05,
+      "loss": 0.0002,
+      "step": 16660
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0121476928986413e-05,
+      "loss": 0.0021,
+      "step": 16661
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0119076199164546e-05,
+      "loss": 0.0009,
+      "step": 16662
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.011667546934268e-05,
+      "loss": 0.0002,
+      "step": 16663
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0114274739520815e-05,
+      "loss": 0.0001,
+      "step": 16664
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0111874009698948e-05,
+      "loss": 0.0002,
+      "step": 16665
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0109473279877082e-05,
+      "loss": 0.0002,
+      "step": 16666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0107072550055217e-05,
+      "loss": 0.0002,
+      "step": 16667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.010467182023335e-05,
+      "loss": 0.0002,
+      "step": 16668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0102271090411486e-05,
+      "loss": 0.0002,
+      "step": 16669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.009987036058962e-05,
+      "loss": 0.001,
+      "step": 16670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0097469630767755e-05,
+      "loss": 0.0003,
+      "step": 16671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0095068900945888e-05,
+      "loss": 0.0002,
+      "step": 16672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0092668171124023e-05,
+      "loss": 0.0024,
+      "step": 16673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0090267441302157e-05,
+      "loss": 0.0002,
+      "step": 16674
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.008786671148029e-05,
+      "loss": 0.0002,
+      "step": 16675
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0085465981658425e-05,
+      "loss": 0.001,
+      "step": 16676
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0083065251836559e-05,
+      "loss": 0.0018,
+      "step": 16677
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0080664522014694e-05,
+      "loss": 0.0002,
+      "step": 16678
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0078263792192828e-05,
+      "loss": 0.0002,
+      "step": 16679
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0075863062370961e-05,
+      "loss": 0.0002,
+      "step": 16680
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0073462332549095e-05,
+      "loss": 0.0002,
+      "step": 16681
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.007106160272723e-05,
+      "loss": 0.0002,
+      "step": 16682
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0068660872905363e-05,
+      "loss": 0.0002,
+      "step": 16683
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0066260143083497e-05,
+      "loss": 0.0002,
+      "step": 16684
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0063859413261632e-05,
+      "loss": 0.0009,
+      "step": 16685
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0061458683439765e-05,
+      "loss": 0.0002,
+      "step": 16686
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.00590579536179e-05,
+      "loss": 0.0002,
+      "step": 16687
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0056657223796034e-05,
+      "loss": 0.0016,
+      "step": 16688
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.005425649397417e-05,
+      "loss": 0.0007,
+      "step": 16689
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0051855764152303e-05,
+      "loss": 0.0003,
+      "step": 16690
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0049455034330436e-05,
+      "loss": 0.0002,
+      "step": 16691
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0047054304508571e-05,
+      "loss": 0.0012,
+      "step": 16692
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0044653574686705e-05,
+      "loss": 0.0002,
+      "step": 16693
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.004225284486484e-05,
+      "loss": 0.0016,
+      "step": 16694
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0039852115042974e-05,
+      "loss": 0.0002,
+      "step": 16695
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0037451385221109e-05,
+      "loss": 0.0002,
+      "step": 16696
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0035050655399242e-05,
+      "loss": 1.0116,
+      "step": 16697
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0032649925577376e-05,
+      "loss": 0.0002,
+      "step": 16698
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0030249195755511e-05,
+      "loss": 0.0002,
+      "step": 16699
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0027848465933644e-05,
+      "loss": 0.0002,
+      "step": 16700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0025447736111778e-05,
+      "loss": 0.0002,
+      "step": 16701
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0023047006289913e-05,
+      "loss": 0.0002,
+      "step": 16702
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0020646276468047e-05,
+      "loss": 0.0011,
+      "step": 16703
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.001824554664618e-05,
+      "loss": 0.0003,
+      "step": 16704
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0015844816824315e-05,
+      "loss": 0.0012,
+      "step": 16705
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0013444087002449e-05,
+      "loss": 0.0014,
+      "step": 16706
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0011043357180582e-05,
+      "loss": 0.0002,
+      "step": 16707
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0008642627358718e-05,
+      "loss": 0.0002,
+      "step": 16708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0006241897536851e-05,
+      "loss": 0.0002,
+      "step": 16709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0003841167714986e-05,
+      "loss": 0.0003,
+      "step": 16710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.000144043789312e-05,
+      "loss": 0.0002,
+      "step": 16711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.999039708071255e-06,
+      "loss": 0.0002,
+      "step": 16712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.996638978249388e-06,
+      "loss": 0.0002,
+      "step": 16713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.994238248427522e-06,
+      "loss": 0.0001,
+      "step": 16714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.991837518605657e-06,
+      "loss": 0.0002,
+      "step": 16715
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.98943678878379e-06,
+      "loss": 0.0002,
+      "step": 16716
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.987036058961926e-06,
+      "loss": 0.0003,
+      "step": 16717
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.98463532914006e-06,
+      "loss": 0.0013,
+      "step": 16718
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.982234599318194e-06,
+      "loss": 0.0002,
+      "step": 16719
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.979833869496328e-06,
+      "loss": 0.0012,
+      "step": 16720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.977433139674461e-06,
+      "loss": 0.0002,
+      "step": 16721
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.975032409852595e-06,
+      "loss": 0.0002,
+      "step": 16722
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.97263168003073e-06,
+      "loss": 0.0003,
+      "step": 16723
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.970230950208864e-06,
+      "loss": 0.0012,
+      "step": 16724
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.967830220386997e-06,
+      "loss": 0.0002,
+      "step": 16725
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.965429490565132e-06,
+      "loss": 0.0008,
+      "step": 16726
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.963028760743266e-06,
+      "loss": 0.0002,
+      "step": 16727
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.960628030921401e-06,
+      "loss": 0.0011,
+      "step": 16728
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.958227301099534e-06,
+      "loss": 0.0009,
+      "step": 16729
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.955826571277668e-06,
+      "loss": 0.0003,
+      "step": 16730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.953425841455803e-06,
+      "loss": 0.0002,
+      "step": 16731
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.951025111633937e-06,
+      "loss": 0.0002,
+      "step": 16732
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.948624381812072e-06,
+      "loss": 0.0003,
+      "step": 16733
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.946223651990205e-06,
+      "loss": 0.0024,
+      "step": 16734
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.94382292216834e-06,
+      "loss": 0.0008,
+      "step": 16735
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.941422192346474e-06,
+      "loss": 0.0003,
+      "step": 16736
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.93902146252461e-06,
+      "loss": 0.0007,
+      "step": 16737
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.936620732702743e-06,
+      "loss": 0.0003,
+      "step": 16738
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.934220002880876e-06,
+      "loss": 0.0003,
+      "step": 16739
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.931819273059011e-06,
+      "loss": 0.0002,
+      "step": 16740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.929418543237145e-06,
+      "loss": 0.0002,
+      "step": 16741
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.927017813415278e-06,
+      "loss": 0.0002,
+      "step": 16742
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.924617083593412e-06,
+      "loss": 0.0002,
+      "step": 16743
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.922216353771547e-06,
+      "loss": 0.0002,
+      "step": 16744
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.91981562394968e-06,
+      "loss": 0.0002,
+      "step": 16745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.917414894127814e-06,
+      "loss": 0.0002,
+      "step": 16746
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.91501416430595e-06,
+      "loss": 0.0002,
+      "step": 16747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.912613434484083e-06,
+      "loss": 0.0002,
+      "step": 16748
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.910212704662218e-06,
+      "loss": 0.0002,
+      "step": 16749
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.907811974840351e-06,
+      "loss": 0.0002,
+      "step": 16750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.905411245018487e-06,
+      "loss": 0.0002,
+      "step": 16751
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.90301051519662e-06,
+      "loss": 0.0002,
+      "step": 16752
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.900609785374755e-06,
+      "loss": 0.0002,
+      "step": 16753
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.898209055552889e-06,
+      "loss": 0.0004,
+      "step": 16754
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.895808325731022e-06,
+      "loss": 0.0007,
+      "step": 16755
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.893407595909157e-06,
+      "loss": 0.0008,
+      "step": 16756
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.891006866087291e-06,
+      "loss": 0.609,
+      "step": 16757
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.888606136265426e-06,
+      "loss": 0.0002,
+      "step": 16758
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.88620540644356e-06,
+      "loss": 0.0002,
+      "step": 16759
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.883804676621695e-06,
+      "loss": 0.0002,
+      "step": 16760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.881403946799828e-06,
+      "loss": 0.0009,
+      "step": 16761
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.879003216977962e-06,
+      "loss": 0.0003,
+      "step": 16762
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.876602487156095e-06,
+      "loss": 0.0007,
+      "step": 16763
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.87420175733423e-06,
+      "loss": 0.0002,
+      "step": 16764
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.871801027512364e-06,
+      "loss": 0.0007,
+      "step": 16765
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.869400297690497e-06,
+      "loss": 0.0002,
+      "step": 16766
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.866999567868633e-06,
+      "loss": 0.0002,
+      "step": 16767
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.864598838046766e-06,
+      "loss": 0.0008,
+      "step": 16768
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.862198108224901e-06,
+      "loss": 0.0007,
+      "step": 16769
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.859797378403035e-06,
+      "loss": 0.0009,
+      "step": 16770
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.857396648581168e-06,
+      "loss": 0.0016,
+      "step": 16771
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.854995918759304e-06,
+      "loss": 0.0002,
+      "step": 16772
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.852595188937437e-06,
+      "loss": 0.0002,
+      "step": 16773
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.850194459115572e-06,
+      "loss": 0.0002,
+      "step": 16774
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.847793729293706e-06,
+      "loss": 0.0002,
+      "step": 16775
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.845392999471841e-06,
+      "loss": 0.0002,
+      "step": 16776
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.842992269649974e-06,
+      "loss": 0.0003,
+      "step": 16777
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.840591539828108e-06,
+      "loss": 0.5244,
+      "step": 16778
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.838190810006243e-06,
+      "loss": 0.0006,
+      "step": 16779
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.835790080184377e-06,
+      "loss": 0.0002,
+      "step": 16780
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.833389350362512e-06,
+      "loss": 0.0001,
+      "step": 16781
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.830988620540645e-06,
+      "loss": 0.0002,
+      "step": 16782
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.828587890718779e-06,
+      "loss": 0.0003,
+      "step": 16783
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.826187160896912e-06,
+      "loss": 0.0002,
+      "step": 16784
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.823786431075047e-06,
+      "loss": 0.0003,
+      "step": 16785
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.821385701253181e-06,
+      "loss": 0.0002,
+      "step": 16786
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.818984971431314e-06,
+      "loss": 0.0004,
+      "step": 16787
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.81658424160945e-06,
+      "loss": 0.0002,
+      "step": 16788
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.814183511787583e-06,
+      "loss": 0.0014,
+      "step": 16789
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.811782781965718e-06,
+      "loss": 0.0001,
+      "step": 16790
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.809382052143852e-06,
+      "loss": 0.0012,
+      "step": 16791
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.806981322321987e-06,
+      "loss": 0.0002,
+      "step": 16792
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.80458059250012e-06,
+      "loss": 0.0003,
+      "step": 16793
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.802179862678254e-06,
+      "loss": 0.0013,
+      "step": 16794
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.799779132856389e-06,
+      "loss": 0.0002,
+      "step": 16795
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.797378403034523e-06,
+      "loss": 0.0003,
+      "step": 16796
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.794977673212658e-06,
+      "loss": 0.0002,
+      "step": 16797
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.792576943390791e-06,
+      "loss": 0.0002,
+      "step": 16798
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.790176213568927e-06,
+      "loss": 0.0019,
+      "step": 16799
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.78777548374706e-06,
+      "loss": 0.0003,
+      "step": 16800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.785374753925193e-06,
+      "loss": 0.0003,
+      "step": 16801
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.782974024103329e-06,
+      "loss": 0.0007,
+      "step": 16802
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.780573294281462e-06,
+      "loss": 0.0003,
+      "step": 16803
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.778172564459596e-06,
+      "loss": 0.0006,
+      "step": 16804
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.77577183463773e-06,
+      "loss": 0.9905,
+      "step": 16805
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.773371104815864e-06,
+      "loss": 0.0007,
+      "step": 16806
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.770970374993998e-06,
+      "loss": 0.0002,
+      "step": 16807
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.768569645172133e-06,
+      "loss": 0.0002,
+      "step": 16808
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.766168915350267e-06,
+      "loss": 0.0003,
+      "step": 16809
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.7637681855284e-06,
+      "loss": 0.0009,
+      "step": 16810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.761367455706535e-06,
+      "loss": 0.5964,
+      "step": 16811
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.758966725884669e-06,
+      "loss": 0.0002,
+      "step": 16812
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.756565996062804e-06,
+      "loss": 0.0002,
+      "step": 16813
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.754165266240937e-06,
+      "loss": 0.0002,
+      "step": 16814
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.751764536419073e-06,
+      "loss": 0.0011,
+      "step": 16815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.749363806597206e-06,
+      "loss": 0.0002,
+      "step": 16816
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.74696307677534e-06,
+      "loss": 0.0002,
+      "step": 16817
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.744562346953475e-06,
+      "loss": 0.0002,
+      "step": 16818
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.742161617131608e-06,
+      "loss": 0.0002,
+      "step": 16819
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.739760887309743e-06,
+      "loss": 0.0002,
+      "step": 16820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.737360157487877e-06,
+      "loss": 0.0011,
+      "step": 16821
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.734959427666012e-06,
+      "loss": 0.0003,
+      "step": 16822
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.732558697844146e-06,
+      "loss": 0.0004,
+      "step": 16823
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.730157968022279e-06,
+      "loss": 0.001,
+      "step": 16824
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.727757238200413e-06,
+      "loss": 0.0002,
+      "step": 16825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.725356508378548e-06,
+      "loss": 0.0012,
+      "step": 16826
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.722955778556681e-06,
+      "loss": 0.0003,
+      "step": 16827
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.720555048734815e-06,
+      "loss": 0.0001,
+      "step": 16828
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.71815431891295e-06,
+      "loss": 0.0002,
+      "step": 16829
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.715753589091083e-06,
+      "loss": 0.0001,
+      "step": 16830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.713352859269219e-06,
+      "loss": 0.0003,
+      "step": 16831
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.710952129447352e-06,
+      "loss": 0.0002,
+      "step": 16832
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.708551399625486e-06,
+      "loss": 0.0002,
+      "step": 16833
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.70615066980362e-06,
+      "loss": 0.0003,
+      "step": 16834
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.703749939981754e-06,
+      "loss": 0.0003,
+      "step": 16835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.70134921015989e-06,
+      "loss": 0.0007,
+      "step": 16836
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.698948480338023e-06,
+      "loss": 0.0002,
+      "step": 16837
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.696547750516158e-06,
+      "loss": 0.0002,
+      "step": 16838
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.694147020694292e-06,
+      "loss": 0.0002,
+      "step": 16839
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.691746290872427e-06,
+      "loss": 0.0001,
+      "step": 16840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.68934556105056e-06,
+      "loss": 0.0006,
+      "step": 16841
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.686944831228694e-06,
+      "loss": 0.0007,
+      "step": 16842
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.684544101406829e-06,
+      "loss": 0.0008,
+      "step": 16843
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.682143371584963e-06,
+      "loss": 0.0002,
+      "step": 16844
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.679742641763096e-06,
+      "loss": 0.0002,
+      "step": 16845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.67734191194123e-06,
+      "loss": 0.0005,
+      "step": 16846
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.674941182119365e-06,
+      "loss": 0.0003,
+      "step": 16847
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.672540452297498e-06,
+      "loss": 0.001,
+      "step": 16848
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.670139722475632e-06,
+      "loss": 0.0003,
+      "step": 16849
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.667738992653767e-06,
+      "loss": 0.0002,
+      "step": 16850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.6653382628319e-06,
+      "loss": 0.0002,
+      "step": 16851
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.662937533010036e-06,
+      "loss": 0.0003,
+      "step": 16852
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.660536803188169e-06,
+      "loss": 0.0002,
+      "step": 16853
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.658136073366304e-06,
+      "loss": 0.0002,
+      "step": 16854
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.655735343544438e-06,
+      "loss": 0.0007,
+      "step": 16855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.653334613722573e-06,
+      "loss": 0.0002,
+      "step": 16856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.650933883900706e-06,
+      "loss": 0.0003,
+      "step": 16857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.64853315407884e-06,
+      "loss": 0.0002,
+      "step": 16858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.646132424256975e-06,
+      "loss": 0.0001,
+      "step": 16859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.643731694435109e-06,
+      "loss": 0.0013,
+      "step": 16860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.641330964613244e-06,
+      "loss": 0.0011,
+      "step": 16861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.638930234791377e-06,
+      "loss": 0.0002,
+      "step": 16862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.636529504969512e-06,
+      "loss": 0.0007,
+      "step": 16863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.634128775147646e-06,
+      "loss": 0.0003,
+      "step": 16864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.63172804532578e-06,
+      "loss": 0.0002,
+      "step": 16865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.629327315503913e-06,
+      "loss": 0.0002,
+      "step": 16866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.626926585682046e-06,
+      "loss": 0.0007,
+      "step": 16867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.624525855860182e-06,
+      "loss": 0.0007,
+      "step": 16868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.622125126038315e-06,
+      "loss": 0.0005,
+      "step": 16869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.61972439621645e-06,
+      "loss": 0.0006,
+      "step": 16870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.617323666394584e-06,
+      "loss": 0.0007,
+      "step": 16871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.614922936572719e-06,
+      "loss": 0.0002,
+      "step": 16872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.612522206750853e-06,
+      "loss": 0.0017,
+      "step": 16873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.610121476928986e-06,
+      "loss": 0.0005,
+      "step": 16874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.607720747107121e-06,
+      "loss": 0.0002,
+      "step": 16875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.605320017285255e-06,
+      "loss": 0.0002,
+      "step": 16876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.60291928746339e-06,
+      "loss": 0.0003,
+      "step": 16877
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.600518557641523e-06,
+      "loss": 0.0006,
+      "step": 16878
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.598117827819659e-06,
+      "loss": 0.0002,
+      "step": 16879
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.595717097997792e-06,
+      "loss": 0.0002,
+      "step": 16880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.593316368175926e-06,
+      "loss": 0.0018,
+      "step": 16881
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.59091563835406e-06,
+      "loss": 0.0002,
+      "step": 16882
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.588514908532194e-06,
+      "loss": 0.0002,
+      "step": 16883
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.58611417871033e-06,
+      "loss": 0.001,
+      "step": 16884
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.583713448888463e-06,
+      "loss": 0.0002,
+      "step": 16885
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.581312719066596e-06,
+      "loss": 0.0002,
+      "step": 16886
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.57891198924473e-06,
+      "loss": 0.0011,
+      "step": 16887
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.576511259422865e-06,
+      "loss": 0.0007,
+      "step": 16888
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.574110529600999e-06,
+      "loss": 0.0006,
+      "step": 16889
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.571709799779132e-06,
+      "loss": 0.0007,
+      "step": 16890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.569309069957267e-06,
+      "loss": 0.0007,
+      "step": 16891
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.5669083401354e-06,
+      "loss": 0.0008,
+      "step": 16892
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.564507610313536e-06,
+      "loss": 0.0002,
+      "step": 16893
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.56210688049167e-06,
+      "loss": 0.7135,
+      "step": 16894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.559706150669805e-06,
+      "loss": 0.0002,
+      "step": 16895
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.557305420847938e-06,
+      "loss": 0.0006,
+      "step": 16896
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.554904691026072e-06,
+      "loss": 0.0004,
+      "step": 16897
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.552503961204207e-06,
+      "loss": 0.0003,
+      "step": 16898
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.55010323138234e-06,
+      "loss": 0.0007,
+      "step": 16899
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.547702501560476e-06,
+      "loss": 0.0006,
+      "step": 16900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.545301771738609e-06,
+      "loss": 0.6751,
+      "step": 16901
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.542901041916744e-06,
+      "loss": 0.0006,
+      "step": 16902
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.540500312094878e-06,
+      "loss": 0.0002,
+      "step": 16903
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.538099582273013e-06,
+      "loss": 0.0002,
+      "step": 16904
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.535698852451146e-06,
+      "loss": 0.0006,
+      "step": 16905
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.53329812262928e-06,
+      "loss": 0.0002,
+      "step": 16906
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.530897392807413e-06,
+      "loss": 0.0002,
+      "step": 16907
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.528496662985547e-06,
+      "loss": 0.0003,
+      "step": 16908
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.526095933163682e-06,
+      "loss": 0.0002,
+      "step": 16909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.523695203341816e-06,
+      "loss": 0.0007,
+      "step": 16910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.52129447351995e-06,
+      "loss": 0.0001,
+      "step": 16911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.518893743698084e-06,
+      "loss": 0.001,
+      "step": 16912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.516493013876218e-06,
+      "loss": 0.0008,
+      "step": 16913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514092284054353e-06,
+      "loss": 0.0001,
+      "step": 16914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.511691554232486e-06,
+      "loss": 0.0008,
+      "step": 16915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.509290824410622e-06,
+      "loss": 0.0002,
+      "step": 16916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.506890094588755e-06,
+      "loss": 0.0001,
+      "step": 16917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.50448936476689e-06,
+      "loss": 0.0001,
+      "step": 16918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.502088634945024e-06,
+      "loss": 0.0002,
+      "step": 16919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.499687905123159e-06,
+      "loss": 0.0002,
+      "step": 16920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.497287175301292e-06,
+      "loss": 0.0014,
+      "step": 16921
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.494886445479426e-06,
+      "loss": 0.0002,
+      "step": 16922
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.492485715657561e-06,
+      "loss": 0.0006,
+      "step": 16923
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.490084985835695e-06,
+      "loss": 0.0002,
+      "step": 16924
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.48768425601383e-06,
+      "loss": 0.0002,
+      "step": 16925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.485283526191963e-06,
+      "loss": 0.0008,
+      "step": 16926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.482882796370097e-06,
+      "loss": 0.0028,
+      "step": 16927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.48048206654823e-06,
+      "loss": 0.0002,
+      "step": 16928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.478081336726364e-06,
+      "loss": 0.0001,
+      "step": 16929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.475680606904499e-06,
+      "loss": 0.0002,
+      "step": 16930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.473279877082632e-06,
+      "loss": 0.0002,
+      "step": 16931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.470879147260768e-06,
+      "loss": 0.0002,
+      "step": 16932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.468478417438901e-06,
+      "loss": 0.0006,
+      "step": 16933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.466077687617036e-06,
+      "loss": 0.0002,
+      "step": 16934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.46367695779517e-06,
+      "loss": 0.0002,
+      "step": 16935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.461276227973305e-06,
+      "loss": 0.0002,
+      "step": 16936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.458875498151439e-06,
+      "loss": 0.0001,
+      "step": 16937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.456474768329572e-06,
+      "loss": 0.0002,
+      "step": 16938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.454074038507707e-06,
+      "loss": 0.0001,
+      "step": 16939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.45167330868584e-06,
+      "loss": 0.0009,
+      "step": 16940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.449272578863976e-06,
+      "loss": 0.0001,
+      "step": 16941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.44687184904211e-06,
+      "loss": 0.0002,
+      "step": 16942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.444471119220245e-06,
+      "loss": 0.0002,
+      "step": 16943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.442070389398378e-06,
+      "loss": 0.0002,
+      "step": 16944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.439669659576512e-06,
+      "loss": 0.0009,
+      "step": 16945
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.437268929754647e-06,
+      "loss": 0.0003,
+      "step": 16946
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.43486819993278e-06,
+      "loss": 0.0003,
+      "step": 16947
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.432467470110914e-06,
+      "loss": 0.0006,
+      "step": 16948
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.430066740289047e-06,
+      "loss": 0.0001,
+      "step": 16949
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.427666010467182e-06,
+      "loss": 0.0002,
+      "step": 16950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.425265280645316e-06,
+      "loss": 0.0007,
+      "step": 16951
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.422864550823451e-06,
+      "loss": 0.0002,
+      "step": 16952
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.420463821001585e-06,
+      "loss": 0.0002,
+      "step": 16953
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.418063091179718e-06,
+      "loss": 0.0002,
+      "step": 16954
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.415662361357853e-06,
+      "loss": 0.0007,
+      "step": 16955
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.413261631535987e-06,
+      "loss": 0.0003,
+      "step": 16956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.410860901714122e-06,
+      "loss": 0.0001,
+      "step": 16957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.408460171892255e-06,
+      "loss": 0.0008,
+      "step": 16958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.40605944207039e-06,
+      "loss": 0.0017,
+      "step": 16959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.403658712248524e-06,
+      "loss": 0.0002,
+      "step": 16960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.401257982426658e-06,
+      "loss": 0.6775,
+      "step": 16961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.398857252604793e-06,
+      "loss": 0.0002,
+      "step": 16962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.396456522782926e-06,
+      "loss": 0.0002,
+      "step": 16963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.394055792961061e-06,
+      "loss": 0.0002,
+      "step": 16964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.391655063139195e-06,
+      "loss": 0.0002,
+      "step": 16965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.38925433331733e-06,
+      "loss": 0.0007,
+      "step": 16966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.386853603495464e-06,
+      "loss": 0.5694,
+      "step": 16967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.384452873673597e-06,
+      "loss": 0.0001,
+      "step": 16968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.38205214385173e-06,
+      "loss": 0.0002,
+      "step": 16969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.379651414029864e-06,
+      "loss": 0.0002,
+      "step": 16970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.377250684208e-06,
+      "loss": 0.0001,
+      "step": 16971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.374849954386133e-06,
+      "loss": 0.0008,
+      "step": 16972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.372449224564268e-06,
+      "loss": 0.0003,
+      "step": 16973
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.370048494742402e-06,
+      "loss": 0.0014,
+      "step": 16974
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.367647764920537e-06,
+      "loss": 0.0002,
+      "step": 16975
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.36524703509867e-06,
+      "loss": 0.0002,
+      "step": 16976
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.362846305276804e-06,
+      "loss": 0.0002,
+      "step": 16977
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.360445575454939e-06,
+      "loss": 0.0002,
+      "step": 16978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.358044845633072e-06,
+      "loss": 0.0026,
+      "step": 16979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.355644115811208e-06,
+      "loss": 0.0002,
+      "step": 16980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.353243385989341e-06,
+      "loss": 0.0002,
+      "step": 16981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.350842656167476e-06,
+      "loss": 0.0011,
+      "step": 16982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.34844192634561e-06,
+      "loss": 0.0002,
+      "step": 16983
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.346041196523745e-06,
+      "loss": 0.0002,
+      "step": 16984
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.343640466701878e-06,
+      "loss": 0.001,
+      "step": 16985
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.341239736880012e-06,
+      "loss": 0.0002,
+      "step": 16986
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.338839007058147e-06,
+      "loss": 0.0002,
+      "step": 16987
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.33643827723628e-06,
+      "loss": 0.0002,
+      "step": 16988
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.334037547414414e-06,
+      "loss": 0.0024,
+      "step": 16989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.331636817592548e-06,
+      "loss": 0.0002,
+      "step": 16990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.329236087770683e-06,
+      "loss": 0.0002,
+      "step": 16991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.326835357948816e-06,
+      "loss": 0.0002,
+      "step": 16992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.32443462812695e-06,
+      "loss": 0.0002,
+      "step": 16993
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.322033898305085e-06,
+      "loss": 0.001,
+      "step": 16994
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.319633168483218e-06,
+      "loss": 0.0001,
+      "step": 16995
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.317232438661354e-06,
+      "loss": 0.0001,
+      "step": 16996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.314831708839487e-06,
+      "loss": 0.0013,
+      "step": 16997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.312430979017622e-06,
+      "loss": 0.0002,
+      "step": 16998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.310030249195756e-06,
+      "loss": 0.0001,
+      "step": 16999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.307629519373891e-06,
+      "loss": 0.0003,
+      "step": 17000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.305228789552025e-06,
+      "loss": 0.0013,
+      "step": 17001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.302828059730158e-06,
+      "loss": 0.0001,
+      "step": 17002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.300427329908293e-06,
+      "loss": 0.0039,
+      "step": 17003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.298026600086427e-06,
+      "loss": 0.0002,
+      "step": 17004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.295625870264562e-06,
+      "loss": 0.0002,
+      "step": 17005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.293225140442695e-06,
+      "loss": 0.0001,
+      "step": 17006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.29082441062083e-06,
+      "loss": 0.0002,
+      "step": 17007
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.288423680798964e-06,
+      "loss": 0.0002,
+      "step": 17008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.286022950977098e-06,
+      "loss": 0.0002,
+      "step": 17009
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.283622221155231e-06,
+      "loss": 0.0002,
+      "step": 17010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.281221491333365e-06,
+      "loss": 0.0002,
+      "step": 17011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.2788207615115e-06,
+      "loss": 0.0008,
+      "step": 17012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.276420031689633e-06,
+      "loss": 0.0001,
+      "step": 17013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.274019301867768e-06,
+      "loss": 0.0029,
+      "step": 17014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.271618572045902e-06,
+      "loss": 0.0004,
+      "step": 17015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.269217842224037e-06,
+      "loss": 0.0002,
+      "step": 17016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.26681711240217e-06,
+      "loss": 0.0001,
+      "step": 17017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.264416382580304e-06,
+      "loss": 0.0002,
+      "step": 17018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.26201565275844e-06,
+      "loss": 0.0002,
+      "step": 17019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.259614922936573e-06,
+      "loss": 0.0002,
+      "step": 17020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.257214193114708e-06,
+      "loss": 0.0001,
+      "step": 17021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.254813463292841e-06,
+      "loss": 0.0009,
+      "step": 17022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.252412733470977e-06,
+      "loss": 0.0018,
+      "step": 17023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.25001200364911e-06,
+      "loss": 0.0002,
+      "step": 17024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.247611273827244e-06,
+      "loss": 0.0002,
+      "step": 17025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.245210544005379e-06,
+      "loss": 0.0001,
+      "step": 17026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.242809814183512e-06,
+      "loss": 0.0002,
+      "step": 17027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.240409084361647e-06,
+      "loss": 0.0001,
+      "step": 17028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.238008354539781e-06,
+      "loss": 0.0002,
+      "step": 17029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.235607624717914e-06,
+      "loss": 0.0002,
+      "step": 17030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.233206894896048e-06,
+      "loss": 0.0011,
+      "step": 17031
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.230806165074183e-06,
+      "loss": 0.0002,
+      "step": 17032
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.228405435252317e-06,
+      "loss": 0.0011,
+      "step": 17033
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.22600470543045e-06,
+      "loss": 0.0026,
+      "step": 17034
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.223603975608585e-06,
+      "loss": 0.0011,
+      "step": 17035
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.221203245786719e-06,
+      "loss": 0.0007,
+      "step": 17036
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.218802515964854e-06,
+      "loss": 0.0009,
+      "step": 17037
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.216401786142988e-06,
+      "loss": 0.0013,
+      "step": 17038
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.214001056321123e-06,
+      "loss": 0.0002,
+      "step": 17039
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.211600326499256e-06,
+      "loss": 0.0016,
+      "step": 17040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.20919959667739e-06,
+      "loss": 0.0001,
+      "step": 17041
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.206798866855525e-06,
+      "loss": 0.0007,
+      "step": 17042
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.204398137033658e-06,
+      "loss": 0.0002,
+      "step": 17043
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.201997407211794e-06,
+      "loss": 0.0001,
+      "step": 17044
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.199596677389927e-06,
+      "loss": 0.001,
+      "step": 17045
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.197195947568062e-06,
+      "loss": 0.0002,
+      "step": 17046
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.194795217746196e-06,
+      "loss": 0.0002,
+      "step": 17047
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.192394487924331e-06,
+      "loss": 0.0001,
+      "step": 17048
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.189993758102464e-06,
+      "loss": 0.5829,
+      "step": 17049
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.187593028280598e-06,
+      "loss": 0.0002,
+      "step": 17050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.185192298458731e-06,
+      "loss": 0.0002,
+      "step": 17051
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.182791568636865e-06,
+      "loss": 0.0011,
+      "step": 17052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.180390838815e-06,
+      "loss": 0.0001,
+      "step": 17053
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.177990108993134e-06,
+      "loss": 0.0002,
+      "step": 17054
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.175589379171269e-06,
+      "loss": 0.0002,
+      "step": 17055
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.173188649349402e-06,
+      "loss": 0.0001,
+      "step": 17056
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.170787919527536e-06,
+      "loss": 0.0002,
+      "step": 17057
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.168387189705671e-06,
+      "loss": 0.0002,
+      "step": 17058
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.165986459883804e-06,
+      "loss": 0.0002,
+      "step": 17059
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.16358573006194e-06,
+      "loss": 0.0002,
+      "step": 17060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.161185000240073e-06,
+      "loss": 0.0008,
+      "step": 17061
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.158784270418208e-06,
+      "loss": 0.0014,
+      "step": 17062
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.156383540596342e-06,
+      "loss": 0.0002,
+      "step": 17063
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.153982810774477e-06,
+      "loss": 0.0002,
+      "step": 17064
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.15158208095261e-06,
+      "loss": 0.0002,
+      "step": 17065
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.149181351130744e-06,
+      "loss": 0.0002,
+      "step": 17066
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.14678062130888e-06,
+      "loss": 0.0002,
+      "step": 17067
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.144379891487013e-06,
+      "loss": 0.0001,
+      "step": 17068
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.141979161665148e-06,
+      "loss": 0.0002,
+      "step": 17069
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.139578431843281e-06,
+      "loss": 0.0002,
+      "step": 17070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.137177702021415e-06,
+      "loss": 0.0002,
+      "step": 17071
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.134776972199548e-06,
+      "loss": 0.0002,
+      "step": 17072
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.132376242377682e-06,
+      "loss": 0.0002,
+      "step": 17073
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.129975512555817e-06,
+      "loss": 0.0001,
+      "step": 17074
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.12757478273395e-06,
+      "loss": 0.0002,
+      "step": 17075
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.125174052912086e-06,
+      "loss": 0.0002,
+      "step": 17076
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.12277332309022e-06,
+      "loss": 0.0002,
+      "step": 17077
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.120372593268354e-06,
+      "loss": 0.0001,
+      "step": 17078
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.117971863446488e-06,
+      "loss": 0.0002,
+      "step": 17079
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.115571133624623e-06,
+      "loss": 0.0002,
+      "step": 17080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.113170403802757e-06,
+      "loss": 0.0001,
+      "step": 17081
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.11076967398089e-06,
+      "loss": 0.0002,
+      "step": 17082
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.108368944159025e-06,
+      "loss": 0.0002,
+      "step": 17083
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.105968214337159e-06,
+      "loss": 0.0002,
+      "step": 17084
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.103567484515294e-06,
+      "loss": 0.0001,
+      "step": 17085
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.101166754693427e-06,
+      "loss": 0.0002,
+      "step": 17086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.098766024871563e-06,
+      "loss": 0.0001,
+      "step": 17087
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.096365295049696e-06,
+      "loss": 0.0001,
+      "step": 17088
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.09396456522783e-06,
+      "loss": 0.7381,
+      "step": 17089
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.091563835405965e-06,
+      "loss": 0.0004,
+      "step": 17090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.089163105584098e-06,
+      "loss": 0.0002,
+      "step": 17091
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.086762375762232e-06,
+      "loss": 0.0001,
+      "step": 17092
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.084361645940365e-06,
+      "loss": 0.0001,
+      "step": 17093
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.0819609161185e-06,
+      "loss": 0.0002,
+      "step": 17094
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.079560186296634e-06,
+      "loss": 0.0002,
+      "step": 17095
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.077159456474769e-06,
+      "loss": 0.0002,
+      "step": 17096
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.074758726652903e-06,
+      "loss": 0.0002,
+      "step": 17097
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.072357996831036e-06,
+      "loss": 0.0002,
+      "step": 17098
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.069957267009171e-06,
+      "loss": 0.0002,
+      "step": 17099
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.067556537187305e-06,
+      "loss": 0.0002,
+      "step": 17100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.06515580736544e-06,
+      "loss": 0.0002,
+      "step": 17101
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.062755077543574e-06,
+      "loss": 0.0002,
+      "step": 17102
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.060354347721709e-06,
+      "loss": 0.0002,
+      "step": 17103
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.057953617899842e-06,
+      "loss": 0.0002,
+      "step": 17104
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.055552888077976e-06,
+      "loss": 0.0003,
+      "step": 17105
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.053152158256111e-06,
+      "loss": 0.0002,
+      "step": 17106
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.050751428434244e-06,
+      "loss": 0.0003,
+      "step": 17107
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.04835069861238e-06,
+      "loss": 0.0002,
+      "step": 17108
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.045949968790513e-06,
+      "loss": 0.0001,
+      "step": 17109
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.043549238968648e-06,
+      "loss": 0.0002,
+      "step": 17110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.041148509146782e-06,
+      "loss": 0.0002,
+      "step": 17111
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.038747779324915e-06,
+      "loss": 0.0002,
+      "step": 17112
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.036347049503049e-06,
+      "loss": 0.0001,
+      "step": 17113
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.033946319681182e-06,
+      "loss": 0.0001,
+      "step": 17114
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.031545589859317e-06,
+      "loss": 0.0002,
+      "step": 17115
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.029144860037451e-06,
+      "loss": 0.0001,
+      "step": 17116
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.026744130215586e-06,
+      "loss": 0.0002,
+      "step": 17117
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.02434340039372e-06,
+      "loss": 0.0002,
+      "step": 17118
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.021942670571855e-06,
+      "loss": 0.0001,
+      "step": 17119
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.019541940749988e-06,
+      "loss": 0.0001,
+      "step": 17120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.017141210928122e-06,
+      "loss": 0.0002,
+      "step": 17121
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.014740481106257e-06,
+      "loss": 0.0001,
+      "step": 17122
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.01233975128439e-06,
+      "loss": 0.0001,
+      "step": 17123
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.009939021462526e-06,
+      "loss": 0.0002,
+      "step": 17124
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.007538291640659e-06,
+      "loss": 0.0001,
+      "step": 17125
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.005137561818794e-06,
+      "loss": 0.0002,
+      "step": 17126
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.002736831996928e-06,
+      "loss": 0.0001,
+      "step": 17127
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.000336102175061e-06,
+      "loss": 0.0002,
+      "step": 17128
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.997935372353196e-06,
+      "loss": 0.0001,
+      "step": 17129
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.99553464253133e-06,
+      "loss": 0.0002,
+      "step": 17130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.993133912709465e-06,
+      "loss": 0.0002,
+      "step": 17131
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.990733182887599e-06,
+      "loss": 0.0002,
+      "step": 17132
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.988332453065732e-06,
+      "loss": 0.0001,
+      "step": 17133
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.985931723243866e-06,
+      "loss": 0.0002,
+      "step": 17134
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.983530993422e-06,
+      "loss": 0.0001,
+      "step": 17135
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.981130263600134e-06,
+      "loss": 0.0002,
+      "step": 17136
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.978729533778268e-06,
+      "loss": 0.0002,
+      "step": 17137
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.976328803956403e-06,
+      "loss": 0.0001,
+      "step": 17138
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.973928074134537e-06,
+      "loss": 0.0002,
+      "step": 17139
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.971527344312672e-06,
+      "loss": 0.0001,
+      "step": 17140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.969126614490805e-06,
+      "loss": 0.0002,
+      "step": 17141
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.96672588466894e-06,
+      "loss": 0.0002,
+      "step": 17142
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.964325154847074e-06,
+      "loss": 0.0001,
+      "step": 17143
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.961924425025207e-06,
+      "loss": 0.0025,
+      "step": 17144
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.959523695203343e-06,
+      "loss": 0.0002,
+      "step": 17145
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.957122965381476e-06,
+      "loss": 0.0007,
+      "step": 17146
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.954722235559611e-06,
+      "loss": 0.0002,
+      "step": 17147
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.952321505737745e-06,
+      "loss": 0.0002,
+      "step": 17148
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.94992077591588e-06,
+      "loss": 0.0001,
+      "step": 17149
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.947520046094013e-06,
+      "loss": 0.0002,
+      "step": 17150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.945119316272149e-06,
+      "loss": 0.0001,
+      "step": 17151
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.942718586450282e-06,
+      "loss": 0.0001,
+      "step": 17152
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.940317856628416e-06,
+      "loss": 0.0007,
+      "step": 17153
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.937917126806549e-06,
+      "loss": 0.0001,
+      "step": 17154
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.935516396984683e-06,
+      "loss": 0.0001,
+      "step": 17155
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.933115667162818e-06,
+      "loss": 0.0002,
+      "step": 17156
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.930714937340951e-06,
+      "loss": 0.0002,
+      "step": 17157
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.928314207519086e-06,
+      "loss": 0.0001,
+      "step": 17158
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.92591347769722e-06,
+      "loss": 0.0001,
+      "step": 17159
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.923512747875353e-06,
+      "loss": 0.0002,
+      "step": 17160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.921112018053489e-06,
+      "loss": 0.0002,
+      "step": 17161
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.918711288231622e-06,
+      "loss": 0.0002,
+      "step": 17162
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.916310558409757e-06,
+      "loss": 1.0095,
+      "step": 17163
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.91390982858789e-06,
+      "loss": 0.0002,
+      "step": 17164
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.911509098766026e-06,
+      "loss": 0.0002,
+      "step": 17165
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.90910836894416e-06,
+      "loss": 0.0001,
+      "step": 17166
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.906707639122295e-06,
+      "loss": 0.0001,
+      "step": 17167
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.904306909300428e-06,
+      "loss": 0.0002,
+      "step": 17168
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.901906179478562e-06,
+      "loss": 0.0001,
+      "step": 17169
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.899505449656697e-06,
+      "loss": 1.1773,
+      "step": 17170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.89710471983483e-06,
+      "loss": 0.0002,
+      "step": 17171
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.894703990012966e-06,
+      "loss": 0.0001,
+      "step": 17172
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.892303260191099e-06,
+      "loss": 0.0002,
+      "step": 17173
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.889902530369233e-06,
+      "loss": 0.0002,
+      "step": 17174
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.887501800547366e-06,
+      "loss": 0.0002,
+      "step": 17175
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.8851010707255e-06,
+      "loss": 0.0002,
+      "step": 17176
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.882700340903635e-06,
+      "loss": 0.0001,
+      "step": 17177
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.880299611081768e-06,
+      "loss": 0.0002,
+      "step": 17178
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.877898881259903e-06,
+      "loss": 0.0002,
+      "step": 17179
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.875498151438037e-06,
+      "loss": 0.0001,
+      "step": 17180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.873097421616172e-06,
+      "loss": 0.0001,
+      "step": 17181
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.870696691794306e-06,
+      "loss": 0.0002,
+      "step": 17182
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.86829596197244e-06,
+      "loss": 0.0002,
+      "step": 17183
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.865895232150574e-06,
+      "loss": 0.0002,
+      "step": 17184
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.863494502328708e-06,
+      "loss": 0.0002,
+      "step": 17185
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.861093772506843e-06,
+      "loss": 0.0002,
+      "step": 17186
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.858693042684976e-06,
+      "loss": 0.0002,
+      "step": 17187
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.856292312863112e-06,
+      "loss": 0.0002,
+      "step": 17188
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.853891583041245e-06,
+      "loss": 0.0003,
+      "step": 17189
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.85149085321938e-06,
+      "loss": 1.0797,
+      "step": 17190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.849090123397514e-06,
+      "loss": 0.0003,
+      "step": 17191
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.846689393575647e-06,
+      "loss": 0.0002,
+      "step": 17192
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.844288663753782e-06,
+      "loss": 0.0002,
+      "step": 17193
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.841887933931916e-06,
+      "loss": 0.0002,
+      "step": 17194
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.83948720411005e-06,
+      "loss": 0.0029,
+      "step": 17195
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.837086474288183e-06,
+      "loss": 0.0002,
+      "step": 17196
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.834685744466318e-06,
+      "loss": 0.0003,
+      "step": 17197
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.832285014644452e-06,
+      "loss": 0.0003,
+      "step": 17198
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.829884284822587e-06,
+      "loss": 0.0003,
+      "step": 17199
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.82748355500072e-06,
+      "loss": 0.0002,
+      "step": 17200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.825082825178854e-06,
+      "loss": 0.5129,
+      "step": 17201
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.822682095356989e-06,
+      "loss": 0.0002,
+      "step": 17202
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.820281365535123e-06,
+      "loss": 0.0003,
+      "step": 17203
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.817880635713258e-06,
+      "loss": 0.0002,
+      "step": 17204
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.815479905891391e-06,
+      "loss": 0.0002,
+      "step": 17205
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.813079176069526e-06,
+      "loss": 0.0003,
+      "step": 17206
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.81067844624766e-06,
+      "loss": 0.0002,
+      "step": 17207
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.808277716425793e-06,
+      "loss": 0.0003,
+      "step": 17208
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.805876986603929e-06,
+      "loss": 0.0002,
+      "step": 17209
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.803476256782062e-06,
+      "loss": 0.0003,
+      "step": 17210
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.801075526960197e-06,
+      "loss": 0.0003,
+      "step": 17211
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.79867479713833e-06,
+      "loss": 0.0003,
+      "step": 17212
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.796274067316466e-06,
+      "loss": 0.0003,
+      "step": 17213
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.7938733374946e-06,
+      "loss": 0.0003,
+      "step": 17214
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.791472607672733e-06,
+      "loss": 0.0002,
+      "step": 17215
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.789071877850866e-06,
+      "loss": 0.0003,
+      "step": 17216
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.786671148029e-06,
+      "loss": 0.0003,
+      "step": 17217
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.784270418207135e-06,
+      "loss": 0.0003,
+      "step": 17218
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.781869688385269e-06,
+      "loss": 0.0002,
+      "step": 17219
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.779468958563404e-06,
+      "loss": 0.0003,
+      "step": 17220
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.777068228741537e-06,
+      "loss": 0.0002,
+      "step": 17221
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.774667498919672e-06,
+      "loss": 0.0003,
+      "step": 17222
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.772266769097806e-06,
+      "loss": 0.0002,
+      "step": 17223
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.76986603927594e-06,
+      "loss": 0.0002,
+      "step": 17224
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.767465309454075e-06,
+      "loss": 0.0002,
+      "step": 17225
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.765064579632208e-06,
+      "loss": 0.0003,
+      "step": 17226
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.762663849810343e-06,
+      "loss": 0.0003,
+      "step": 17227
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.760263119988477e-06,
+      "loss": 0.0002,
+      "step": 17228
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.757862390166612e-06,
+      "loss": 0.0003,
+      "step": 17229
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.755461660344745e-06,
+      "loss": 0.0002,
+      "step": 17230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.75306093052288e-06,
+      "loss": 0.0002,
+      "step": 17231
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.750660200701014e-06,
+      "loss": 0.0002,
+      "step": 17232
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.748259470879148e-06,
+      "loss": 0.0002,
+      "step": 17233
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.745858741057283e-06,
+      "loss": 0.0003,
+      "step": 17234
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.743458011235416e-06,
+      "loss": 0.0002,
+      "step": 17235
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.74105728141355e-06,
+      "loss": 0.0002,
+      "step": 17236
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.738656551591683e-06,
+      "loss": 0.0003,
+      "step": 17237
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.736255821769819e-06,
+      "loss": 0.0002,
+      "step": 17238
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.733855091947952e-06,
+      "loss": 0.0002,
+      "step": 17239
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.731454362126086e-06,
+      "loss": 0.0002,
+      "step": 17240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.72905363230422e-06,
+      "loss": 0.0003,
+      "step": 17241
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.726652902482354e-06,
+      "loss": 0.0003,
+      "step": 17242
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.72425217266049e-06,
+      "loss": 0.0003,
+      "step": 17243
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.721851442838623e-06,
+      "loss": 0.0007,
+      "step": 17244
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.719450713016758e-06,
+      "loss": 0.0002,
+      "step": 17245
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.717049983194892e-06,
+      "loss": 0.0003,
+      "step": 17246
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.714649253373027e-06,
+      "loss": 0.0003,
+      "step": 17247
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.71224852355116e-06,
+      "loss": 0.0002,
+      "step": 17248
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.709847793729294e-06,
+      "loss": 0.0003,
+      "step": 17249
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.707447063907429e-06,
+      "loss": 0.0002,
+      "step": 17250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.705046334085562e-06,
+      "loss": 0.0002,
+      "step": 17251
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.702645604263698e-06,
+      "loss": 0.0002,
+      "step": 17252
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.700244874441831e-06,
+      "loss": 0.0003,
+      "step": 17253
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.697844144619965e-06,
+      "loss": 0.0003,
+      "step": 17254
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.6954434147981e-06,
+      "loss": 0.0002,
+      "step": 17255
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.693042684976233e-06,
+      "loss": 0.0002,
+      "step": 17256
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.690641955154367e-06,
+      "loss": 0.0002,
+      "step": 17257
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.6882412253325e-06,
+      "loss": 0.0003,
+      "step": 17258
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.685840495510635e-06,
+      "loss": 0.0002,
+      "step": 17259
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.683439765688769e-06,
+      "loss": 0.0003,
+      "step": 17260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.681039035866904e-06,
+      "loss": 0.0002,
+      "step": 17261
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.678638306045038e-06,
+      "loss": 0.0003,
+      "step": 17262
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.676237576223173e-06,
+      "loss": 0.0002,
+      "step": 17263
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.673836846401306e-06,
+      "loss": 0.0002,
+      "step": 17264
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.67143611657944e-06,
+      "loss": 0.0002,
+      "step": 17265
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.669035386757575e-06,
+      "loss": 0.0002,
+      "step": 17266
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.666634656935709e-06,
+      "loss": 0.0003,
+      "step": 17267
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.664233927113844e-06,
+      "loss": 0.0002,
+      "step": 17268
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.661833197291977e-06,
+      "loss": 0.0003,
+      "step": 17269
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.659432467470112e-06,
+      "loss": 0.0002,
+      "step": 17270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.657031737648246e-06,
+      "loss": 0.0003,
+      "step": 17271
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.65463100782638e-06,
+      "loss": 0.0002,
+      "step": 17272
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.652230278004515e-06,
+      "loss": 0.0002,
+      "step": 17273
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.649829548182648e-06,
+      "loss": 0.0003,
+      "step": 17274
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.647428818360783e-06,
+      "loss": 0.0002,
+      "step": 17275
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.645028088538917e-06,
+      "loss": 0.0002,
+      "step": 17276
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.64262735871705e-06,
+      "loss": 0.0002,
+      "step": 17277
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.640226628895184e-06,
+      "loss": 0.0002,
+      "step": 17278
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.637825899073319e-06,
+      "loss": 0.0002,
+      "step": 17279
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.635425169251452e-06,
+      "loss": 0.0002,
+      "step": 17280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.633024439429586e-06,
+      "loss": 0.0002,
+      "step": 17281
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.630623709607721e-06,
+      "loss": 0.0002,
+      "step": 17282
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.628222979785855e-06,
+      "loss": 0.0003,
+      "step": 17283
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.62582224996399e-06,
+      "loss": 0.0002,
+      "step": 17284
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.623421520142123e-06,
+      "loss": 0.0002,
+      "step": 17285
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.621020790320258e-06,
+      "loss": 0.0002,
+      "step": 17286
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.618620060498392e-06,
+      "loss": 0.0002,
+      "step": 17287
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.616219330676525e-06,
+      "loss": 0.0002,
+      "step": 17288
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.61381860085466e-06,
+      "loss": 0.0002,
+      "step": 17289
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.611417871032794e-06,
+      "loss": 0.0002,
+      "step": 17290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.60901714121093e-06,
+      "loss": 0.0002,
+      "step": 17291
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.606616411389063e-06,
+      "loss": 0.0002,
+      "step": 17292
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.604215681567198e-06,
+      "loss": 0.0003,
+      "step": 17293
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.601814951745331e-06,
+      "loss": 0.0002,
+      "step": 17294
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.599414221923465e-06,
+      "loss": 0.0002,
+      "step": 17295
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.5970134921016e-06,
+      "loss": 0.0002,
+      "step": 17296
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.594612762279734e-06,
+      "loss": 0.0002,
+      "step": 17297
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.592212032457867e-06,
+      "loss": 0.0002,
+      "step": 17298
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.589811302636e-06,
+      "loss": 0.0002,
+      "step": 17299
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.587410572814136e-06,
+      "loss": 0.0065,
+      "step": 17300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.58500984299227e-06,
+      "loss": 0.322,
+      "step": 17301
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.582609113170405e-06,
+      "loss": 0.0003,
+      "step": 17302
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.580208383348538e-06,
+      "loss": 0.0003,
+      "step": 17303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.577807653526672e-06,
+      "loss": 0.0002,
+      "step": 17304
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.575406923704807e-06,
+      "loss": 0.0002,
+      "step": 17305
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.57300619388294e-06,
+      "loss": 0.0002,
+      "step": 17306
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.570605464061075e-06,
+      "loss": 0.0003,
+      "step": 17307
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.568204734239209e-06,
+      "loss": 0.0002,
+      "step": 17308
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.565804004417344e-06,
+      "loss": 0.0011,
+      "step": 17309
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.563403274595478e-06,
+      "loss": 0.0003,
+      "step": 17310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.561002544773613e-06,
+      "loss": 0.0002,
+      "step": 17311
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.558601814951746e-06,
+      "loss": 0.0002,
+      "step": 17312
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.55620108512988e-06,
+      "loss": 0.0002,
+      "step": 17313
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.553800355308015e-06,
+      "loss": 0.0002,
+      "step": 17314
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.551399625486148e-06,
+      "loss": 0.0027,
+      "step": 17315
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.548998895664282e-06,
+      "loss": 0.0002,
+      "step": 17316
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.546598165842417e-06,
+      "loss": 0.0002,
+      "step": 17317
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.54419743602055e-06,
+      "loss": 0.0002,
+      "step": 17318
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.541796706198684e-06,
+      "loss": 0.0001,
+      "step": 17319
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.539395976376818e-06,
+      "loss": 0.0002,
+      "step": 17320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.536995246554953e-06,
+      "loss": 0.0002,
+      "step": 17321
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.534594516733086e-06,
+      "loss": 0.0002,
+      "step": 17322
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.532193786911221e-06,
+      "loss": 0.0002,
+      "step": 17323
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.529793057089355e-06,
+      "loss": 0.0003,
+      "step": 17324
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.52739232726749e-06,
+      "loss": 0.0002,
+      "step": 17325
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.524991597445624e-06,
+      "loss": 0.0002,
+      "step": 17326
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.522590867623759e-06,
+      "loss": 0.0002,
+      "step": 17327
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.520190137801892e-06,
+      "loss": 0.0002,
+      "step": 17328
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.517789407980026e-06,
+      "loss": 0.0002,
+      "step": 17329
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.515388678158161e-06,
+      "loss": 0.0003,
+      "step": 17330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.512987948336294e-06,
+      "loss": 0.0002,
+      "step": 17331
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.51058721851443e-06,
+      "loss": 0.0002,
+      "step": 17332
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.508186488692563e-06,
+      "loss": 0.0002,
+      "step": 17333
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.505785758870698e-06,
+      "loss": 0.0002,
+      "step": 17334
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.503385029048832e-06,
+      "loss": 0.0002,
+      "step": 17335
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.500984299226965e-06,
+      "loss": 0.0002,
+      "step": 17336
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.4985835694051e-06,
+      "loss": 0.0002,
+      "step": 17337
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.496182839583234e-06,
+      "loss": 0.0002,
+      "step": 17338
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.493782109761368e-06,
+      "loss": 0.0002,
+      "step": 17339
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.491381379939501e-06,
+      "loss": 0.0003,
+      "step": 17340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.488980650117636e-06,
+      "loss": 0.0002,
+      "step": 17341
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.48657992029577e-06,
+      "loss": 0.0002,
+      "step": 17342
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.484179190473905e-06,
+      "loss": 0.0002,
+      "step": 17343
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.481778460652038e-06,
+      "loss": 0.0002,
+      "step": 17344
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.479377730830172e-06,
+      "loss": 0.0002,
+      "step": 17345
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.476977001008307e-06,
+      "loss": 0.0002,
+      "step": 17346
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 0.0002,
+      "step": 17347
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.472175541364576e-06,
+      "loss": 0.0002,
+      "step": 17348
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.46977481154271e-06,
+      "loss": 0.0002,
+      "step": 17349
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.467374081720844e-06,
+      "loss": 1.1029,
+      "step": 17350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.464973351898978e-06,
+      "loss": 0.0002,
+      "step": 17351
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.462572622077111e-06,
+      "loss": 0.0002,
+      "step": 17352
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.460171892255247e-06,
+      "loss": 0.0002,
+      "step": 17353
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.45777116243338e-06,
+      "loss": 0.0002,
+      "step": 17354
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.455370432611515e-06,
+      "loss": 0.0002,
+      "step": 17355
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.452969702789649e-06,
+      "loss": 0.0003,
+      "step": 17356
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.450568972967782e-06,
+      "loss": 0.0002,
+      "step": 17357
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.448168243145917e-06,
+      "loss": 0.0002,
+      "step": 17358
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.445767513324051e-06,
+      "loss": 0.0002,
+      "step": 17359
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.443366783502184e-06,
+      "loss": 0.6384,
+      "step": 17360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.440966053680318e-06,
+      "loss": 0.0002,
+      "step": 17361
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.438565323858453e-06,
+      "loss": 0.0002,
+      "step": 17362
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.436164594036587e-06,
+      "loss": 0.0002,
+      "step": 17363
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.433763864214722e-06,
+      "loss": 0.0002,
+      "step": 17364
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.431363134392855e-06,
+      "loss": 0.0002,
+      "step": 17365
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.42896240457099e-06,
+      "loss": 0.0003,
+      "step": 17366
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.426561674749124e-06,
+      "loss": 0.0002,
+      "step": 17367
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.424160944927258e-06,
+      "loss": 0.0002,
+      "step": 17368
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.421760215105393e-06,
+      "loss": 0.0002,
+      "step": 17369
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.419359485283526e-06,
+      "loss": 0.0002,
+      "step": 17370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.416958755461661e-06,
+      "loss": 0.0003,
+      "step": 17371
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.414558025639795e-06,
+      "loss": 0.0002,
+      "step": 17372
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.41215729581793e-06,
+      "loss": 0.001,
+      "step": 17373
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.409756565996064e-06,
+      "loss": 0.0002,
+      "step": 17374
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.407355836174199e-06,
+      "loss": 0.0003,
+      "step": 17375
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.404955106352332e-06,
+      "loss": 0.0002,
+      "step": 17376
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.402554376530466e-06,
+      "loss": 0.0002,
+      "step": 17377
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.4001536467086e-06,
+      "loss": 0.0002,
+      "step": 17378
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.397752916886734e-06,
+      "loss": 0.0002,
+      "step": 17379
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.395352187064868e-06,
+      "loss": 0.0002,
+      "step": 17380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.392951457243001e-06,
+      "loss": 0.0002,
+      "step": 17381
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.390550727421137e-06,
+      "loss": 0.0002,
+      "step": 17382
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.38814999759927e-06,
+      "loss": 0.0003,
+      "step": 17383
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.385749267777404e-06,
+      "loss": 0.0002,
+      "step": 17384
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.383348537955539e-06,
+      "loss": 0.0002,
+      "step": 17385
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.380947808133672e-06,
+      "loss": 0.0002,
+      "step": 17386
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.378547078311807e-06,
+      "loss": 0.0002,
+      "step": 17387
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.376146348489941e-06,
+      "loss": 0.0002,
+      "step": 17388
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.373745618668076e-06,
+      "loss": 0.0002,
+      "step": 17389
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.37134488884621e-06,
+      "loss": 0.0002,
+      "step": 17390
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.368944159024345e-06,
+      "loss": 0.0002,
+      "step": 17391
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.366543429202478e-06,
+      "loss": 0.0002,
+      "step": 17392
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.364142699380612e-06,
+      "loss": 0.0003,
+      "step": 17393
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.361741969558747e-06,
+      "loss": 0.0002,
+      "step": 17394
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.35934123973688e-06,
+      "loss": 0.0002,
+      "step": 17395
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.356940509915016e-06,
+      "loss": 0.0002,
+      "step": 17396
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.35453978009315e-06,
+      "loss": 0.0003,
+      "step": 17397
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.352139050271283e-06,
+      "loss": 0.0002,
+      "step": 17398
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.349738320449418e-06,
+      "loss": 0.0002,
+      "step": 17399
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.347337590627551e-06,
+      "loss": 0.0002,
+      "step": 17400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.344936860805685e-06,
+      "loss": 1.0764,
+      "step": 17401
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.342536130983818e-06,
+      "loss": 0.0003,
+      "step": 17402
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.340135401161954e-06,
+      "loss": 0.0009,
+      "step": 17403
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.337734671340087e-06,
+      "loss": 0.0002,
+      "step": 17404
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.335333941518222e-06,
+      "loss": 0.0003,
+      "step": 17405
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.332933211696356e-06,
+      "loss": 0.0002,
+      "step": 17406
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.330532481874491e-06,
+      "loss": 0.0003,
+      "step": 17407
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.328131752052624e-06,
+      "loss": 0.0003,
+      "step": 17408
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.325731022230758e-06,
+      "loss": 0.0003,
+      "step": 17409
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.323330292408893e-06,
+      "loss": 0.0003,
+      "step": 17410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.320929562587027e-06,
+      "loss": 0.0003,
+      "step": 17411
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.318528832765162e-06,
+      "loss": 0.0003,
+      "step": 17412
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.316128102943295e-06,
+      "loss": 0.0003,
+      "step": 17413
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.31372737312143e-06,
+      "loss": 0.0003,
+      "step": 17414
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.311326643299564e-06,
+      "loss": 0.0003,
+      "step": 17415
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.308925913477697e-06,
+      "loss": 0.0003,
+      "step": 17416
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.306525183655833e-06,
+      "loss": 0.0003,
+      "step": 17417
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.304124453833966e-06,
+      "loss": 0.0003,
+      "step": 17418
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.3017237240121e-06,
+      "loss": 0.0003,
+      "step": 17419
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.299322994190235e-06,
+      "loss": 0.0002,
+      "step": 17420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.296922264368368e-06,
+      "loss": 0.0003,
+      "step": 17421
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.294521534546502e-06,
+      "loss": 0.0003,
+      "step": 17422
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.292120804724637e-06,
+      "loss": 0.0004,
+      "step": 17423
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.28972007490277e-06,
+      "loss": 0.0003,
+      "step": 17424
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.287319345080904e-06,
+      "loss": 0.0003,
+      "step": 17425
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.284918615259039e-06,
+      "loss": 0.0003,
+      "step": 17426
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.282517885437173e-06,
+      "loss": 0.0003,
+      "step": 17427
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.280117155615308e-06,
+      "loss": 0.0003,
+      "step": 17428
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.277716425793441e-06,
+      "loss": 0.0003,
+      "step": 17429
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.275315695971577e-06,
+      "loss": 0.0004,
+      "step": 17430
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.27291496614971e-06,
+      "loss": 0.0003,
+      "step": 17431
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.270514236327843e-06,
+      "loss": 0.0003,
+      "step": 17432
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.268113506505979e-06,
+      "loss": 1.0819,
+      "step": 17433
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.265712776684112e-06,
+      "loss": 0.0003,
+      "step": 17434
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.263312046862247e-06,
+      "loss": 0.0003,
+      "step": 17435
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.260911317040381e-06,
+      "loss": 0.0003,
+      "step": 17436
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.258510587218516e-06,
+      "loss": 0.0003,
+      "step": 17437
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.25610985739665e-06,
+      "loss": 0.0003,
+      "step": 17438
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.253709127574783e-06,
+      "loss": 0.0004,
+      "step": 17439
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.251308397752917e-06,
+      "loss": 0.0003,
+      "step": 17440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.248907667931052e-06,
+      "loss": 0.0003,
+      "step": 17441
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.246506938109185e-06,
+      "loss": 0.0004,
+      "step": 17442
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.244106208287319e-06,
+      "loss": 0.0003,
+      "step": 17443
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.241705478465454e-06,
+      "loss": 0.0003,
+      "step": 17444
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.239304748643587e-06,
+      "loss": 0.0003,
+      "step": 17445
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.236904018821723e-06,
+      "loss": 0.0003,
+      "step": 17446
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.234503288999856e-06,
+      "loss": 0.0003,
+      "step": 17447
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.23210255917799e-06,
+      "loss": 0.0003,
+      "step": 17448
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.229701829356125e-06,
+      "loss": 0.0003,
+      "step": 17449
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.227301099534258e-06,
+      "loss": 0.0003,
+      "step": 17450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.224900369712393e-06,
+      "loss": 0.0003,
+      "step": 17451
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.222499639890527e-06,
+      "loss": 0.0003,
+      "step": 17452
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.220098910068662e-06,
+      "loss": 0.0003,
+      "step": 17453
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.217698180246796e-06,
+      "loss": 0.0003,
+      "step": 17454
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.21529745042493e-06,
+      "loss": 0.0003,
+      "step": 17455
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.212896720603064e-06,
+      "loss": 0.0003,
+      "step": 17456
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.210495990781198e-06,
+      "loss": 0.0003,
+      "step": 17457
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.208095260959333e-06,
+      "loss": 0.0003,
+      "step": 17458
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.205694531137466e-06,
+      "loss": 0.0003,
+      "step": 17459
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.2032938013156e-06,
+      "loss": 0.0003,
+      "step": 17460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.200893071493735e-06,
+      "loss": 0.0003,
+      "step": 17461
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.198492341671869e-06,
+      "loss": 0.0003,
+      "step": 17462
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.196091611850002e-06,
+      "loss": 0.0003,
+      "step": 17463
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.193690882028136e-06,
+      "loss": 0.0003,
+      "step": 17464
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.19129015220627e-06,
+      "loss": 0.0003,
+      "step": 17465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.188889422384404e-06,
+      "loss": 0.0003,
+      "step": 17466
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.18648869256254e-06,
+      "loss": 0.0003,
+      "step": 17467
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.184087962740673e-06,
+      "loss": 0.0003,
+      "step": 17468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.181687232918808e-06,
+      "loss": 0.0003,
+      "step": 17469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.179286503096942e-06,
+      "loss": 0.0003,
+      "step": 17470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.176885773275075e-06,
+      "loss": 0.0004,
+      "step": 17471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.17448504345321e-06,
+      "loss": 0.0003,
+      "step": 17472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.172084313631344e-06,
+      "loss": 0.0011,
+      "step": 17473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.169683583809479e-06,
+      "loss": 0.0003,
+      "step": 17474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.167282853987613e-06,
+      "loss": 0.0003,
+      "step": 17475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.164882124165748e-06,
+      "loss": 0.0003,
+      "step": 17476
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.162481394343881e-06,
+      "loss": 0.0003,
+      "step": 17477
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.160080664522016e-06,
+      "loss": 0.0003,
+      "step": 17478
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.15767993470015e-06,
+      "loss": 0.0003,
+      "step": 17479
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.155279204878283e-06,
+      "loss": 0.0003,
+      "step": 17480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.152878475056417e-06,
+      "loss": 0.0003,
+      "step": 17481
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.150477745234552e-06,
+      "loss": 0.0002,
+      "step": 17482
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.148077015412686e-06,
+      "loss": 0.0003,
+      "step": 17483
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.145676285590819e-06,
+      "loss": 0.0003,
+      "step": 17484
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.143275555768954e-06,
+      "loss": 0.0002,
+      "step": 17485
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.140874825947088e-06,
+      "loss": 0.0003,
+      "step": 17486
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.138474096125221e-06,
+      "loss": 0.0003,
+      "step": 17487
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.136073366303356e-06,
+      "loss": 0.0003,
+      "step": 17488
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.13367263648149e-06,
+      "loss": 0.0003,
+      "step": 17489
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.131271906659625e-06,
+      "loss": 0.0003,
+      "step": 17490
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.128871176837759e-06,
+      "loss": 0.0003,
+      "step": 17491
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.126470447015894e-06,
+      "loss": 0.0003,
+      "step": 17492
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.124069717194027e-06,
+      "loss": 0.0003,
+      "step": 17493
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.121668987372162e-06,
+      "loss": 0.0003,
+      "step": 17494
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.119268257550296e-06,
+      "loss": 0.0002,
+      "step": 17495
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.11686752772843e-06,
+      "loss": 0.0003,
+      "step": 17496
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.114466797906565e-06,
+      "loss": 0.0003,
+      "step": 17497
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.112066068084698e-06,
+      "loss": 0.0003,
+      "step": 17498
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.109665338262833e-06,
+      "loss": 0.0003,
+      "step": 17499
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.107264608440967e-06,
+      "loss": 0.0003,
+      "step": 17500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.1048638786191e-06,
+      "loss": 0.0003,
+      "step": 17501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.102463148797234e-06,
+      "loss": 0.0003,
+      "step": 17502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.100062418975369e-06,
+      "loss": 0.0002,
+      "step": 17503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.097661689153503e-06,
+      "loss": 0.0002,
+      "step": 17504
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.095260959331636e-06,
+      "loss": 0.0003,
+      "step": 17505
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.092860229509771e-06,
+      "loss": 0.0003,
+      "step": 17506
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.090459499687905e-06,
+      "loss": 0.0002,
+      "step": 17507
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.08805876986604e-06,
+      "loss": 0.0002,
+      "step": 17508
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.085658040044173e-06,
+      "loss": 0.0003,
+      "step": 17509
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.083257310222309e-06,
+      "loss": 0.0002,
+      "step": 17510
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.080856580400442e-06,
+      "loss": 0.0002,
+      "step": 17511
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.078455850578576e-06,
+      "loss": 0.0002,
+      "step": 17512
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.07605512075671e-06,
+      "loss": 0.0002,
+      "step": 17513
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.073654390934844e-06,
+      "loss": 0.0002,
+      "step": 17514
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.07125366111298e-06,
+      "loss": 0.0003,
+      "step": 17515
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.068852931291113e-06,
+      "loss": 0.0002,
+      "step": 17516
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.066452201469248e-06,
+      "loss": 0.0003,
+      "step": 17517
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.064051471647382e-06,
+      "loss": 0.0003,
+      "step": 17518
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.061650741825515e-06,
+      "loss": 0.0003,
+      "step": 17519
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.05925001200365e-06,
+      "loss": 0.0003,
+      "step": 17520
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.056849282181784e-06,
+      "loss": 0.0003,
+      "step": 17521
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.054448552359917e-06,
+      "loss": 0.0002,
+      "step": 17522
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.05204782253805e-06,
+      "loss": 0.6894,
+      "step": 17523
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.049647092716186e-06,
+      "loss": 0.0002,
+      "step": 17524
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.04724636289432e-06,
+      "loss": 0.0003,
+      "step": 17525
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.044845633072455e-06,
+      "loss": 0.0003,
+      "step": 17526
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.042444903250588e-06,
+      "loss": 0.0003,
+      "step": 17527
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.040044173428722e-06,
+      "loss": 0.0003,
+      "step": 17528
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.037643443606857e-06,
+      "loss": 0.0003,
+      "step": 17529
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.03524271378499e-06,
+      "loss": 0.0003,
+      "step": 17530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.032841983963126e-06,
+      "loss": 0.0002,
+      "step": 17531
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.030441254141259e-06,
+      "loss": 0.0002,
+      "step": 17532
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.028040524319394e-06,
+      "loss": 0.0003,
+      "step": 17533
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.025639794497528e-06,
+      "loss": 0.0003,
+      "step": 17534
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.023239064675661e-06,
+      "loss": 0.0003,
+      "step": 17535
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.020838334853796e-06,
+      "loss": 0.0003,
+      "step": 17536
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.01843760503193e-06,
+      "loss": 0.0003,
+      "step": 17537
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.016036875210065e-06,
+      "loss": 0.0002,
+      "step": 17538
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.013636145388199e-06,
+      "loss": 0.0002,
+      "step": 17539
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.011235415566334e-06,
+      "loss": 0.0003,
+      "step": 17540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.008834685744467e-06,
+      "loss": 0.0002,
+      "step": 17541
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.0064339559226e-06,
+      "loss": 0.0002,
+      "step": 17542
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.004033226100734e-06,
+      "loss": 0.0002,
+      "step": 17543
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.00163249627887e-06,
+      "loss": 0.0003,
+      "step": 17544
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.999231766457003e-06,
+      "loss": 0.0002,
+      "step": 17545
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.996831036635136e-06,
+      "loss": 0.0003,
+      "step": 17546
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.994430306813272e-06,
+      "loss": 0.0002,
+      "step": 17547
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.992029576991405e-06,
+      "loss": 0.0002,
+      "step": 17548
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.98962884716954e-06,
+      "loss": 0.0003,
+      "step": 17549
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.987228117347674e-06,
+      "loss": 0.0003,
+      "step": 17550
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.984827387525807e-06,
+      "loss": 0.0002,
+      "step": 17551
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.982426657703942e-06,
+      "loss": 0.0002,
+      "step": 17552
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.980025927882076e-06,
+      "loss": 0.0011,
+      "step": 17553
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.977625198060211e-06,
+      "loss": 0.0006,
+      "step": 17554
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.975224468238345e-06,
+      "loss": 0.0002,
+      "step": 17555
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.97282373841648e-06,
+      "loss": 0.0002,
+      "step": 17556
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.970423008594613e-06,
+      "loss": 1.0134,
+      "step": 17557
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.968022278772748e-06,
+      "loss": 0.0002,
+      "step": 17558
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.965621548950882e-06,
+      "loss": 0.0002,
+      "step": 17559
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.963220819129015e-06,
+      "loss": 1.0356,
+      "step": 17560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.96082008930715e-06,
+      "loss": 0.0002,
+      "step": 17561
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.958419359485284e-06,
+      "loss": 0.0003,
+      "step": 17562
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.956018629663418e-06,
+      "loss": 0.0002,
+      "step": 17563
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.953617899841551e-06,
+      "loss": 0.0003,
+      "step": 17564
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.951217170019686e-06,
+      "loss": 0.0002,
+      "step": 17565
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.94881644019782e-06,
+      "loss": 0.0003,
+      "step": 17566
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.946415710375953e-06,
+      "loss": 0.0002,
+      "step": 17567
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.944014980554089e-06,
+      "loss": 0.0003,
+      "step": 17568
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.941614250732222e-06,
+      "loss": 0.0002,
+      "step": 17569
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.939213520910357e-06,
+      "loss": 0.0003,
+      "step": 17570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.93681279108849e-06,
+      "loss": 0.0003,
+      "step": 17571
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.934412061266626e-06,
+      "loss": 1.0716,
+      "step": 17572
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.93201133144476e-06,
+      "loss": 0.0003,
+      "step": 17573
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.929610601622895e-06,
+      "loss": 0.0003,
+      "step": 17574
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.927209871801028e-06,
+      "loss": 0.0003,
+      "step": 17575
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.924809141979162e-06,
+      "loss": 0.0003,
+      "step": 17576
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.922408412157297e-06,
+      "loss": 0.0003,
+      "step": 17577
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.92000768233543e-06,
+      "loss": 0.0003,
+      "step": 17578
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.917606952513565e-06,
+      "loss": 0.0003,
+      "step": 17579
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.915206222691699e-06,
+      "loss": 0.0004,
+      "step": 17580
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.912805492869834e-06,
+      "loss": 0.0003,
+      "step": 17581
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.910404763047968e-06,
+      "loss": 0.0038,
+      "step": 17582
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.908004033226101e-06,
+      "loss": 0.0003,
+      "step": 17583
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.905603303404235e-06,
+      "loss": 0.0003,
+      "step": 17584
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.903202573582368e-06,
+      "loss": 0.0003,
+      "step": 17585
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.900801843760503e-06,
+      "loss": 0.0003,
+      "step": 17586
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.898401113938637e-06,
+      "loss": 0.0003,
+      "step": 17587
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.896000384116772e-06,
+      "loss": 0.0004,
+      "step": 17588
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.893599654294905e-06,
+      "loss": 0.0004,
+      "step": 17589
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.89119892447304e-06,
+      "loss": 0.0003,
+      "step": 17590
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.888798194651174e-06,
+      "loss": 0.0003,
+      "step": 17591
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.886397464829308e-06,
+      "loss": 0.0003,
+      "step": 17592
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.883996735007443e-06,
+      "loss": 0.0004,
+      "step": 17593
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.881596005185576e-06,
+      "loss": 0.0003,
+      "step": 17594
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.879195275363711e-06,
+      "loss": 0.0004,
+      "step": 17595
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.876794545541845e-06,
+      "loss": 0.0004,
+      "step": 17596
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.87439381571998e-06,
+      "loss": 0.5441,
+      "step": 17597
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.871993085898114e-06,
+      "loss": 0.0003,
+      "step": 17598
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.869592356076247e-06,
+      "loss": 0.0004,
+      "step": 17599
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.867191626254382e-06,
+      "loss": 0.0003,
+      "step": 17600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.864790896432516e-06,
+      "loss": 0.0003,
+      "step": 17601
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.862390166610651e-06,
+      "loss": 0.0003,
+      "step": 17602
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.859989436788785e-06,
+      "loss": 0.0003,
+      "step": 17603
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.857588706966918e-06,
+      "loss": 0.0003,
+      "step": 17604
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.855187977145052e-06,
+      "loss": 0.0003,
+      "step": 17605
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.852787247323187e-06,
+      "loss": 0.0003,
+      "step": 17606
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.85038651750132e-06,
+      "loss": 0.0004,
+      "step": 17607
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.847985787679454e-06,
+      "loss": 0.0003,
+      "step": 17608
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.845585057857589e-06,
+      "loss": 0.0009,
+      "step": 17609
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.843184328035722e-06,
+      "loss": 0.0003,
+      "step": 17610
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.840783598213858e-06,
+      "loss": 0.0003,
+      "step": 17611
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.838382868391991e-06,
+      "loss": 0.0003,
+      "step": 17612
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.835982138570126e-06,
+      "loss": 0.0004,
+      "step": 17613
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.83358140874826e-06,
+      "loss": 0.0007,
+      "step": 17614
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.831180678926393e-06,
+      "loss": 0.0004,
+      "step": 17615
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.828779949104528e-06,
+      "loss": 0.0003,
+      "step": 17616
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.826379219282662e-06,
+      "loss": 0.0003,
+      "step": 17617
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.823978489460797e-06,
+      "loss": 0.0027,
+      "step": 17618
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.82157775963893e-06,
+      "loss": 0.0003,
+      "step": 17619
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.819177029817066e-06,
+      "loss": 0.0003,
+      "step": 17620
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.8167762999952e-06,
+      "loss": 0.0003,
+      "step": 17621
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.814375570173334e-06,
+      "loss": 0.0004,
+      "step": 17622
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.811974840351468e-06,
+      "loss": 0.0003,
+      "step": 17623
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.809574110529601e-06,
+      "loss": 0.0003,
+      "step": 17624
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.807173380707735e-06,
+      "loss": 0.0003,
+      "step": 17625
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.804772650885868e-06,
+      "loss": 0.0003,
+      "step": 17626
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.802371921064004e-06,
+      "loss": 0.0003,
+      "step": 17627
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.799971191242137e-06,
+      "loss": 0.0004,
+      "step": 17628
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.797570461420272e-06,
+      "loss": 0.0003,
+      "step": 17629
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.795169731598406e-06,
+      "loss": 0.0003,
+      "step": 17630
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.79276900177654e-06,
+      "loss": 0.0003,
+      "step": 17631
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.790368271954675e-06,
+      "loss": 0.0003,
+      "step": 17632
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.787967542132808e-06,
+      "loss": 0.0003,
+      "step": 17633
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.785566812310943e-06,
+      "loss": 0.0004,
+      "step": 17634
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.783166082489077e-06,
+      "loss": 0.0004,
+      "step": 17635
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.780765352667212e-06,
+      "loss": 0.0003,
+      "step": 17636
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.778364622845345e-06,
+      "loss": 0.0003,
+      "step": 17637
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.77596389302348e-06,
+      "loss": 0.0003,
+      "step": 17638
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.773563163201614e-06,
+      "loss": 0.0003,
+      "step": 17639
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.771162433379748e-06,
+      "loss": 0.0003,
+      "step": 17640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.768761703557883e-06,
+      "loss": 0.0003,
+      "step": 17641
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.766360973736016e-06,
+      "loss": 0.0003,
+      "step": 17642
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.763960243914151e-06,
+      "loss": 0.0003,
+      "step": 17643
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.761559514092285e-06,
+      "loss": 0.0003,
+      "step": 17644
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.759158784270418e-06,
+      "loss": 0.0003,
+      "step": 17645
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.756758054448552e-06,
+      "loss": 0.0003,
+      "step": 17646
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.754357324626685e-06,
+      "loss": 0.0003,
+      "step": 17647
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.75195659480482e-06,
+      "loss": 0.0004,
+      "step": 17648
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.749555864982954e-06,
+      "loss": 0.0003,
+      "step": 17649
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.74715513516109e-06,
+      "loss": 0.0003,
+      "step": 17650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.744754405339223e-06,
+      "loss": 0.0003,
+      "step": 17651
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.742353675517358e-06,
+      "loss": 0.0003,
+      "step": 17652
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.739952945695491e-06,
+      "loss": 0.0003,
+      "step": 17653
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.737552215873627e-06,
+      "loss": 0.0003,
+      "step": 17654
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.73515148605176e-06,
+      "loss": 0.0003,
+      "step": 17655
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.732750756229894e-06,
+      "loss": 0.0002,
+      "step": 17656
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.730350026408029e-06,
+      "loss": 0.0003,
+      "step": 17657
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.727949296586162e-06,
+      "loss": 0.0011,
+      "step": 17658
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.725548566764297e-06,
+      "loss": 0.0003,
+      "step": 17659
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.723147836942431e-06,
+      "loss": 0.0003,
+      "step": 17660
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.720747107120566e-06,
+      "loss": 0.0003,
+      "step": 17661
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.7183463772987e-06,
+      "loss": 0.0003,
+      "step": 17662
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.715945647476833e-06,
+      "loss": 0.0003,
+      "step": 17663
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.713544917654968e-06,
+      "loss": 0.0003,
+      "step": 17664
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.711144187833102e-06,
+      "loss": 0.6509,
+      "step": 17665
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.708743458011235e-06,
+      "loss": 0.0003,
+      "step": 17666
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.706342728189369e-06,
+      "loss": 0.0003,
+      "step": 17667
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.703941998367504e-06,
+      "loss": 0.0003,
+      "step": 17668
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.701541268545638e-06,
+      "loss": 0.0003,
+      "step": 17669
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.699140538723773e-06,
+      "loss": 0.9661,
+      "step": 17670
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.696739808901906e-06,
+      "loss": 0.0003,
+      "step": 17671
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.69433907908004e-06,
+      "loss": 0.0003,
+      "step": 17672
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.691938349258175e-06,
+      "loss": 0.0003,
+      "step": 17673
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.689537619436308e-06,
+      "loss": 0.0003,
+      "step": 17674
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.687136889614444e-06,
+      "loss": 0.0003,
+      "step": 17675
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.684736159792577e-06,
+      "loss": 0.0003,
+      "step": 17676
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.682335429970712e-06,
+      "loss": 0.0003,
+      "step": 17677
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.679934700148846e-06,
+      "loss": 0.0003,
+      "step": 17678
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.67753397032698e-06,
+      "loss": 0.0004,
+      "step": 17679
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.675133240505114e-06,
+      "loss": 0.0003,
+      "step": 17680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.672732510683248e-06,
+      "loss": 0.0003,
+      "step": 17681
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.670331780861383e-06,
+      "loss": 0.0003,
+      "step": 17682
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.667931051039517e-06,
+      "loss": 0.0004,
+      "step": 17683
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.665530321217652e-06,
+      "loss": 1.0044,
+      "step": 17684
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.663129591395785e-06,
+      "loss": 0.0003,
+      "step": 17685
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.660728861573919e-06,
+      "loss": 0.0038,
+      "step": 17686
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.658328131752052e-06,
+      "loss": 0.0003,
+      "step": 17687
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.655927401930186e-06,
+      "loss": 0.0003,
+      "step": 17688
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.653526672108321e-06,
+      "loss": 0.0004,
+      "step": 17689
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.651125942286454e-06,
+      "loss": 0.0003,
+      "step": 17690
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.64872521246459e-06,
+      "loss": 0.0004,
+      "step": 17691
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.646324482642723e-06,
+      "loss": 0.0004,
+      "step": 17692
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.643923752820858e-06,
+      "loss": 0.0003,
+      "step": 17693
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.641523022998992e-06,
+      "loss": 0.0003,
+      "step": 17694
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.639122293177125e-06,
+      "loss": 0.0004,
+      "step": 17695
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.63672156335526e-06,
+      "loss": 0.0003,
+      "step": 17696
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.634320833533394e-06,
+      "loss": 0.0004,
+      "step": 17697
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.63192010371153e-06,
+      "loss": 0.0004,
+      "step": 17698
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.629519373889663e-06,
+      "loss": 0.0003,
+      "step": 17699
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.627118644067798e-06,
+      "loss": 0.0004,
+      "step": 17700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.624717914245931e-06,
+      "loss": 0.0004,
+      "step": 17701
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.622317184424066e-06,
+      "loss": 0.0003,
+      "step": 17702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.619916454602199e-06,
+      "loss": 0.0004,
+      "step": 17703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.617515724780333e-06,
+      "loss": 0.0004,
+      "step": 17704
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.615114994958468e-06,
+      "loss": 0.0004,
+      "step": 17705
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.612714265136601e-06,
+      "loss": 0.0003,
+      "step": 17706
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.6103135353147366e-06,
+      "loss": 0.6692,
+      "step": 17707
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.60791280549287e-06,
+      "loss": 0.0004,
+      "step": 17708
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.605512075671005e-06,
+      "loss": 0.0004,
+      "step": 17709
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.603111345849139e-06,
+      "loss": 0.0004,
+      "step": 17710
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.600710616027272e-06,
+      "loss": 0.003,
+      "step": 17711
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5983098862054066e-06,
+      "loss": 0.0003,
+      "step": 17712
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.595909156383541e-06,
+      "loss": 0.0003,
+      "step": 17713
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.593508426561675e-06,
+      "loss": 0.0003,
+      "step": 17714
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.591107696739809e-06,
+      "loss": 0.0004,
+      "step": 17715
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.588706966917944e-06,
+      "loss": 0.0004,
+      "step": 17716
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5863062370960774e-06,
+      "loss": 0.0003,
+      "step": 17717
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.583905507274213e-06,
+      "loss": 0.0004,
+      "step": 17718
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.581504777452346e-06,
+      "loss": 0.0003,
+      "step": 17719
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.57910404763048e-06,
+      "loss": 0.0004,
+      "step": 17720
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.576703317808615e-06,
+      "loss": 0.0004,
+      "step": 17721
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.574302587986748e-06,
+      "loss": 0.0003,
+      "step": 17722
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.571901858164883e-06,
+      "loss": 0.0004,
+      "step": 17723
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.569501128343016e-06,
+      "loss": 0.0004,
+      "step": 17724
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.567100398521151e-06,
+      "loss": 0.0004,
+      "step": 17725
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.564699668699285e-06,
+      "loss": 0.0004,
+      "step": 17726
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.562298938877418e-06,
+      "loss": 0.0004,
+      "step": 17727
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5598982090555535e-06,
+      "loss": 0.0004,
+      "step": 17728
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.557497479233687e-06,
+      "loss": 0.0004,
+      "step": 17729
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.555096749411822e-06,
+      "loss": 0.0004,
+      "step": 17730
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.552696019589956e-06,
+      "loss": 0.0004,
+      "step": 17731
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.55029528976809e-06,
+      "loss": 0.0003,
+      "step": 17732
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5478945599462235e-06,
+      "loss": 0.0003,
+      "step": 17733
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.545493830124359e-06,
+      "loss": 0.0004,
+      "step": 17734
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.543093100302492e-06,
+      "loss": 0.9935,
+      "step": 17735
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.540692370480626e-06,
+      "loss": 0.0004,
+      "step": 17736
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.538291640658761e-06,
+      "loss": 0.0004,
+      "step": 17737
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.535890910836894e-06,
+      "loss": 0.0004,
+      "step": 17738
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5334901810150296e-06,
+      "loss": 0.0004,
+      "step": 17739
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.531089451193163e-06,
+      "loss": 0.0004,
+      "step": 17740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.528688721371298e-06,
+      "loss": 0.0004,
+      "step": 17741
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.526287991549432e-06,
+      "loss": 0.0004,
+      "step": 17742
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.523887261727565e-06,
+      "loss": 0.0004,
+      "step": 17743
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5214865319056996e-06,
+      "loss": 0.0004,
+      "step": 17744
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.519085802083833e-06,
+      "loss": 0.0005,
+      "step": 17745
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.516685072261968e-06,
+      "loss": 0.0004,
+      "step": 17746
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.514284342440102e-06,
+      "loss": 0.0005,
+      "step": 17747
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.511883612618237e-06,
+      "loss": 0.0004,
+      "step": 17748
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5094828827963704e-06,
+      "loss": 0.0004,
+      "step": 17749
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.507082152974506e-06,
+      "loss": 0.9134,
+      "step": 17750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.504681423152639e-06,
+      "loss": 0.9528,
+      "step": 17751
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.502280693330773e-06,
+      "loss": 0.0004,
+      "step": 17752
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.499879963508907e-06,
+      "loss": 0.0005,
+      "step": 17753
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4974792336870404e-06,
+      "loss": 0.0006,
+      "step": 17754
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.495078503865176e-06,
+      "loss": 0.0005,
+      "step": 17755
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.492677774043309e-06,
+      "loss": 0.0006,
+      "step": 17756
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.490277044221444e-06,
+      "loss": 0.0006,
+      "step": 17757
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.487876314399578e-06,
+      "loss": 0.0008,
+      "step": 17758
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.485475584577711e-06,
+      "loss": 0.0005,
+      "step": 17759
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4830748547558465e-06,
+      "loss": 0.0008,
+      "step": 17760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.48067412493398e-06,
+      "loss": 0.0007,
+      "step": 17761
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.478273395112115e-06,
+      "loss": 0.0007,
+      "step": 17762
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.475872665290249e-06,
+      "loss": 0.0008,
+      "step": 17763
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.473471935468383e-06,
+      "loss": 0.0007,
+      "step": 17764
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4710712056465165e-06,
+      "loss": 0.8482,
+      "step": 17765
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.468670475824652e-06,
+      "loss": 0.0008,
+      "step": 17766
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.466269746002785e-06,
+      "loss": 0.0006,
+      "step": 17767
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.463869016180919e-06,
+      "loss": 0.0005,
+      "step": 17768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.461468286359054e-06,
+      "loss": 0.0014,
+      "step": 17769
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.459067556537187e-06,
+      "loss": 0.0009,
+      "step": 17770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4566668267153226e-06,
+      "loss": 0.0009,
+      "step": 17771
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.454266096893456e-06,
+      "loss": 0.0011,
+      "step": 17772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.45186536707159e-06,
+      "loss": 0.0009,
+      "step": 17773
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.449464637249724e-06,
+      "loss": 0.0012,
+      "step": 17774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.447063907427858e-06,
+      "loss": 0.0007,
+      "step": 17775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4446631776059926e-06,
+      "loss": 0.0009,
+      "step": 17776
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.442262447784126e-06,
+      "loss": 0.0014,
+      "step": 17777
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.439861717962261e-06,
+      "loss": 0.0014,
+      "step": 17778
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.437460988140395e-06,
+      "loss": 0.0014,
+      "step": 17779
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.43506025831853e-06,
+      "loss": 0.0045,
+      "step": 17780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.4326595284966634e-06,
+      "loss": 0.0012,
+      "step": 17781
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.430258798674798e-06,
+      "loss": 0.0011,
+      "step": 17782
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.427858068852932e-06,
+      "loss": 0.001,
+      "step": 17783
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.425457339031066e-06,
+      "loss": 0.0014,
+      "step": 17784
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.4230566092092e-06,
+      "loss": 0.0009,
+      "step": 17785
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.4206558793873334e-06,
+      "loss": 0.0015,
+      "step": 17786
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.418255149565469e-06,
+      "loss": 0.0011,
+      "step": 17787
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.415854419743602e-06,
+      "loss": 0.0012,
+      "step": 17788
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.413453689921737e-06,
+      "loss": 0.0009,
+      "step": 17789
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.411052960099871e-06,
+      "loss": 0.001,
+      "step": 17790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.408652230278004e-06,
+      "loss": 0.001,
+      "step": 17791
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.4062515004561395e-06,
+      "loss": 0.0008,
+      "step": 17792
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.403850770634273e-06,
+      "loss": 0.001,
+      "step": 17793
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.401450040812407e-06,
+      "loss": 0.0006,
+      "step": 17794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.399049310990541e-06,
+      "loss": 0.0008,
+      "step": 17795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.396648581168676e-06,
+      "loss": 0.4461,
+      "step": 17796
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3942478513468095e-06,
+      "loss": 0.0009,
+      "step": 17797
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.391847121524945e-06,
+      "loss": 0.001,
+      "step": 17798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.389446391703078e-06,
+      "loss": 0.0015,
+      "step": 17799
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.387045661881212e-06,
+      "loss": 0.001,
+      "step": 17800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.384644932059347e-06,
+      "loss": 0.0007,
+      "step": 17801
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.38224420223748e-06,
+      "loss": 0.0008,
+      "step": 17802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3798434724156155e-06,
+      "loss": 0.0007,
+      "step": 17803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.377442742593749e-06,
+      "loss": 0.0008,
+      "step": 17804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.375042012771883e-06,
+      "loss": 0.0008,
+      "step": 17805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.372641282950017e-06,
+      "loss": 0.0006,
+      "step": 17806
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.37024055312815e-06,
+      "loss": 0.0005,
+      "step": 17807
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3678398233062856e-06,
+      "loss": 0.0006,
+      "step": 17808
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.365439093484419e-06,
+      "loss": 0.001,
+      "step": 17809
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.363038363662554e-06,
+      "loss": 0.0009,
+      "step": 17810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.360637633840688e-06,
+      "loss": 0.0008,
+      "step": 17811
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.358236904018823e-06,
+      "loss": 0.0009,
+      "step": 17812
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.355836174196956e-06,
+      "loss": 0.0006,
+      "step": 17813
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.35343544437509e-06,
+      "loss": 0.0008,
+      "step": 17814
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.351034714553224e-06,
+      "loss": 0.0008,
+      "step": 17815
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.348633984731358e-06,
+      "loss": 0.0007,
+      "step": 17816
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.346233254909493e-06,
+      "loss": 0.0007,
+      "step": 17817
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3438325250876264e-06,
+      "loss": 0.0006,
+      "step": 17818
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.341431795265762e-06,
+      "loss": 0.0006,
+      "step": 17819
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.339031065443895e-06,
+      "loss": 0.0007,
+      "step": 17820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.33663033562203e-06,
+      "loss": 0.0005,
+      "step": 17821
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.334229605800164e-06,
+      "loss": 0.0006,
+      "step": 17822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.331828875978297e-06,
+      "loss": 0.0005,
+      "step": 17823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3294281461564325e-06,
+      "loss": 0.0005,
+      "step": 17824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.327027416334566e-06,
+      "loss": 0.0004,
+      "step": 17825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3246266865127e-06,
+      "loss": 0.0005,
+      "step": 17826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.322225956690834e-06,
+      "loss": 0.0007,
+      "step": 17827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.319825226868969e-06,
+      "loss": 0.0006,
+      "step": 17828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3174244970471025e-06,
+      "loss": 0.0006,
+      "step": 17829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.315023767225236e-06,
+      "loss": 0.0005,
+      "step": 17830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.312623037403371e-06,
+      "loss": 0.0004,
+      "step": 17831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.310222307581505e-06,
+      "loss": 0.0007,
+      "step": 17832
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.30782157775964e-06,
+      "loss": 0.0004,
+      "step": 17833
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.305420847937773e-06,
+      "loss": 0.0007,
+      "step": 17834
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.303020118115908e-06,
+      "loss": 0.0005,
+      "step": 17835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.300619388294041e-06,
+      "loss": 0.0006,
+      "step": 17836
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.298218658472176e-06,
+      "loss": 0.0006,
+      "step": 17837
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.29581792865031e-06,
+      "loss": 0.0007,
+      "step": 17838
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.293417198828443e-06,
+      "loss": 0.0006,
+      "step": 17839
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.2910164690065786e-06,
+      "loss": 0.0006,
+      "step": 17840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.288615739184712e-06,
+      "loss": 0.0005,
+      "step": 17841
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.286215009362847e-06,
+      "loss": 0.0005,
+      "step": 17842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.283814279540981e-06,
+      "loss": 0.0006,
+      "step": 17843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.281413549719115e-06,
+      "loss": 0.0007,
+      "step": 17844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.279012819897249e-06,
+      "loss": 0.0006,
+      "step": 17845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.276612090075383e-06,
+      "loss": 0.0005,
+      "step": 17846
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.274211360253517e-06,
+      "loss": 0.0007,
+      "step": 17847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.271810630431651e-06,
+      "loss": 0.0006,
+      "step": 17848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.269409900609786e-06,
+      "loss": 0.0004,
+      "step": 17849
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2670091707879194e-06,
+      "loss": 0.0005,
+      "step": 17850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.264608440966055e-06,
+      "loss": 0.0006,
+      "step": 17851
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.262207711144188e-06,
+      "loss": 0.0006,
+      "step": 17852
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.259806981322323e-06,
+      "loss": 0.0006,
+      "step": 17853
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.257406251500457e-06,
+      "loss": 0.0006,
+      "step": 17854
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.25500552167859e-06,
+      "loss": 0.0006,
+      "step": 17855
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.252604791856725e-06,
+      "loss": 0.0006,
+      "step": 17856
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.250204062034858e-06,
+      "loss": 0.0006,
+      "step": 17857
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.247803332212993e-06,
+      "loss": 0.0005,
+      "step": 17858
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.245402602391127e-06,
+      "loss": 0.0004,
+      "step": 17859
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.243001872569262e-06,
+      "loss": 0.0005,
+      "step": 17860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2406011427473955e-06,
+      "loss": 0.0005,
+      "step": 17861
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.238200412925529e-06,
+      "loss": 0.0005,
+      "step": 17862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.235799683103664e-06,
+      "loss": 0.0005,
+      "step": 17863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.233398953281798e-06,
+      "loss": 0.0005,
+      "step": 17864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.230998223459933e-06,
+      "loss": 0.0005,
+      "step": 17865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.228597493638066e-06,
+      "loss": 0.0004,
+      "step": 17866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.226196763816201e-06,
+      "loss": 0.0006,
+      "step": 17867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.223796033994334e-06,
+      "loss": 0.0004,
+      "step": 17868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.221395304172469e-06,
+      "loss": 0.0005,
+      "step": 17869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.218994574350603e-06,
+      "loss": 0.0004,
+      "step": 17870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.216593844528736e-06,
+      "loss": 0.0006,
+      "step": 17871
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2141931147068716e-06,
+      "loss": 0.0005,
+      "step": 17872
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.211792384885005e-06,
+      "loss": 0.0006,
+      "step": 17873
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.20939165506314e-06,
+      "loss": 0.0005,
+      "step": 17874
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.206990925241274e-06,
+      "loss": 0.0005,
+      "step": 17875
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.204590195419408e-06,
+      "loss": 0.0005,
+      "step": 17876
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2021894655975416e-06,
+      "loss": 0.0004,
+      "step": 17877
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.199788735775675e-06,
+      "loss": 0.0006,
+      "step": 17878
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.19738800595381e-06,
+      "loss": 0.0005,
+      "step": 17879
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.194987276131944e-06,
+      "loss": 0.0005,
+      "step": 17880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.192586546310079e-06,
+      "loss": 0.0004,
+      "step": 17881
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1901858164882124e-06,
+      "loss": 0.0005,
+      "step": 17882
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.187785086666348e-06,
+      "loss": 0.0004,
+      "step": 17883
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.185384356844481e-06,
+      "loss": 0.0004,
+      "step": 17884
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1829836270226155e-06,
+      "loss": 0.0004,
+      "step": 17885
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.18058289720075e-06,
+      "loss": 0.0005,
+      "step": 17886
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.178182167378883e-06,
+      "loss": 0.0005,
+      "step": 17887
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.175781437557018e-06,
+      "loss": 0.0004,
+      "step": 17888
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.173380707735151e-06,
+      "loss": 0.0004,
+      "step": 17889
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.170979977913286e-06,
+      "loss": 0.7009,
+      "step": 17890
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.16857924809142e-06,
+      "loss": 0.0004,
+      "step": 17891
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.166178518269555e-06,
+      "loss": 0.0004,
+      "step": 17892
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1637777884476885e-06,
+      "loss": 0.0004,
+      "step": 17893
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.161377058625822e-06,
+      "loss": 0.0004,
+      "step": 17894
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.158976328803957e-06,
+      "loss": 0.0004,
+      "step": 17895
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.156575598982091e-06,
+      "loss": 0.0004,
+      "step": 17896
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.154174869160225e-06,
+      "loss": 0.0004,
+      "step": 17897
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1517741393383585e-06,
+      "loss": 0.0003,
+      "step": 17898
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.149373409516494e-06,
+      "loss": 0.0004,
+      "step": 17899
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.146972679694627e-06,
+      "loss": 0.0005,
+      "step": 17900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.144571949872762e-06,
+      "loss": 0.0004,
+      "step": 17901
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.142171220050896e-06,
+      "loss": 0.0004,
+      "step": 17902
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.139770490229029e-06,
+      "loss": 0.0005,
+      "step": 17903
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1373697604071646e-06,
+      "loss": 0.0005,
+      "step": 17904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.134969030585298e-06,
+      "loss": 0.0004,
+      "step": 17905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.132568300763432e-06,
+      "loss": 0.0004,
+      "step": 17906
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.130167570941567e-06,
+      "loss": 0.0004,
+      "step": 17907
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.127766841119701e-06,
+      "loss": 0.0004,
+      "step": 17908
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1253661112978346e-06,
+      "loss": 0.0004,
+      "step": 17909
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.122965381475968e-06,
+      "loss": 0.0004,
+      "step": 17910
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.120564651654103e-06,
+      "loss": 0.0003,
+      "step": 17911
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.118163921832237e-06,
+      "loss": 0.0005,
+      "step": 17912
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.115763192010372e-06,
+      "loss": 0.0004,
+      "step": 17913
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1133624621885054e-06,
+      "loss": 0.0004,
+      "step": 17914
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.110961732366641e-06,
+      "loss": 0.0003,
+      "step": 17915
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.108561002544774e-06,
+      "loss": 0.0005,
+      "step": 17916
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1061602727229084e-06,
+      "loss": 0.0004,
+      "step": 17917
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.103759542901042e-06,
+      "loss": 0.0004,
+      "step": 17918
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.1013588130791754e-06,
+      "loss": 0.0004,
+      "step": 17919
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.098958083257311e-06,
+      "loss": 0.0003,
+      "step": 17920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.096557353435444e-06,
+      "loss": 0.0005,
+      "step": 17921
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.094156623613579e-06,
+      "loss": 0.0006,
+      "step": 17922
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.091755893791713e-06,
+      "loss": 0.0004,
+      "step": 17923
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.089355163969848e-06,
+      "loss": 0.0004,
+      "step": 17924
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0869544341479815e-06,
+      "loss": 0.0004,
+      "step": 17925
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.084553704326115e-06,
+      "loss": 0.0003,
+      "step": 17926
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.08215297450425e-06,
+      "loss": 0.0004,
+      "step": 17927
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.079752244682384e-06,
+      "loss": 0.0004,
+      "step": 17928
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.077351514860518e-06,
+      "loss": 0.0006,
+      "step": 17929
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0749507850386515e-06,
+      "loss": 0.9266,
+      "step": 17930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.072550055216787e-06,
+      "loss": 0.0004,
+      "step": 17931
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.07014932539492e-06,
+      "loss": 0.0004,
+      "step": 17932
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.067748595573055e-06,
+      "loss": 0.0004,
+      "step": 17933
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.065347865751189e-06,
+      "loss": 0.0004,
+      "step": 17934
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.062947135929322e-06,
+      "loss": 0.0004,
+      "step": 17935
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0605464061074575e-06,
+      "loss": 0.0004,
+      "step": 17936
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.058145676285591e-06,
+      "loss": 0.0004,
+      "step": 17937
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.055744946463725e-06,
+      "loss": 0.0005,
+      "step": 17938
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.053344216641859e-06,
+      "loss": 0.0004,
+      "step": 17939
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.050943486819994e-06,
+      "loss": 0.0004,
+      "step": 17940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0485427569981276e-06,
+      "loss": 0.0004,
+      "step": 17941
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.046142027176261e-06,
+      "loss": 0.0004,
+      "step": 17942
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.043741297354396e-06,
+      "loss": 0.0004,
+      "step": 17943
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.04134056753253e-06,
+      "loss": 0.0004,
+      "step": 17944
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.038939837710665e-06,
+      "loss": 0.0004,
+      "step": 17945
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.036539107888798e-06,
+      "loss": 0.0004,
+      "step": 17946
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.034138378066933e-06,
+      "loss": 0.0004,
+      "step": 17947
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.031737648245067e-06,
+      "loss": 0.0003,
+      "step": 17948
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0293369184232014e-06,
+      "loss": 0.9934,
+      "step": 17949
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.026936188601335e-06,
+      "loss": 0.0004,
+      "step": 17950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0245354587794684e-06,
+      "loss": 0.0004,
+      "step": 17951
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.022134728957604e-06,
+      "loss": 0.0004,
+      "step": 17952
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.019733999135737e-06,
+      "loss": 0.0004,
+      "step": 17953
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.017333269313872e-06,
+      "loss": 0.0005,
+      "step": 17954
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.014932539492006e-06,
+      "loss": 0.0004,
+      "step": 17955
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.012531809670141e-06,
+      "loss": 0.0005,
+      "step": 17956
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0101310798482745e-06,
+      "loss": 0.0004,
+      "step": 17957
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.007730350026408e-06,
+      "loss": 0.0005,
+      "step": 17958
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.005329620204542e-06,
+      "loss": 0.0005,
+      "step": 17959
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.002928890382676e-06,
+      "loss": 0.0004,
+      "step": 17960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.000528160560811e-06,
+      "loss": 0.0004,
+      "step": 17961
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9981274307389445e-06,
+      "loss": 0.0004,
+      "step": 17962
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.99572670091708e-06,
+      "loss": 0.0005,
+      "step": 17963
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.993325971095213e-06,
+      "loss": 0.0005,
+      "step": 17964
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.990925241273348e-06,
+      "loss": 0.0005,
+      "step": 17965
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.988524511451482e-06,
+      "loss": 0.0004,
+      "step": 17966
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.986123781629615e-06,
+      "loss": 0.0004,
+      "step": 17967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.98372305180775e-06,
+      "loss": 0.0005,
+      "step": 17968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.981322321985884e-06,
+      "loss": 0.0004,
+      "step": 17969
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.978921592164018e-06,
+      "loss": 0.0006,
+      "step": 17970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.976520862342152e-06,
+      "loss": 0.0004,
+      "step": 17971
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.974120132520287e-06,
+      "loss": 0.0005,
+      "step": 17972
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9717194026984206e-06,
+      "loss": 0.0005,
+      "step": 17973
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.969318672876554e-06,
+      "loss": 0.0004,
+      "step": 17974
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.966917943054689e-06,
+      "loss": 0.0005,
+      "step": 17975
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.964517213232823e-06,
+      "loss": 0.0004,
+      "step": 17976
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.962116483410958e-06,
+      "loss": 0.0004,
+      "step": 17977
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.959715753589091e-06,
+      "loss": 0.0005,
+      "step": 17978
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.957315023767226e-06,
+      "loss": 0.0005,
+      "step": 17979
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.954914293945359e-06,
+      "loss": 0.0004,
+      "step": 17980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9525135641234944e-06,
+      "loss": 0.0004,
+      "step": 17981
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.950112834301628e-06,
+      "loss": 0.0005,
+      "step": 17982
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9477121044797614e-06,
+      "loss": 0.0004,
+      "step": 17983
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.945311374657897e-06,
+      "loss": 0.0005,
+      "step": 17984
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.94291064483603e-06,
+      "loss": 0.0005,
+      "step": 17985
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.940509915014165e-06,
+      "loss": 0.0004,
+      "step": 17986
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.938109185192299e-06,
+      "loss": 0.0004,
+      "step": 17987
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.935708455370433e-06,
+      "loss": 0.0005,
+      "step": 17988
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.9333077255485675e-06,
+      "loss": 0.0004,
+      "step": 17989
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.930906995726701e-06,
+      "loss": 0.0005,
+      "step": 17990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.928506265904835e-06,
+      "loss": 0.0004,
+      "step": 17991
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.926105536082969e-06,
+      "loss": 0.0004,
+      "step": 17992
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.923704806261104e-06,
+      "loss": 0.0005,
+      "step": 17993
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.9213040764392375e-06,
+      "loss": 0.0018,
+      "step": 17994
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.918903346617373e-06,
+      "loss": 0.0004,
+      "step": 17995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.916502616795506e-06,
+      "loss": 0.0005,
+      "step": 17996
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.914101886973641e-06,
+      "loss": 0.0004,
+      "step": 17997
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.911701157151775e-06,
+      "loss": 0.0004,
+      "step": 17998
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.909300427329908e-06,
+      "loss": 0.4228,
+      "step": 17999
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.906899697508043e-06,
+      "loss": 0.0005,
+      "step": 18000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.904498967686176e-06,
+      "loss": 0.0004,
+      "step": 18001
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.902098237864311e-06,
+      "loss": 0.0004,
+      "step": 18002
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.899697508042445e-06,
+      "loss": 0.0004,
+      "step": 18003
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.89729677822058e-06,
+      "loss": 0.0004,
+      "step": 18004
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8948960483987136e-06,
+      "loss": 0.0004,
+      "step": 18005
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.892495318576847e-06,
+      "loss": 0.0005,
+      "step": 18006
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.890094588754982e-06,
+      "loss": 0.0005,
+      "step": 18007
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.887693858933116e-06,
+      "loss": 0.0005,
+      "step": 18008
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.88529312911125e-06,
+      "loss": 0.0005,
+      "step": 18009
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.882892399289384e-06,
+      "loss": 0.0004,
+      "step": 18010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.880491669467519e-06,
+      "loss": 0.0004,
+      "step": 18011
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.878090939645652e-06,
+      "loss": 0.0004,
+      "step": 18012
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8756902098237874e-06,
+      "loss": 0.0004,
+      "step": 18013
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.873289480001921e-06,
+      "loss": 0.0004,
+      "step": 18014
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8708887501800544e-06,
+      "loss": 0.0004,
+      "step": 18015
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.86848802035819e-06,
+      "loss": 0.0004,
+      "step": 18016
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.866087290536323e-06,
+      "loss": 0.0004,
+      "step": 18017
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.863686560714458e-06,
+      "loss": 0.0004,
+      "step": 18018
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.861285830892592e-06,
+      "loss": 0.0004,
+      "step": 18019
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.858885101070726e-06,
+      "loss": 0.0004,
+      "step": 18020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.85648437124886e-06,
+      "loss": 0.0004,
+      "step": 18021
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.854083641426993e-06,
+      "loss": 0.0004,
+      "step": 18022
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.851682911605128e-06,
+      "loss": 0.0004,
+      "step": 18023
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.849282181783262e-06,
+      "loss": 0.0004,
+      "step": 18024
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.846881451961397e-06,
+      "loss": 0.0004,
+      "step": 18025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8444807221395305e-06,
+      "loss": 0.0005,
+      "step": 18026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.842079992317666e-06,
+      "loss": 0.0009,
+      "step": 18027
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.839679262495799e-06,
+      "loss": 0.0004,
+      "step": 18028
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8372785326739335e-06,
+      "loss": 0.0004,
+      "step": 18029
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.834877802852067e-06,
+      "loss": 0.0004,
+      "step": 18030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.832477073030201e-06,
+      "loss": 0.0004,
+      "step": 18031
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.830076343208336e-06,
+      "loss": 0.0004,
+      "step": 18032
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.827675613386469e-06,
+      "loss": 0.0004,
+      "step": 18033
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.825274883564604e-06,
+      "loss": 0.0004,
+      "step": 18034
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.822874153742738e-06,
+      "loss": 0.0004,
+      "step": 18035
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.820473423920873e-06,
+      "loss": 0.0004,
+      "step": 18036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8180726940990065e-06,
+      "loss": 0.921,
+      "step": 18037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.81567196427714e-06,
+      "loss": 0.0004,
+      "step": 18038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.813271234455275e-06,
+      "loss": 0.0004,
+      "step": 18039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.810870504633409e-06,
+      "loss": 0.0004,
+      "step": 18040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.808469774811543e-06,
+      "loss": 0.0004,
+      "step": 18041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8060690449896766e-06,
+      "loss": 0.0004,
+      "step": 18042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.803668315167812e-06,
+      "loss": 0.0004,
+      "step": 18043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.801267585345945e-06,
+      "loss": 0.0004,
+      "step": 18044
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.7988668555240804e-06,
+      "loss": 0.0004,
+      "step": 18045
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.796466125702214e-06,
+      "loss": 0.0004,
+      "step": 18046
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.794065395880347e-06,
+      "loss": 0.0004,
+      "step": 18047
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.791664666058483e-06,
+      "loss": 0.0004,
+      "step": 18048
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.789263936236616e-06,
+      "loss": 0.0004,
+      "step": 18049
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.7868632064147504e-06,
+      "loss": 0.0004,
+      "step": 18050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.784462476592885e-06,
+      "loss": 0.0003,
+      "step": 18051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.782061746771019e-06,
+      "loss": 0.0003,
+      "step": 18052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 0.9237,
+      "step": 18053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.777260287127286e-06,
+      "loss": 0.0004,
+      "step": 18054
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.774859557305421e-06,
+      "loss": 0.0004,
+      "step": 18055
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.772458827483555e-06,
+      "loss": 0.0004,
+      "step": 18056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.77005809766169e-06,
+      "loss": 0.0005,
+      "step": 18057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.7676573678398235e-06,
+      "loss": 0.0003,
+      "step": 18058
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.765256638017959e-06,
+      "loss": 0.0004,
+      "step": 18059
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.762855908196092e-06,
+      "loss": 0.0004,
+      "step": 18060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7604551783742265e-06,
+      "loss": 1.0217,
+      "step": 18061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.75805444855236e-06,
+      "loss": 0.0004,
+      "step": 18062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7556537187304935e-06,
+      "loss": 0.0004,
+      "step": 18063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.753252988908629e-06,
+      "loss": 0.0005,
+      "step": 18064
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.750852259086762e-06,
+      "loss": 0.0005,
+      "step": 18065
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.748451529264897e-06,
+      "loss": 0.0005,
+      "step": 18066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.746050799443031e-06,
+      "loss": 0.0004,
+      "step": 18067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.743650069621166e-06,
+      "loss": 0.0005,
+      "step": 18068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7412493397992995e-06,
+      "loss": 0.0005,
+      "step": 18069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.738848609977433e-06,
+      "loss": 0.0004,
+      "step": 18070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.736447880155567e-06,
+      "loss": 0.0004,
+      "step": 18071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.734047150333702e-06,
+      "loss": 0.0005,
+      "step": 18072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.731646420511836e-06,
+      "loss": 0.0004,
+      "step": 18073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7292456906899696e-06,
+      "loss": 0.0004,
+      "step": 18074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.726844960868105e-06,
+      "loss": 0.0005,
+      "step": 18075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.724444231046238e-06,
+      "loss": 0.0005,
+      "step": 18076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7220435012243734e-06,
+      "loss": 0.0005,
+      "step": 18077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.719642771402507e-06,
+      "loss": 0.0005,
+      "step": 18078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.71724204158064e-06,
+      "loss": 0.0006,
+      "step": 18079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.714841311758776e-06,
+      "loss": 0.0004,
+      "step": 18080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.712440581936909e-06,
+      "loss": 0.0004,
+      "step": 18081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7100398521150434e-06,
+      "loss": 0.0004,
+      "step": 18082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.707639122293177e-06,
+      "loss": 0.0005,
+      "step": 18083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.705238392471312e-06,
+      "loss": 0.0004,
+      "step": 18084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.702837662649446e-06,
+      "loss": 0.0005,
+      "step": 18085
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.700436932827579e-06,
+      "loss": 0.0004,
+      "step": 18086
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.698036203005714e-06,
+      "loss": 0.0005,
+      "step": 18087
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.695635473183848e-06,
+      "loss": 0.0005,
+      "step": 18088
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.693234743361983e-06,
+      "loss": 0.0004,
+      "step": 18089
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6908340135401165e-06,
+      "loss": 0.0005,
+      "step": 18090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.688433283718251e-06,
+      "loss": 0.0004,
+      "step": 18091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.686032553896384e-06,
+      "loss": 0.0005,
+      "step": 18092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6836318240745195e-06,
+      "loss": 0.0005,
+      "step": 18093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.681231094252653e-06,
+      "loss": 0.0004,
+      "step": 18094
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6788303644307865e-06,
+      "loss": 0.0007,
+      "step": 18095
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.676429634608922e-06,
+      "loss": 0.0005,
+      "step": 18096
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.674028904787055e-06,
+      "loss": 0.0005,
+      "step": 18097
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.67162817496519e-06,
+      "loss": 0.0004,
+      "step": 18098
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.669227445143324e-06,
+      "loss": 0.0005,
+      "step": 18099
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.666826715321459e-06,
+      "loss": 0.0004,
+      "step": 18100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6644259854995925e-06,
+      "loss": 0.0004,
+      "step": 18101
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.662025255677726e-06,
+      "loss": 0.0005,
+      "step": 18102
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.65962452585586e-06,
+      "loss": 0.0004,
+      "step": 18103
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.657223796033994e-06,
+      "loss": 0.0005,
+      "step": 18104
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.654823066212129e-06,
+      "loss": 0.0004,
+      "step": 18105
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6524223363902626e-06,
+      "loss": 0.0004,
+      "step": 18106
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.650021606568398e-06,
+      "loss": 0.0004,
+      "step": 18107
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.647620876746531e-06,
+      "loss": 0.0005,
+      "step": 18108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6452201469246664e-06,
+      "loss": 0.0005,
+      "step": 18109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6428194171028e-06,
+      "loss": 0.0004,
+      "step": 18110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.640418687280933e-06,
+      "loss": 0.0004,
+      "step": 18111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.638017957459068e-06,
+      "loss": 0.0004,
+      "step": 18112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.635617227637201e-06,
+      "loss": 0.0004,
+      "step": 18113
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6332164978153364e-06,
+      "loss": 0.0004,
+      "step": 18114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.63081576799347e-06,
+      "loss": 0.0004,
+      "step": 18115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.628415038171605e-06,
+      "loss": 0.0005,
+      "step": 18116
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.626014308349739e-06,
+      "loss": 0.0004,
+      "step": 18117
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.623613578527872e-06,
+      "loss": 0.4819,
+      "step": 18118
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.621212848706007e-06,
+      "loss": 0.0004,
+      "step": 18119
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.618812118884141e-06,
+      "loss": 0.0004,
+      "step": 18120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.616411389062276e-06,
+      "loss": 0.0004,
+      "step": 18121
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6140106592404095e-06,
+      "loss": 0.0004,
+      "step": 18122
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.611609929418544e-06,
+      "loss": 0.0004,
+      "step": 18123
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.609209199596677e-06,
+      "loss": 0.0004,
+      "step": 18124
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6068084697748125e-06,
+      "loss": 0.0005,
+      "step": 18125
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.604407739952946e-06,
+      "loss": 0.0005,
+      "step": 18126
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.6020070101310795e-06,
+      "loss": 0.0005,
+      "step": 18127
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.599606280309215e-06,
+      "loss": 0.0004,
+      "step": 18128
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.597205550487348e-06,
+      "loss": 0.0004,
+      "step": 18129
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.594804820665483e-06,
+      "loss": 0.0004,
+      "step": 18130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.592404090843617e-06,
+      "loss": 0.0004,
+      "step": 18131
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.590003361021751e-06,
+      "loss": 0.0049,
+      "step": 18132
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.587602631199885e-06,
+      "loss": 0.0004,
+      "step": 18133
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.585201901378019e-06,
+      "loss": 0.0005,
+      "step": 18134
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.582801171556153e-06,
+      "loss": 0.0004,
+      "step": 18135
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.580400441734287e-06,
+      "loss": 0.0004,
+      "step": 18136
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.577999711912422e-06,
+      "loss": 0.0004,
+      "step": 18137
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5755989820905555e-06,
+      "loss": 0.0004,
+      "step": 18138
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.573198252268691e-06,
+      "loss": 0.0004,
+      "step": 18139
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.570797522446824e-06,
+      "loss": 0.0004,
+      "step": 18140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5683967926249586e-06,
+      "loss": 0.0005,
+      "step": 18141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.565996062803093e-06,
+      "loss": 0.0004,
+      "step": 18142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.563595332981226e-06,
+      "loss": 0.0004,
+      "step": 18143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.561194603159361e-06,
+      "loss": 0.0005,
+      "step": 18144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.558793873337494e-06,
+      "loss": 0.0004,
+      "step": 18145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5563931435156294e-06,
+      "loss": 0.0004,
+      "step": 18146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.553992413693763e-06,
+      "loss": 0.0004,
+      "step": 18147
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.551591683871898e-06,
+      "loss": 0.0005,
+      "step": 18148
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.549190954050032e-06,
+      "loss": 0.003,
+      "step": 18149
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.546790224228165e-06,
+      "loss": 0.0004,
+      "step": 18150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5443894944063e-06,
+      "loss": 0.0004,
+      "step": 18151
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.541988764584434e-06,
+      "loss": 0.0004,
+      "step": 18152
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.539588034762568e-06,
+      "loss": 0.0004,
+      "step": 18153
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.537187304940702e-06,
+      "loss": 0.0004,
+      "step": 18154
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.534786575118837e-06,
+      "loss": 0.0004,
+      "step": 18155
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.53238584529697e-06,
+      "loss": 0.0005,
+      "step": 18156
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.529985115475104e-06,
+      "loss": 0.0004,
+      "step": 18157
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.527584385653239e-06,
+      "loss": 0.0004,
+      "step": 18158
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5251836558313725e-06,
+      "loss": 0.0004,
+      "step": 18159
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.522782926009508e-06,
+      "loss": 0.0004,
+      "step": 18160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.520382196187641e-06,
+      "loss": 0.0004,
+      "step": 18161
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.517981466365776e-06,
+      "loss": 0.0003,
+      "step": 18162
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.51558073654391e-06,
+      "loss": 0.0004,
+      "step": 18163
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.513180006722044e-06,
+      "loss": 0.0004,
+      "step": 18164
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.510779276900178e-06,
+      "loss": 0.0004,
+      "step": 18165
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.508378547078311e-06,
+      "loss": 0.0004,
+      "step": 18166
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.505977817256446e-06,
+      "loss": 0.0004,
+      "step": 18167
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.50357708743458e-06,
+      "loss": 0.0003,
+      "step": 18168
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.501176357612715e-06,
+      "loss": 0.0004,
+      "step": 18169
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4987756277908485e-06,
+      "loss": 0.0004,
+      "step": 18170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.496374897968984e-06,
+      "loss": 0.0004,
+      "step": 18171
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.493974168147117e-06,
+      "loss": 0.0004,
+      "step": 18172
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.491573438325251e-06,
+      "loss": 0.0003,
+      "step": 18173
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.489172708503385e-06,
+      "loss": 0.0029,
+      "step": 18174
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4867719786815186e-06,
+      "loss": 0.0004,
+      "step": 18175
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.484371248859654e-06,
+      "loss": 0.0004,
+      "step": 18176
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.481970519037787e-06,
+      "loss": 0.0004,
+      "step": 18177
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4795697892159224e-06,
+      "loss": 0.0004,
+      "step": 18178
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.477169059394056e-06,
+      "loss": 0.0004,
+      "step": 18179
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.474768329572191e-06,
+      "loss": 0.0004,
+      "step": 18180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.472367599750325e-06,
+      "loss": 0.6689,
+      "step": 18181
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.469966869928458e-06,
+      "loss": 0.0004,
+      "step": 18182
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.467566140106593e-06,
+      "loss": 0.0004,
+      "step": 18183
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.465165410284727e-06,
+      "loss": 0.0004,
+      "step": 18184
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.462764680462861e-06,
+      "loss": 0.0004,
+      "step": 18185
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.460363950640995e-06,
+      "loss": 0.0004,
+      "step": 18186
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.45796322081913e-06,
+      "loss": 0.0004,
+      "step": 18187
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.455562490997263e-06,
+      "loss": 0.0004,
+      "step": 18188
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.453161761175397e-06,
+      "loss": 0.0004,
+      "step": 18189
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.450761031353532e-06,
+      "loss": 0.0004,
+      "step": 18190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4483603015316655e-06,
+      "loss": 0.0003,
+      "step": 18191
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.445959571709801e-06,
+      "loss": 0.0004,
+      "step": 18192
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.443558841887934e-06,
+      "loss": 0.0004,
+      "step": 18193
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4411581120660685e-06,
+      "loss": 0.0004,
+      "step": 18194
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.438757382244202e-06,
+      "loss": 0.0004,
+      "step": 18195
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.436356652422337e-06,
+      "loss": 0.0004,
+      "step": 18196
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.433955922600471e-06,
+      "loss": 0.0004,
+      "step": 18197
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.431555192778604e-06,
+      "loss": 0.0003,
+      "step": 18198
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.429154462956739e-06,
+      "loss": 0.0034,
+      "step": 18199
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.426753733134873e-06,
+      "loss": 0.0004,
+      "step": 18200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.424353003313008e-06,
+      "loss": 0.0004,
+      "step": 18201
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4219522734911415e-06,
+      "loss": 0.0004,
+      "step": 18202
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.419551543669276e-06,
+      "loss": 0.0004,
+      "step": 18203
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.41715081384741e-06,
+      "loss": 0.0004,
+      "step": 18204
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.414750084025544e-06,
+      "loss": 0.0004,
+      "step": 18205
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.412349354203678e-06,
+      "loss": 0.0004,
+      "step": 18206
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4099486243818116e-06,
+      "loss": 0.0003,
+      "step": 18207
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.407547894559947e-06,
+      "loss": 0.0004,
+      "step": 18208
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.40514716473808e-06,
+      "loss": 0.0004,
+      "step": 18209
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4027464349162154e-06,
+      "loss": 0.0004,
+      "step": 18210
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.400345705094349e-06,
+      "loss": 0.0003,
+      "step": 18211
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.397944975272484e-06,
+      "loss": 0.0068,
+      "step": 18212
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.395544245450618e-06,
+      "loss": 0.0004,
+      "step": 18213
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.393143515628751e-06,
+      "loss": 0.0004,
+      "step": 18214
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3907427858068854e-06,
+      "loss": 0.0005,
+      "step": 18215
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.388342055985019e-06,
+      "loss": 0.0003,
+      "step": 18216
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.385941326163154e-06,
+      "loss": 0.0004,
+      "step": 18217
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.383540596341288e-06,
+      "loss": 0.0003,
+      "step": 18218
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.381139866519423e-06,
+      "loss": 0.0003,
+      "step": 18219
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.378739136697556e-06,
+      "loss": 0.0003,
+      "step": 18220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.37633840687569e-06,
+      "loss": 0.0004,
+      "step": 18221
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.373937677053825e-06,
+      "loss": 0.0003,
+      "step": 18222
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3715369472319585e-06,
+      "loss": 0.0003,
+      "step": 18223
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.369136217410094e-06,
+      "loss": 0.0004,
+      "step": 18224
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.366735487588227e-06,
+      "loss": 0.0004,
+      "step": 18225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3643347577663615e-06,
+      "loss": 0.0003,
+      "step": 18226
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.361934027944495e-06,
+      "loss": 0.0003,
+      "step": 18227
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.35953329812263e-06,
+      "loss": 0.0003,
+      "step": 18228
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.357132568300764e-06,
+      "loss": 0.0004,
+      "step": 18229
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.354731838478897e-06,
+      "loss": 0.0004,
+      "step": 18230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.352331108657032e-06,
+      "loss": 0.0003,
+      "step": 18231
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.349930378835166e-06,
+      "loss": 0.0003,
+      "step": 18232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.347529649013301e-06,
+      "loss": 0.0004,
+      "step": 18233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3451289191914345e-06,
+      "loss": 0.0004,
+      "step": 18234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.342728189369569e-06,
+      "loss": 0.0003,
+      "step": 18235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.340327459547702e-06,
+      "loss": 0.0004,
+      "step": 18236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.337926729725836e-06,
+      "loss": 0.0004,
+      "step": 18237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.335525999903971e-06,
+      "loss": 0.0004,
+      "step": 18238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3331252700821046e-06,
+      "loss": 0.0003,
+      "step": 18239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.33072454026024e-06,
+      "loss": 0.0004,
+      "step": 18240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.328323810438373e-06,
+      "loss": 0.0004,
+      "step": 18241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.325923080616508e-06,
+      "loss": 0.0003,
+      "step": 18242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.323522350794642e-06,
+      "loss": 0.0003,
+      "step": 18243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.321121620972776e-06,
+      "loss": 0.0004,
+      "step": 18244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.318720891150911e-06,
+      "loss": 0.0003,
+      "step": 18245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.316320161329044e-06,
+      "loss": 0.0003,
+      "step": 18246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3139194315071784e-06,
+      "loss": 0.0003,
+      "step": 18247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.311518701685312e-06,
+      "loss": 0.0004,
+      "step": 18248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.309117971863447e-06,
+      "loss": 0.0003,
+      "step": 18249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.306717242041581e-06,
+      "loss": 0.0003,
+      "step": 18250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.304316512219716e-06,
+      "loss": 0.0004,
+      "step": 18251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.301915782397849e-06,
+      "loss": 0.0003,
+      "step": 18252
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.299515052575983e-06,
+      "loss": 0.0003,
+      "step": 18253
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.297114322754118e-06,
+      "loss": 0.5113,
+      "step": 18254
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2947135929322515e-06,
+      "loss": 0.0003,
+      "step": 18255
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.292312863110386e-06,
+      "loss": 0.0003,
+      "step": 18256
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.289912133288519e-06,
+      "loss": 0.0004,
+      "step": 18257
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2875114034666545e-06,
+      "loss": 0.0003,
+      "step": 18258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.285110673644788e-06,
+      "loss": 0.0003,
+      "step": 18259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.282709943822923e-06,
+      "loss": 0.0003,
+      "step": 18260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.280309214001057e-06,
+      "loss": 0.0003,
+      "step": 18261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.27790848417919e-06,
+      "loss": 0.0003,
+      "step": 18262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.275507754357325e-06,
+      "loss": 0.0003,
+      "step": 18263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.273107024535459e-06,
+      "loss": 0.0067,
+      "step": 18264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.270706294713593e-06,
+      "loss": 0.0003,
+      "step": 18265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2683055648917275e-06,
+      "loss": 0.0004,
+      "step": 18266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.265904835069862e-06,
+      "loss": 0.0003,
+      "step": 18267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.263504105247995e-06,
+      "loss": 0.0003,
+      "step": 18268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.261103375426129e-06,
+      "loss": 0.0003,
+      "step": 18269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.258702645604264e-06,
+      "loss": 0.0003,
+      "step": 18270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2563019157823975e-06,
+      "loss": 0.0004,
+      "step": 18271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.253901185960533e-06,
+      "loss": 0.0003,
+      "step": 18272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.251500456138666e-06,
+      "loss": 0.0003,
+      "step": 18273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2490997263168006e-06,
+      "loss": 0.0003,
+      "step": 18274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.246698996494935e-06,
+      "loss": 0.0003,
+      "step": 18275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.244298266673069e-06,
+      "loss": 0.0004,
+      "step": 18276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.241897536851203e-06,
+      "loss": 0.0003,
+      "step": 18277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.239496807029337e-06,
+      "loss": 0.0003,
+      "step": 18278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2370960772074714e-06,
+      "loss": 0.0003,
+      "step": 18279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.234695347385606e-06,
+      "loss": 0.0003,
+      "step": 18280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.232294617563739e-06,
+      "loss": 0.0005,
+      "step": 18281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.229893887741874e-06,
+      "loss": 0.0003,
+      "step": 18282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.227493157920008e-06,
+      "loss": 0.0003,
+      "step": 18283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.225092428098142e-06,
+      "loss": 0.0028,
+      "step": 18284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.222691698276277e-06,
+      "loss": 0.0003,
+      "step": 18285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.220290968454411e-06,
+      "loss": 0.0003,
+      "step": 18286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.2178902386325445e-06,
+      "loss": 0.0004,
+      "step": 18287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.215489508810679e-06,
+      "loss": 0.0003,
+      "step": 18288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.213088778988812e-06,
+      "loss": 0.0051,
+      "step": 18289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.210688049166947e-06,
+      "loss": 0.0003,
+      "step": 18290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.208287319345081e-06,
+      "loss": 0.0003,
+      "step": 18291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.205886589523215e-06,
+      "loss": 0.0004,
+      "step": 18292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.20348585970135e-06,
+      "loss": 0.0035,
+      "step": 18293
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.201085129879484e-06,
+      "loss": 0.0003,
+      "step": 18294
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.198684400057618e-06,
+      "loss": 0.0003,
+      "step": 18295
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.196283670235752e-06,
+      "loss": 0.0003,
+      "step": 18296
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.193882940413886e-06,
+      "loss": 0.0003,
+      "step": 18297
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.19148221059202e-06,
+      "loss": 0.0003,
+      "step": 18298
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.189081480770154e-06,
+      "loss": 0.0003,
+      "step": 18299
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.186680750948288e-06,
+      "loss": 0.0003,
+      "step": 18300
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.184280021126423e-06,
+      "loss": 0.0003,
+      "step": 18301
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.181879291304557e-06,
+      "loss": 0.0003,
+      "step": 18302
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.179478561482691e-06,
+      "loss": 0.0003,
+      "step": 18303
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.177077831660826e-06,
+      "loss": 0.0003,
+      "step": 18304
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.174677101838959e-06,
+      "loss": 0.0003,
+      "step": 18305
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1722763720170936e-06,
+      "loss": 0.0003,
+      "step": 18306
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.169875642195228e-06,
+      "loss": 0.0003,
+      "step": 18307
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.167474912373361e-06,
+      "loss": 0.0003,
+      "step": 18308
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.165074182551496e-06,
+      "loss": 0.0003,
+      "step": 18309
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.16267345272963e-06,
+      "loss": 0.4785,
+      "step": 18310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1602727229077644e-06,
+      "loss": 0.0003,
+      "step": 18311
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.157871993085899e-06,
+      "loss": 1.0268,
+      "step": 18312
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.155471263264032e-06,
+      "loss": 0.0003,
+      "step": 18313
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.153070533442167e-06,
+      "loss": 0.0003,
+      "step": 18314
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.150669803620301e-06,
+      "loss": 0.0004,
+      "step": 18315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.148269073798435e-06,
+      "loss": 0.0004,
+      "step": 18316
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.14586834397657e-06,
+      "loss": 0.0003,
+      "step": 18317
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.143467614154703e-06,
+      "loss": 0.0003,
+      "step": 18318
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1410668843328375e-06,
+      "loss": 0.0003,
+      "step": 18319
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.138666154510972e-06,
+      "loss": 0.0003,
+      "step": 18320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.136265424689105e-06,
+      "loss": 0.0003,
+      "step": 18321
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.13386469486724e-06,
+      "loss": 0.0004,
+      "step": 18322
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.131463965045374e-06,
+      "loss": 0.0004,
+      "step": 18323
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.129063235223508e-06,
+      "loss": 0.0003,
+      "step": 18324
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.126662505401643e-06,
+      "loss": 0.0005,
+      "step": 18325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.124261775579777e-06,
+      "loss": 0.0003,
+      "step": 18326
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1218610457579105e-06,
+      "loss": 0.0003,
+      "step": 18327
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.119460315936045e-06,
+      "loss": 0.0004,
+      "step": 18328
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.117059586114178e-06,
+      "loss": 0.0003,
+      "step": 18329
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.114658856292313e-06,
+      "loss": 0.0004,
+      "step": 18330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.112258126470447e-06,
+      "loss": 0.0004,
+      "step": 18331
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.109857396648581e-06,
+      "loss": 0.0003,
+      "step": 18332
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.107456666826716e-06,
+      "loss": 0.0003,
+      "step": 18333
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.10505593700485e-06,
+      "loss": 0.0004,
+      "step": 18334
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.102655207182984e-06,
+      "loss": 0.0004,
+      "step": 18335
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.100254477361119e-06,
+      "loss": 0.0003,
+      "step": 18336
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.097853747539252e-06,
+      "loss": 0.0003,
+      "step": 18337
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0954530177173866e-06,
+      "loss": 0.0004,
+      "step": 18338
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.09305228789552e-06,
+      "loss": 0.0003,
+      "step": 18339
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.090651558073654e-06,
+      "loss": 0.0004,
+      "step": 18340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.088250828251789e-06,
+      "loss": 0.0003,
+      "step": 18341
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.085850098429923e-06,
+      "loss": 0.0003,
+      "step": 18342
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0834493686080574e-06,
+      "loss": 0.0003,
+      "step": 18343
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.081048638786192e-06,
+      "loss": 0.0004,
+      "step": 18344
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.078647908964325e-06,
+      "loss": 0.0003,
+      "step": 18345
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.07624717914246e-06,
+      "loss": 0.0003,
+      "step": 18346
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.073846449320594e-06,
+      "loss": 0.0003,
+      "step": 18347
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.071445719498728e-06,
+      "loss": 0.0003,
+      "step": 18348
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.069044989676862e-06,
+      "loss": 0.0003,
+      "step": 18349
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.066644259854996e-06,
+      "loss": 0.0003,
+      "step": 18350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0642435300331305e-06,
+      "loss": 0.0004,
+      "step": 18351
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.061842800211265e-06,
+      "loss": 0.0003,
+      "step": 18352
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.059442070389398e-06,
+      "loss": 0.0003,
+      "step": 18353
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.057041340567533e-06,
+      "loss": 0.0003,
+      "step": 18354
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.054640610745667e-06,
+      "loss": 0.0003,
+      "step": 18355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.052239880923801e-06,
+      "loss": 0.0003,
+      "step": 18356
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.049839151101936e-06,
+      "loss": 0.0003,
+      "step": 18357
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.047438421280069e-06,
+      "loss": 0.0003,
+      "step": 18358
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.0450376914582035e-06,
+      "loss": 0.0004,
+      "step": 18359
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.042636961636338e-06,
+      "loss": 0.0003,
+      "step": 18360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.040236231814471e-06,
+      "loss": 0.0004,
+      "step": 18361
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.037835501992606e-06,
+      "loss": 0.0003,
+      "step": 18362
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.03543477217074e-06,
+      "loss": 0.0003,
+      "step": 18363
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.033034042348874e-06,
+      "loss": 0.0003,
+      "step": 18364
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.030633312527009e-06,
+      "loss": 0.0003,
+      "step": 18365
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.028232582705143e-06,
+      "loss": 0.0003,
+      "step": 18366
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.025831852883277e-06,
+      "loss": 0.0003,
+      "step": 18367
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.023431123061411e-06,
+      "loss": 0.0004,
+      "step": 18368
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.021030393239545e-06,
+      "loss": 0.0003,
+      "step": 18369
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.018629663417679e-06,
+      "loss": 0.0003,
+      "step": 18370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.016228933595813e-06,
+      "loss": 0.0003,
+      "step": 18371
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.013828203773947e-06,
+      "loss": 0.0003,
+      "step": 18372
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.011427473952082e-06,
+      "loss": 0.0035,
+      "step": 18373
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.009026744130216e-06,
+      "loss": 0.0004,
+      "step": 18374
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.00662601430835e-06,
+      "loss": 0.0003,
+      "step": 18375
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.004225284486485e-06,
+      "loss": 0.0003,
+      "step": 18376
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.001824554664618e-06,
+      "loss": 0.0004,
+      "step": 18377
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.999423824842753e-06,
+      "loss": 0.0004,
+      "step": 18378
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.997023095020887e-06,
+      "loss": 0.0004,
+      "step": 18379
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9946223651990204e-06,
+      "loss": 0.0003,
+      "step": 18380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.992221635377155e-06,
+      "loss": 0.0003,
+      "step": 18381
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.989820905555289e-06,
+      "loss": 0.0003,
+      "step": 18382
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9874201757334235e-06,
+      "loss": 0.0003,
+      "step": 18383
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.985019445911558e-06,
+      "loss": 0.0003,
+      "step": 18384
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.982618716089691e-06,
+      "loss": 0.0003,
+      "step": 18385
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.980217986267826e-06,
+      "loss": 0.0003,
+      "step": 18386
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.97781725644596e-06,
+      "loss": 0.0003,
+      "step": 18387
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.975416526624094e-06,
+      "loss": 0.0003,
+      "step": 18388
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.973015796802228e-06,
+      "loss": 0.0003,
+      "step": 18389
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.970615066980362e-06,
+      "loss": 0.0003,
+      "step": 18390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9682143371584965e-06,
+      "loss": 0.0003,
+      "step": 18391
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.965813607336631e-06,
+      "loss": 0.0003,
+      "step": 18392
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.963412877514764e-06,
+      "loss": 0.0003,
+      "step": 18393
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.961012147692899e-06,
+      "loss": 0.0003,
+      "step": 18394
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.958611417871033e-06,
+      "loss": 0.0003,
+      "step": 18395
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.956210688049167e-06,
+      "loss": 0.0003,
+      "step": 18396
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.953809958227302e-06,
+      "loss": 0.0003,
+      "step": 18397
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.951409228405436e-06,
+      "loss": 0.0003,
+      "step": 18398
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9490084985835695e-06,
+      "loss": 0.0003,
+      "step": 18399
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.946607768761704e-06,
+      "loss": 0.0003,
+      "step": 18400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.944207038939837e-06,
+      "loss": 0.0003,
+      "step": 18401
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.941806309117972e-06,
+      "loss": 0.0003,
+      "step": 18402
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.939405579296106e-06,
+      "loss": 1.0217,
+      "step": 18403
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.93700484947424e-06,
+      "loss": 0.0003,
+      "step": 18404
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.934604119652375e-06,
+      "loss": 0.0003,
+      "step": 18405
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.932203389830509e-06,
+      "loss": 0.0003,
+      "step": 18406
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.929802660008643e-06,
+      "loss": 0.0003,
+      "step": 18407
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.927401930186778e-06,
+      "loss": 0.0003,
+      "step": 18408
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.925001200364911e-06,
+      "loss": 0.0003,
+      "step": 18409
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.922600470543046e-06,
+      "loss": 0.0003,
+      "step": 18410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.920199740721179e-06,
+      "loss": 0.0003,
+      "step": 18411
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9177990108993134e-06,
+      "loss": 0.0003,
+      "step": 18412
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.915398281077448e-06,
+      "loss": 0.0003,
+      "step": 18413
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.912997551255582e-06,
+      "loss": 0.0003,
+      "step": 18414
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9105968214337165e-06,
+      "loss": 0.0003,
+      "step": 18415
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.908196091611851e-06,
+      "loss": 0.0003,
+      "step": 18416
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.905795361789984e-06,
+      "loss": 0.0003,
+      "step": 18417
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.903394631968119e-06,
+      "loss": 0.0003,
+      "step": 18418
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.900993902146253e-06,
+      "loss": 0.0003,
+      "step": 18419
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8985931723243865e-06,
+      "loss": 0.0003,
+      "step": 18420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.896192442502521e-06,
+      "loss": 0.0003,
+      "step": 18421
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.893791712680655e-06,
+      "loss": 0.0004,
+      "step": 18422
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8913909828587895e-06,
+      "loss": 0.0004,
+      "step": 18423
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.888990253036924e-06,
+      "loss": 0.0003,
+      "step": 18424
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.886589523215057e-06,
+      "loss": 0.0003,
+      "step": 18425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.884188793393192e-06,
+      "loss": 0.0004,
+      "step": 18426
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.881788063571326e-06,
+      "loss": 0.0003,
+      "step": 18427
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.87938733374946e-06,
+      "loss": 0.0003,
+      "step": 18428
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.876986603927595e-06,
+      "loss": 0.0003,
+      "step": 18429
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.874585874105728e-06,
+      "loss": 0.0004,
+      "step": 18430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8721851442838625e-06,
+      "loss": 0.0003,
+      "step": 18431
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.869784414461997e-06,
+      "loss": 0.0004,
+      "step": 18432
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.86738368464013e-06,
+      "loss": 0.0003,
+      "step": 18433
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.864982954818265e-06,
+      "loss": 0.0003,
+      "step": 18434
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.862582224996399e-06,
+      "loss": 0.0003,
+      "step": 18435
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.860181495174533e-06,
+      "loss": 0.0003,
+      "step": 18436
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.857780765352668e-06,
+      "loss": 0.0004,
+      "step": 18437
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.855380035530802e-06,
+      "loss": 0.0004,
+      "step": 18438
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.852979305708936e-06,
+      "loss": 0.0003,
+      "step": 18439
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.85057857588707e-06,
+      "loss": 0.0004,
+      "step": 18440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.848177846065204e-06,
+      "loss": 0.0004,
+      "step": 18441
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.845777116243338e-06,
+      "loss": 0.0004,
+      "step": 18442
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.843376386421472e-06,
+      "loss": 0.3356,
+      "step": 18443
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8409756565996064e-06,
+      "loss": 0.0003,
+      "step": 18444
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.838574926777741e-06,
+      "loss": 0.0003,
+      "step": 18445
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.836174196955875e-06,
+      "loss": 0.0003,
+      "step": 18446
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8337734671340094e-06,
+      "loss": 0.0003,
+      "step": 18447
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.831372737312144e-06,
+      "loss": 0.0003,
+      "step": 18448
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.828972007490277e-06,
+      "loss": 0.0003,
+      "step": 18449
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.826571277668412e-06,
+      "loss": 0.0004,
+      "step": 18450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.824170547846545e-06,
+      "loss": 0.0003,
+      "step": 18451
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8217698180246795e-06,
+      "loss": 0.0004,
+      "step": 18452
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.819369088202814e-06,
+      "loss": 0.0003,
+      "step": 18453
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.816968358380948e-06,
+      "loss": 0.0003,
+      "step": 18454
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.8145676285590825e-06,
+      "loss": 0.0003,
+      "step": 18455
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.812166898737217e-06,
+      "loss": 0.0003,
+      "step": 18456
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.80976616891535e-06,
+      "loss": 0.0003,
+      "step": 18457
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.807365439093485e-06,
+      "loss": 0.0003,
+      "step": 18458
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.804964709271619e-06,
+      "loss": 0.0004,
+      "step": 18459
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.802563979449753e-06,
+      "loss": 0.0004,
+      "step": 18460
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.800163249627887e-06,
+      "loss": 0.0003,
+      "step": 18461
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.797762519806021e-06,
+      "loss": 0.0003,
+      "step": 18462
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.7953617899841555e-06,
+      "loss": 0.0004,
+      "step": 18463
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.79296106016229e-06,
+      "loss": 0.0003,
+      "step": 18464
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.790560330340423e-06,
+      "loss": 0.0004,
+      "step": 18465
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.788159600518558e-06,
+      "loss": 0.0003,
+      "step": 18466
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.785758870696692e-06,
+      "loss": 0.0003,
+      "step": 18467
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.783358140874826e-06,
+      "loss": 0.0003,
+      "step": 18468
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.780957411052961e-06,
+      "loss": 0.0003,
+      "step": 18469
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.778556681231095e-06,
+      "loss": 0.0004,
+      "step": 18470
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.7761559514092286e-06,
+      "loss": 0.0003,
+      "step": 18471
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.773755221587363e-06,
+      "loss": 0.4937,
+      "step": 18472
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.771354491765496e-06,
+      "loss": 0.0003,
+      "step": 18473
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.768953761943631e-06,
+      "loss": 0.0003,
+      "step": 18474
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.766553032121765e-06,
+      "loss": 0.0003,
+      "step": 18475
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.764152302299899e-06,
+      "loss": 0.4495,
+      "step": 18476
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.761751572478034e-06,
+      "loss": 0.0003,
+      "step": 18477
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.759350842656168e-06,
+      "loss": 0.0022,
+      "step": 18478
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7569501128343024e-06,
+      "loss": 0.0003,
+      "step": 18479
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.754549383012437e-06,
+      "loss": 0.0003,
+      "step": 18480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.75214865319057e-06,
+      "loss": 0.0003,
+      "step": 18481
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.749747923368704e-06,
+      "loss": 0.0003,
+      "step": 18482
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.747347193546838e-06,
+      "loss": 0.0004,
+      "step": 18483
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7449464637249725e-06,
+      "loss": 0.0003,
+      "step": 18484
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.742545733903107e-06,
+      "loss": 0.0003,
+      "step": 18485
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.740145004081241e-06,
+      "loss": 0.0003,
+      "step": 18486
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7377442742593755e-06,
+      "loss": 0.0003,
+      "step": 18487
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.735343544437509e-06,
+      "loss": 0.0004,
+      "step": 18488
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.732942814615643e-06,
+      "loss": 0.0003,
+      "step": 18489
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.730542084793778e-06,
+      "loss": 0.0003,
+      "step": 18490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.728141354971912e-06,
+      "loss": 0.0003,
+      "step": 18491
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7257406251500455e-06,
+      "loss": 0.0003,
+      "step": 18492
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.72333989532818e-06,
+      "loss": 0.0003,
+      "step": 18493
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.720939165506314e-06,
+      "loss": 0.0003,
+      "step": 18494
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7185384356844485e-06,
+      "loss": 0.0003,
+      "step": 18495
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.716137705862582e-06,
+      "loss": 0.0004,
+      "step": 18496
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.713736976040716e-06,
+      "loss": 0.0002,
+      "step": 18497
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.711336246218851e-06,
+      "loss": 0.0003,
+      "step": 18498
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.708935516396985e-06,
+      "loss": 0.0003,
+      "step": 18499
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.706534786575119e-06,
+      "loss": 0.0016,
+      "step": 18500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.704134056753254e-06,
+      "loss": 0.0003,
+      "step": 18501
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.701733326931387e-06,
+      "loss": 0.0003,
+      "step": 18502
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6993325971095216e-06,
+      "loss": 0.0004,
+      "step": 18503
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.696931867287655e-06,
+      "loss": 0.0003,
+      "step": 18504
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.694531137465789e-06,
+      "loss": 0.0003,
+      "step": 18505
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.692130407643924e-06,
+      "loss": 0.0003,
+      "step": 18506
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.689729677822058e-06,
+      "loss": 0.0003,
+      "step": 18507
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.687328948000192e-06,
+      "loss": 0.0003,
+      "step": 18508
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.684928218178327e-06,
+      "loss": 0.0003,
+      "step": 18509
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.682527488356461e-06,
+      "loss": 0.0004,
+      "step": 18510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6801267585345954e-06,
+      "loss": 0.0003,
+      "step": 18511
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.677726028712729e-06,
+      "loss": 0.0003,
+      "step": 18512
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6753252988908624e-06,
+      "loss": 0.0003,
+      "step": 18513
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.672924569068997e-06,
+      "loss": 0.0003,
+      "step": 18514
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.670523839247131e-06,
+      "loss": 0.0003,
+      "step": 18515
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6681231094252655e-06,
+      "loss": 0.0004,
+      "step": 18516
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6657223796034e-06,
+      "loss": 0.0003,
+      "step": 18517
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.663321649781534e-06,
+      "loss": 0.0003,
+      "step": 18518
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6609209199596685e-06,
+      "loss": 0.0003,
+      "step": 18519
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.658520190137802e-06,
+      "loss": 0.0003,
+      "step": 18520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.656119460315936e-06,
+      "loss": 0.0003,
+      "step": 18521
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.653718730494071e-06,
+      "loss": 0.0003,
+      "step": 18522
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.651318000672204e-06,
+      "loss": 0.0003,
+      "step": 18523
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6489172708503385e-06,
+      "loss": 0.0003,
+      "step": 18524
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.646516541028473e-06,
+      "loss": 0.0003,
+      "step": 18525
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.644115811206607e-06,
+      "loss": 0.0003,
+      "step": 18526
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6417150813847415e-06,
+      "loss": 0.8742,
+      "step": 18527
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.639314351562875e-06,
+      "loss": 0.0003,
+      "step": 18528
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.636913621741009e-06,
+      "loss": 0.0003,
+      "step": 18529
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.634512891919144e-06,
+      "loss": 0.0003,
+      "step": 18530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.632112162097278e-06,
+      "loss": 0.0003,
+      "step": 18531
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.629711432275412e-06,
+      "loss": 0.0003,
+      "step": 18532
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.627310702453546e-06,
+      "loss": 0.0003,
+      "step": 18533
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.62490997263168e-06,
+      "loss": 0.0003,
+      "step": 18534
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6225092428098146e-06,
+      "loss": 0.0004,
+      "step": 18535
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.620108512987948e-06,
+      "loss": 0.0003,
+      "step": 18536
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.617707783166082e-06,
+      "loss": 0.0003,
+      "step": 18537
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.615307053344217e-06,
+      "loss": 0.0003,
+      "step": 18538
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.612906323522351e-06,
+      "loss": 0.0003,
+      "step": 18539
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.610505593700485e-06,
+      "loss": 0.0003,
+      "step": 18540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.60810486387862e-06,
+      "loss": 0.0003,
+      "step": 18541
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.605704134056754e-06,
+      "loss": 0.0003,
+      "step": 18542
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.603303404234888e-06,
+      "loss": 0.0002,
+      "step": 18543
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.600902674413021e-06,
+      "loss": 0.0003,
+      "step": 18544
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.5985019445911554e-06,
+      "loss": 0.0004,
+      "step": 18545
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.59610121476929e-06,
+      "loss": 0.0019,
+      "step": 18546
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.593700484947424e-06,
+      "loss": 0.0003,
+      "step": 18547
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5912997551255585e-06,
+      "loss": 0.0003,
+      "step": 18548
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.588899025303693e-06,
+      "loss": 0.0002,
+      "step": 18549
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.586498295481827e-06,
+      "loss": 0.0003,
+      "step": 18550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5840975656599615e-06,
+      "loss": 0.0003,
+      "step": 18551
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.581696835838095e-06,
+      "loss": 0.0003,
+      "step": 18552
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.579296106016229e-06,
+      "loss": 0.0003,
+      "step": 18553
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.576895376194363e-06,
+      "loss": 0.0003,
+      "step": 18554
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.574494646372497e-06,
+      "loss": 0.5581,
+      "step": 18555
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5720939165506315e-06,
+      "loss": 0.0003,
+      "step": 18556
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.569693186728766e-06,
+      "loss": 0.0003,
+      "step": 18557
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5672924569069e-06,
+      "loss": 0.0003,
+      "step": 18558
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5648917270850345e-06,
+      "loss": 0.0003,
+      "step": 18559
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.562490997263168e-06,
+      "loss": 0.0003,
+      "step": 18560
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.560090267441302e-06,
+      "loss": 1.0315,
+      "step": 18561
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.557689537619437e-06,
+      "loss": 0.0003,
+      "step": 18562
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.555288807797571e-06,
+      "loss": 0.0003,
+      "step": 18563
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5528880779757045e-06,
+      "loss": 0.0004,
+      "step": 18564
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.550487348153839e-06,
+      "loss": 0.0003,
+      "step": 18565
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.548086618331973e-06,
+      "loss": 0.0003,
+      "step": 18566
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5456858885101075e-06,
+      "loss": 0.0003,
+      "step": 18567
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.543285158688241e-06,
+      "loss": 0.0003,
+      "step": 18568
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.540884428866375e-06,
+      "loss": 0.0003,
+      "step": 18569
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.53848369904451e-06,
+      "loss": 0.0003,
+      "step": 18570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.536082969222644e-06,
+      "loss": 0.0003,
+      "step": 18571
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.533682239400778e-06,
+      "loss": 0.0003,
+      "step": 18572
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.531281509578913e-06,
+      "loss": 0.0003,
+      "step": 18573
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.528880779757046e-06,
+      "loss": 0.0003,
+      "step": 18574
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.526480049935181e-06,
+      "loss": 0.0003,
+      "step": 18575
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.524079320113314e-06,
+      "loss": 0.0003,
+      "step": 18576
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.521678590291448e-06,
+      "loss": 0.0003,
+      "step": 18577
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.519277860469583e-06,
+      "loss": 0.0004,
+      "step": 18578
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.516877130647717e-06,
+      "loss": 0.0003,
+      "step": 18579
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5144764008258514e-06,
+      "loss": 0.0003,
+      "step": 18580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.512075671003986e-06,
+      "loss": 0.0003,
+      "step": 18581
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.50967494118212e-06,
+      "loss": 0.0003,
+      "step": 18582
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5072742113602545e-06,
+      "loss": 0.0003,
+      "step": 18583
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.504873481538388e-06,
+      "loss": 0.0003,
+      "step": 18584
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5024727517165215e-06,
+      "loss": 0.0003,
+      "step": 18585
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.500072021894656e-06,
+      "loss": 0.0003,
+      "step": 18586
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.49767129207279e-06,
+      "loss": 0.0003,
+      "step": 18587
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4952705622509245e-06,
+      "loss": 0.0003,
+      "step": 18588
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.492869832429059e-06,
+      "loss": 0.0002,
+      "step": 18589
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.490469102607193e-06,
+      "loss": 0.0003,
+      "step": 18590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4880683727853275e-06,
+      "loss": 0.0003,
+      "step": 18591
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.485667642963461e-06,
+      "loss": 0.0003,
+      "step": 18592
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.483266913141595e-06,
+      "loss": 0.0004,
+      "step": 18593
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.48086618331973e-06,
+      "loss": 0.0003,
+      "step": 18594
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.478465453497863e-06,
+      "loss": 0.0003,
+      "step": 18595
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4760647236759975e-06,
+      "loss": 0.0003,
+      "step": 18596
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.473663993854132e-06,
+      "loss": 0.0003,
+      "step": 18597
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.471263264032266e-06,
+      "loss": 0.0003,
+      "step": 18598
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4688625342104005e-06,
+      "loss": 0.0003,
+      "step": 18599
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.466461804388534e-06,
+      "loss": 0.0003,
+      "step": 18600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.464061074566668e-06,
+      "loss": 0.0003,
+      "step": 18601
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.461660344744803e-06,
+      "loss": 0.0003,
+      "step": 18602
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.459259614922937e-06,
+      "loss": 0.0003,
+      "step": 18603
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.456858885101071e-06,
+      "loss": 0.0003,
+      "step": 18604
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.454458155279205e-06,
+      "loss": 0.0003,
+      "step": 18605
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.452057425457339e-06,
+      "loss": 0.0003,
+      "step": 18606
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.449656695635474e-06,
+      "loss": 0.0003,
+      "step": 18607
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.447255965813607e-06,
+      "loss": 0.0003,
+      "step": 18608
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.444855235991741e-06,
+      "loss": 0.0003,
+      "step": 18609
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.442454506169876e-06,
+      "loss": 0.0003,
+      "step": 18610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.44005377634801e-06,
+      "loss": 0.0003,
+      "step": 18611
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4376530465261444e-06,
+      "loss": 0.0003,
+      "step": 18612
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.435252316704279e-06,
+      "loss": 0.4083,
+      "step": 18613
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.432851586882413e-06,
+      "loss": 0.0003,
+      "step": 18614
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.430450857060547e-06,
+      "loss": 0.0003,
+      "step": 18615
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.42805012723868e-06,
+      "loss": 0.0003,
+      "step": 18616
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4256493974168145e-06,
+      "loss": 0.0003,
+      "step": 18617
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.423248667594949e-06,
+      "loss": 0.0003,
+      "step": 18618
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.420847937773083e-06,
+      "loss": 0.0003,
+      "step": 18619
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4184472079512175e-06,
+      "loss": 0.0003,
+      "step": 18620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.416046478129352e-06,
+      "loss": 0.0002,
+      "step": 18621
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.413645748307486e-06,
+      "loss": 0.0003,
+      "step": 18622
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4112450184856205e-06,
+      "loss": 0.0003,
+      "step": 18623
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.408844288663754e-06,
+      "loss": 0.0003,
+      "step": 18624
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.406443558841888e-06,
+      "loss": 0.0003,
+      "step": 18625
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.404042829020022e-06,
+      "loss": 0.0003,
+      "step": 18626
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.401642099198156e-06,
+      "loss": 0.0003,
+      "step": 18627
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3992413693762905e-06,
+      "loss": 0.0003,
+      "step": 18628
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.396840639554425e-06,
+      "loss": 0.0003,
+      "step": 18629
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.394439909732559e-06,
+      "loss": 0.0003,
+      "step": 18630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3920391799106935e-06,
+      "loss": 0.0003,
+      "step": 18631
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.389638450088827e-06,
+      "loss": 0.0002,
+      "step": 18632
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.387237720266961e-06,
+      "loss": 0.0003,
+      "step": 18633
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.384836990445096e-06,
+      "loss": 0.0003,
+      "step": 18634
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.38243626062323e-06,
+      "loss": 0.0002,
+      "step": 18635
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3800355308013636e-06,
+      "loss": 0.0003,
+      "step": 18636
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.377634800979498e-06,
+      "loss": 0.0003,
+      "step": 18637
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.375234071157632e-06,
+      "loss": 0.0003,
+      "step": 18638
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.372833341335767e-06,
+      "loss": 0.0003,
+      "step": 18639
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3704326115139e-06,
+      "loss": 0.0003,
+      "step": 18640
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.368031881692034e-06,
+      "loss": 0.0003,
+      "step": 18641
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.365631151870169e-06,
+      "loss": 0.0003,
+      "step": 18642
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.363230422048303e-06,
+      "loss": 0.0003,
+      "step": 18643
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3608296922264374e-06,
+      "loss": 0.0003,
+      "step": 18644
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.358428962404572e-06,
+      "loss": 0.0003,
+      "step": 18645
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.356028232582705e-06,
+      "loss": 0.0002,
+      "step": 18646
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.35362750276084e-06,
+      "loss": 0.0003,
+      "step": 18647
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.351226772938973e-06,
+      "loss": 0.0002,
+      "step": 18648
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3488260431171075e-06,
+      "loss": 0.0003,
+      "step": 18649
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.346425313295242e-06,
+      "loss": 0.0003,
+      "step": 18650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.344024583473376e-06,
+      "loss": 0.0003,
+      "step": 18651
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3416238536515105e-06,
+      "loss": 0.0003,
+      "step": 18652
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.339223123829645e-06,
+      "loss": 0.0002,
+      "step": 18653
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.336822394007779e-06,
+      "loss": 0.0002,
+      "step": 18654
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.334421664185913e-06,
+      "loss": 0.0003,
+      "step": 18655
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.332020934364047e-06,
+      "loss": 0.0003,
+      "step": 18656
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3296202045421805e-06,
+      "loss": 0.0002,
+      "step": 18657
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.327219474720315e-06,
+      "loss": 0.0003,
+      "step": 18658
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.324818744898449e-06,
+      "loss": 0.0002,
+      "step": 18659
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3224180150765835e-06,
+      "loss": 0.0003,
+      "step": 18660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.320017285254718e-06,
+      "loss": 0.0003,
+      "step": 18661
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.317616555432852e-06,
+      "loss": 0.0003,
+      "step": 18662
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3152158256109865e-06,
+      "loss": 0.0003,
+      "step": 18663
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.31281509578912e-06,
+      "loss": 0.0003,
+      "step": 18664
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.310414365967254e-06,
+      "loss": 0.0003,
+      "step": 18665
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.308013636145389e-06,
+      "loss": 0.0002,
+      "step": 18666
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.305612906323522e-06,
+      "loss": 0.0003,
+      "step": 18667
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3032121765016566e-06,
+      "loss": 0.0002,
+      "step": 18668
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.300811446679791e-06,
+      "loss": 0.0003,
+      "step": 18669
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.298410716857925e-06,
+      "loss": 0.0003,
+      "step": 18670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2960099870360596e-06,
+      "loss": 0.0003,
+      "step": 18671
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.293609257214193e-06,
+      "loss": 0.0003,
+      "step": 18672
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.291208527392327e-06,
+      "loss": 0.0003,
+      "step": 18673
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.288807797570462e-06,
+      "loss": 0.0003,
+      "step": 18674
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.286407067748596e-06,
+      "loss": 0.0003,
+      "step": 18675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2840063379267304e-06,
+      "loss": 0.0003,
+      "step": 18676
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.281605608104864e-06,
+      "loss": 0.0002,
+      "step": 18677
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.279204878282998e-06,
+      "loss": 0.0002,
+      "step": 18678
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.276804148461133e-06,
+      "loss": 0.0003,
+      "step": 18679
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.274403418639266e-06,
+      "loss": 0.0003,
+      "step": 18680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2720026888174004e-06,
+      "loss": 0.0003,
+      "step": 18681
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.269601958995535e-06,
+      "loss": 0.0003,
+      "step": 18682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.267201229173669e-06,
+      "loss": 0.0002,
+      "step": 18683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.2648004993518035e-06,
+      "loss": 0.0031,
+      "step": 18684
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.262399769529938e-06,
+      "loss": 1.0399,
+      "step": 18685
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.259999039708071e-06,
+      "loss": 0.0003,
+      "step": 18686
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.257598309886206e-06,
+      "loss": 0.0002,
+      "step": 18687
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.255197580064339e-06,
+      "loss": 0.0003,
+      "step": 18688
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2527968502424735e-06,
+      "loss": 0.0003,
+      "step": 18689
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.250396120420608e-06,
+      "loss": 0.0004,
+      "step": 18690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.247995390598742e-06,
+      "loss": 0.0003,
+      "step": 18691
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2455946607768765e-06,
+      "loss": 0.0002,
+      "step": 18692
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.243193930955011e-06,
+      "loss": 0.0003,
+      "step": 18693
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.240793201133145e-06,
+      "loss": 0.0003,
+      "step": 18694
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2383924713112795e-06,
+      "loss": 0.0003,
+      "step": 18695
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.235991741489413e-06,
+      "loss": 0.0005,
+      "step": 18696
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.233591011667547e-06,
+      "loss": 0.0003,
+      "step": 18697
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.231190281845681e-06,
+      "loss": 0.0003,
+      "step": 18698
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.228789552023815e-06,
+      "loss": 0.0003,
+      "step": 18699
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2263888222019495e-06,
+      "loss": 0.0003,
+      "step": 18700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.223988092380084e-06,
+      "loss": 0.0003,
+      "step": 18701
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.221587362558218e-06,
+      "loss": 0.0003,
+      "step": 18702
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2191866327363526e-06,
+      "loss": 0.0003,
+      "step": 18703
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.216785902914486e-06,
+      "loss": 0.0003,
+      "step": 18704
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.21438517309262e-06,
+      "loss": 0.0003,
+      "step": 18705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.211984443270755e-06,
+      "loss": 0.0016,
+      "step": 18706
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.209583713448889e-06,
+      "loss": 0.0002,
+      "step": 18707
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.207182983627023e-06,
+      "loss": 0.0002,
+      "step": 18708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.204782253805157e-06,
+      "loss": 0.0003,
+      "step": 18709
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.202381523983291e-06,
+      "loss": 0.0002,
+      "step": 18710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.199980794161426e-06,
+      "loss": 0.0003,
+      "step": 18711
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.197580064339559e-06,
+      "loss": 0.0003,
+      "step": 18712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1951793345176934e-06,
+      "loss": 0.0003,
+      "step": 18713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.192778604695828e-06,
+      "loss": 0.0003,
+      "step": 18714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.190377874873962e-06,
+      "loss": 0.0003,
+      "step": 18715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1879771450520965e-06,
+      "loss": 0.0003,
+      "step": 18716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.18557641523023e-06,
+      "loss": 0.0003,
+      "step": 18717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.183175685408364e-06,
+      "loss": 0.0003,
+      "step": 18718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.180774955586499e-06,
+      "loss": 0.0003,
+      "step": 18719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.178374225764632e-06,
+      "loss": 0.0003,
+      "step": 18720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1759734959427665e-06,
+      "loss": 0.0003,
+      "step": 18721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.173572766120901e-06,
+      "loss": 0.0003,
+      "step": 18722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.171172036299035e-06,
+      "loss": 0.0003,
+      "step": 18723
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1687713064771695e-06,
+      "loss": 0.0003,
+      "step": 18724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.166370576655304e-06,
+      "loss": 0.0002,
+      "step": 18725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.163969846833438e-06,
+      "loss": 0.0003,
+      "step": 18726
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.161569117011572e-06,
+      "loss": 0.0003,
+      "step": 18727
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.159168387189706e-06,
+      "loss": 0.0003,
+      "step": 18728
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1567676573678395e-06,
+      "loss": 0.0003,
+      "step": 18729
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.154366927545974e-06,
+      "loss": 0.0003,
+      "step": 18730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.151966197724108e-06,
+      "loss": 0.0003,
+      "step": 18731
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1495654679022425e-06,
+      "loss": 0.0003,
+      "step": 18732
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.147164738080377e-06,
+      "loss": 0.0003,
+      "step": 18733
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.144764008258511e-06,
+      "loss": 0.0004,
+      "step": 18734
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1423632784366456e-06,
+      "loss": 0.0003,
+      "step": 18735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.139962548614779e-06,
+      "loss": 0.0003,
+      "step": 18736
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.137561818792913e-06,
+      "loss": 0.0003,
+      "step": 18737
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.135161088971048e-06,
+      "loss": 0.0003,
+      "step": 18738
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.132760359149181e-06,
+      "loss": 0.0003,
+      "step": 18739
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.130359629327316e-06,
+      "loss": 0.0003,
+      "step": 18740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.12795889950545e-06,
+      "loss": 0.0003,
+      "step": 18741
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.125558169683584e-06,
+      "loss": 0.0003,
+      "step": 18742
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.123157439861719e-06,
+      "loss": 0.0002,
+      "step": 18743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.120756710039852e-06,
+      "loss": 0.0003,
+      "step": 18744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1183559802179864e-06,
+      "loss": 0.0004,
+      "step": 18745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.115955250396121e-06,
+      "loss": 0.0003,
+      "step": 18746
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.113554520574255e-06,
+      "loss": 0.0003,
+      "step": 18747
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.111153790752389e-06,
+      "loss": 0.0003,
+      "step": 18748
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.108753060930523e-06,
+      "loss": 0.0002,
+      "step": 18749
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.106352331108657e-06,
+      "loss": 0.0003,
+      "step": 18750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.103951601286792e-06,
+      "loss": 0.0023,
+      "step": 18751
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.101550871464925e-06,
+      "loss": 0.0003,
+      "step": 18752
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.0991501416430595e-06,
+      "loss": 0.0004,
+      "step": 18753
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.096749411821194e-06,
+      "loss": 0.0003,
+      "step": 18754
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.094348681999328e-06,
+      "loss": 0.0003,
+      "step": 18755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0919479521774625e-06,
+      "loss": 0.0003,
+      "step": 18756
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.089547222355597e-06,
+      "loss": 0.0003,
+      "step": 18757
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08714649253373e-06,
+      "loss": 0.0003,
+      "step": 18758
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 0.0002,
+      "step": 18759
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.082345032889998e-06,
+      "loss": 0.0002,
+      "step": 18760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0799443030681325e-06,
+      "loss": 0.0002,
+      "step": 18761
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.077543573246267e-06,
+      "loss": 0.0002,
+      "step": 18762
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.075142843424401e-06,
+      "loss": 0.0002,
+      "step": 18763
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0727421136025355e-06,
+      "loss": 0.0003,
+      "step": 18764
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.07034138378067e-06,
+      "loss": 0.0002,
+      "step": 18765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.067940653958804e-06,
+      "loss": 0.0003,
+      "step": 18766
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0655399241369386e-06,
+      "loss": 0.0003,
+      "step": 18767
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.063139194315072e-06,
+      "loss": 0.0002,
+      "step": 18768
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.060738464493206e-06,
+      "loss": 0.0003,
+      "step": 18769
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.05833773467134e-06,
+      "loss": 0.0003,
+      "step": 18770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.055937004849474e-06,
+      "loss": 0.0003,
+      "step": 18771
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0535362750276086e-06,
+      "loss": 0.0002,
+      "step": 18772
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.051135545205743e-06,
+      "loss": 0.9522,
+      "step": 18773
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048734815383877e-06,
+      "loss": 0.0003,
+      "step": 18774
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.046334085562012e-06,
+      "loss": 0.0003,
+      "step": 18775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.043933355740145e-06,
+      "loss": 0.0003,
+      "step": 18776
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0415326259182794e-06,
+      "loss": 0.0002,
+      "step": 18777
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.039131896096414e-06,
+      "loss": 0.0003,
+      "step": 18778
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.036731166274547e-06,
+      "loss": 0.0003,
+      "step": 18779
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.034330436452682e-06,
+      "loss": 0.0003,
+      "step": 18780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.031929706630816e-06,
+      "loss": 0.0004,
+      "step": 18781
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.02952897680895e-06,
+      "loss": 0.0002,
+      "step": 18782
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.027128246987085e-06,
+      "loss": 0.0005,
+      "step": 18783
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.024727517165218e-06,
+      "loss": 0.0003,
+      "step": 18784
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0223267873433525e-06,
+      "loss": 0.0002,
+      "step": 18785
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.019926057521487e-06,
+      "loss": 0.0003,
+      "step": 18786
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.017525327699621e-06,
+      "loss": 0.0003,
+      "step": 18787
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0151245978777555e-06,
+      "loss": 0.0003,
+      "step": 18788
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.012723868055889e-06,
+      "loss": 0.0003,
+      "step": 18789
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.010323138234023e-06,
+      "loss": 0.0004,
+      "step": 18790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.007922408412158e-06,
+      "loss": 0.0003,
+      "step": 18791
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.005521678590291e-06,
+      "loss": 0.0003,
+      "step": 18792
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0031209487684255e-06,
+      "loss": 0.0002,
+      "step": 18793
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.00072021894656e-06,
+      "loss": 0.0002,
+      "step": 18794
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.998319489124694e-06,
+      "loss": 0.0003,
+      "step": 18795
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9959187593028285e-06,
+      "loss": 0.0003,
+      "step": 18796
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.993518029480963e-06,
+      "loss": 0.0002,
+      "step": 18797
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.991117299659097e-06,
+      "loss": 0.0003,
+      "step": 18798
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.988716569837231e-06,
+      "loss": 0.0005,
+      "step": 18799
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.986315840015365e-06,
+      "loss": 0.0003,
+      "step": 18800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9839151101934985e-06,
+      "loss": 0.0003,
+      "step": 18801
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.981514380371633e-06,
+      "loss": 0.0003,
+      "step": 18802
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.979113650549767e-06,
+      "loss": 0.0002,
+      "step": 18803
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9767129207279016e-06,
+      "loss": 0.0002,
+      "step": 18804
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.974312190906036e-06,
+      "loss": 0.0004,
+      "step": 18805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.97191146108417e-06,
+      "loss": 1.0484,
+      "step": 18806
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.969510731262305e-06,
+      "loss": 0.0003,
+      "step": 18807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.967110001440438e-06,
+      "loss": 0.0003,
+      "step": 18808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9647092716185724e-06,
+      "loss": 0.0004,
+      "step": 18809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.962308541796706e-06,
+      "loss": 0.2014,
+      "step": 18810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.95990781197484e-06,
+      "loss": 0.5119,
+      "step": 18811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.957507082152975e-06,
+      "loss": 0.0002,
+      "step": 18812
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.955106352331109e-06,
+      "loss": 0.0002,
+      "step": 18813
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.952705622509243e-06,
+      "loss": 0.0003,
+      "step": 18814
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.950304892687378e-06,
+      "loss": 0.0003,
+      "step": 18815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.947904162865511e-06,
+      "loss": 0.0003,
+      "step": 18816
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9455034330436455e-06,
+      "loss": 0.0003,
+      "step": 18817
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.94310270322178e-06,
+      "loss": 0.0003,
+      "step": 18818
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.940701973399914e-06,
+      "loss": 0.0004,
+      "step": 18819
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.938301243578048e-06,
+      "loss": 0.0004,
+      "step": 18820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.935900513756182e-06,
+      "loss": 0.0026,
+      "step": 18821
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.933499783934316e-06,
+      "loss": 0.0003,
+      "step": 18822
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.931099054112451e-06,
+      "loss": 0.0004,
+      "step": 18823
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.928698324290584e-06,
+      "loss": 0.0003,
+      "step": 18824
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9262975944687185e-06,
+      "loss": 0.0003,
+      "step": 18825
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.923896864646853e-06,
+      "loss": 0.0003,
+      "step": 18826
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921496134824987e-06,
+      "loss": 0.0003,
+      "step": 18827
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9190954050031215e-06,
+      "loss": 0.0003,
+      "step": 18828
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916694675181256e-06,
+      "loss": 0.0019,
+      "step": 18829
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914293945359389e-06,
+      "loss": 0.0003,
+      "step": 18830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.911893215537524e-06,
+      "loss": 0.0002,
+      "step": 18831
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.909492485715657e-06,
+      "loss": 0.0002,
+      "step": 18832
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9070917558937915e-06,
+      "loss": 0.0003,
+      "step": 18833
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.904691026071926e-06,
+      "loss": 0.0002,
+      "step": 18834
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.90229029625006e-06,
+      "loss": 0.0003,
+      "step": 18835
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8998895664281946e-06,
+      "loss": 0.0002,
+      "step": 18836
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.897488836606329e-06,
+      "loss": 0.0002,
+      "step": 18837
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.895088106784463e-06,
+      "loss": 0.0002,
+      "step": 18838
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.892687376962597e-06,
+      "loss": 0.0003,
+      "step": 18839
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.890286647140731e-06,
+      "loss": 0.0002,
+      "step": 18840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.887885917318865e-06,
+      "loss": 0.0003,
+      "step": 18841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.885485187496999e-06,
+      "loss": 0.0002,
+      "step": 18842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.883084457675133e-06,
+      "loss": 0.0002,
+      "step": 18843
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.880683727853268e-06,
+      "loss": 0.0003,
+      "step": 18844
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.878282998031402e-06,
+      "loss": 0.0002,
+      "step": 18845
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.875882268209536e-06,
+      "loss": 0.0002,
+      "step": 18846
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.87348153838767e-06,
+      "loss": 0.0002,
+      "step": 18847
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.871080808565804e-06,
+      "loss": 0.0003,
+      "step": 18848
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8686800787439385e-06,
+      "loss": 0.0002,
+      "step": 18849
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.866279348922073e-06,
+      "loss": 0.0005,
+      "step": 18850
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.863878619100206e-06,
+      "loss": 0.0003,
+      "step": 18851
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.861477889278341e-06,
+      "loss": 0.0003,
+      "step": 18852
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.859077159456475e-06,
+      "loss": 0.0003,
+      "step": 18853
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.856676429634609e-06,
+      "loss": 0.0002,
+      "step": 18854
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.854275699812743e-06,
+      "loss": 0.0004,
+      "step": 18855
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.851874969990877e-06,
+      "loss": 0.0002,
+      "step": 18856
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8494742401690115e-06,
+      "loss": 0.0002,
+      "step": 18857
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.847073510347146e-06,
+      "loss": 0.0003,
+      "step": 18858
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.84467278052528e-06,
+      "loss": 0.0002,
+      "step": 18859
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8422720507034145e-06,
+      "loss": 0.0012,
+      "step": 18860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.839871320881548e-06,
+      "loss": 0.0003,
+      "step": 18861
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.837470591059682e-06,
+      "loss": 0.0003,
+      "step": 18862
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.835069861237816e-06,
+      "loss": 0.0002,
+      "step": 18863
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.83266913141595e-06,
+      "loss": 0.0003,
+      "step": 18864
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8302684015940845e-06,
+      "loss": 0.0019,
+      "step": 18865
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.827867671772219e-06,
+      "loss": 0.0002,
+      "step": 18866
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.825466941950353e-06,
+      "loss": 0.0004,
+      "step": 18867
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8230662121284876e-06,
+      "loss": 0.0003,
+      "step": 18868
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.820665482306622e-06,
+      "loss": 1.0708,
+      "step": 18869
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.818264752484756e-06,
+      "loss": 0.0003,
+      "step": 18870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.81586402266289e-06,
+      "loss": 0.0003,
+      "step": 18871
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.813463292841023e-06,
+      "loss": 0.0002,
+      "step": 18872
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.811062563019158e-06,
+      "loss": 0.0003,
+      "step": 18873
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.808661833197292e-06,
+      "loss": 0.0003,
+      "step": 18874
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.806261103375426e-06,
+      "loss": 0.0003,
+      "step": 18875
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.803860373553561e-06,
+      "loss": 0.0003,
+      "step": 18876
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.801459643731695e-06,
+      "loss": 0.0003,
+      "step": 18877
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.799058913909829e-06,
+      "loss": 0.0003,
+      "step": 18878
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796658184087963e-06,
+      "loss": 0.0003,
+      "step": 18879
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.794257454266097e-06,
+      "loss": 0.0003,
+      "step": 18880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7918567244442315e-06,
+      "loss": 0.0003,
+      "step": 18881
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.789455994622365e-06,
+      "loss": 0.0003,
+      "step": 18882
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.787055264800499e-06,
+      "loss": 0.0003,
+      "step": 18883
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.784654534978634e-06,
+      "loss": 0.0003,
+      "step": 18884
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.782253805156768e-06,
+      "loss": 0.0003,
+      "step": 18885
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.779853075334902e-06,
+      "loss": 0.0003,
+      "step": 18886
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.777452345513036e-06,
+      "loss": 0.0003,
+      "step": 18887
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.77505161569117e-06,
+      "loss": 0.0003,
+      "step": 18888
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7726508858693045e-06,
+      "loss": 0.0003,
+      "step": 18889
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.770250156047439e-06,
+      "loss": 0.0003,
+      "step": 18890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.767849426225573e-06,
+      "loss": 0.0003,
+      "step": 18891
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.765448696403707e-06,
+      "loss": 0.0003,
+      "step": 18892
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.763047966581841e-06,
+      "loss": 0.0002,
+      "step": 18893
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.760647236759975e-06,
+      "loss": 0.0003,
+      "step": 18894
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.758246506938109e-06,
+      "loss": 0.0003,
+      "step": 18895
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.755845777116243e-06,
+      "loss": 0.0003,
+      "step": 18896
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7534450472943775e-06,
+      "loss": 0.0003,
+      "step": 18897
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.751044317472512e-06,
+      "loss": 0.0002,
+      "step": 18898
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.748643587650646e-06,
+      "loss": 0.0003,
+      "step": 18899
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7462428578287806e-06,
+      "loss": 0.0003,
+      "step": 18900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.743842128006915e-06,
+      "loss": 0.0003,
+      "step": 18901
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.741441398185048e-06,
+      "loss": 0.0003,
+      "step": 18902
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.739040668363182e-06,
+      "loss": 0.0003,
+      "step": 18903
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.736639938541316e-06,
+      "loss": 0.0003,
+      "step": 18904
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.7342392087194506e-06,
+      "loss": 0.0002,
+      "step": 18905
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.731838478897585e-06,
+      "loss": 0.0002,
+      "step": 18906
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.729437749075719e-06,
+      "loss": 0.0003,
+      "step": 18907
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.727037019253854e-06,
+      "loss": 0.0002,
+      "step": 18908
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.724636289431988e-06,
+      "loss": 0.0003,
+      "step": 18909
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.722235559610122e-06,
+      "loss": 0.002,
+      "step": 18910
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.719834829788256e-06,
+      "loss": 0.0003,
+      "step": 18911
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.71743409996639e-06,
+      "loss": 0.0003,
+      "step": 18912
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.715033370144524e-06,
+      "loss": 0.0035,
+      "step": 18913
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.712632640322658e-06,
+      "loss": 0.0003,
+      "step": 18914
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.710231910500792e-06,
+      "loss": 0.0003,
+      "step": 18915
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.707831180678927e-06,
+      "loss": 0.0003,
+      "step": 18916
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.705430450857061e-06,
+      "loss": 0.0003,
+      "step": 18917
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.703029721035195e-06,
+      "loss": 0.0003,
+      "step": 18918
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.700628991213329e-06,
+      "loss": 0.0002,
+      "step": 18919
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.698228261391463e-06,
+      "loss": 0.0002,
+      "step": 18920
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6958275315695975e-06,
+      "loss": 0.0002,
+      "step": 18921
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.693426801747732e-06,
+      "loss": 0.0004,
+      "step": 18922
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.691026071925865e-06,
+      "loss": 0.0003,
+      "step": 18923
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.688625342104e-06,
+      "loss": 0.0003,
+      "step": 18924
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.686224612282134e-06,
+      "loss": 0.0002,
+      "step": 18925
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.683823882460268e-06,
+      "loss": 0.0003,
+      "step": 18926
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.681423152638402e-06,
+      "loss": 0.0002,
+      "step": 18927
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.679022422816536e-06,
+      "loss": 0.0003,
+      "step": 18928
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6766216929946705e-06,
+      "loss": 0.0002,
+      "step": 18929
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.674220963172805e-06,
+      "loss": 0.0002,
+      "step": 18930
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.671820233350939e-06,
+      "loss": 0.0002,
+      "step": 18931
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6694195035290736e-06,
+      "loss": 0.0003,
+      "step": 18932
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.667018773707207e-06,
+      "loss": 0.0002,
+      "step": 18933
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.664618043885341e-06,
+      "loss": 0.0003,
+      "step": 18934
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.662217314063475e-06,
+      "loss": 0.0002,
+      "step": 18935
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.659816584241609e-06,
+      "loss": 0.0003,
+      "step": 18936
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6574158544197436e-06,
+      "loss": 0.0003,
+      "step": 18937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.655015124597878e-06,
+      "loss": 0.0002,
+      "step": 18938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.652614394776012e-06,
+      "loss": 0.0002,
+      "step": 18939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.650213664954147e-06,
+      "loss": 0.0003,
+      "step": 18940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.647812935132281e-06,
+      "loss": 0.0002,
+      "step": 18941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.645412205310415e-06,
+      "loss": 0.0003,
+      "step": 18942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.643011475488549e-06,
+      "loss": 0.0002,
+      "step": 18943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.640610745666682e-06,
+      "loss": 0.0003,
+      "step": 18944
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.638210015844817e-06,
+      "loss": 0.0003,
+      "step": 18945
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.635809286022951e-06,
+      "loss": 0.0003,
+      "step": 18946
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.633408556201085e-06,
+      "loss": 0.0002,
+      "step": 18947
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.63100782637922e-06,
+      "loss": 0.0002,
+      "step": 18948
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.628607096557354e-06,
+      "loss": 0.0002,
+      "step": 18949
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.626206366735488e-06,
+      "loss": 0.0002,
+      "step": 18950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.623805636913622e-06,
+      "loss": 0.0002,
+      "step": 18951
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.621404907091756e-06,
+      "loss": 0.0002,
+      "step": 18952
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6190041772698905e-06,
+      "loss": 0.0002,
+      "step": 18953
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.616603447448024e-06,
+      "loss": 0.0003,
+      "step": 18954
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.614202717626158e-06,
+      "loss": 0.0003,
+      "step": 18955
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.611801987804293e-06,
+      "loss": 0.0003,
+      "step": 18956
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.609401257982427e-06,
+      "loss": 0.0003,
+      "step": 18957
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.607000528160561e-06,
+      "loss": 0.0002,
+      "step": 18958
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.604599798338695e-06,
+      "loss": 0.0003,
+      "step": 18959
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.602199068516829e-06,
+      "loss": 0.0002,
+      "step": 18960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5997983386949635e-06,
+      "loss": 0.0002,
+      "step": 18961
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.597397608873098e-06,
+      "loss": 0.0002,
+      "step": 18962
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.594996879051232e-06,
+      "loss": 0.0002,
+      "step": 18963
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.592596149229366e-06,
+      "loss": 0.0002,
+      "step": 18964
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5901954194075e-06,
+      "loss": 0.0003,
+      "step": 18965
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.587794689585634e-06,
+      "loss": 0.0002,
+      "step": 18966
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.585393959763768e-06,
+      "loss": 0.0003,
+      "step": 18967
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.582993229941902e-06,
+      "loss": 0.0002,
+      "step": 18968
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5805925001200366e-06,
+      "loss": 0.0003,
+      "step": 18969
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.578191770298171e-06,
+      "loss": 0.0003,
+      "step": 18970
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.575791040476305e-06,
+      "loss": 0.0002,
+      "step": 18971
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.57339031065444e-06,
+      "loss": 0.0003,
+      "step": 18972
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.570989580832574e-06,
+      "loss": 0.0002,
+      "step": 18973
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5685888510107074e-06,
+      "loss": 0.0002,
+      "step": 18974
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.566188121188841e-06,
+      "loss": 0.0002,
+      "step": 18975
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.563787391366975e-06,
+      "loss": 0.0003,
+      "step": 18976
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.56138666154511e-06,
+      "loss": 0.0021,
+      "step": 18977
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.558985931723244e-06,
+      "loss": 0.0003,
+      "step": 18978
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.556585201901378e-06,
+      "loss": 0.0002,
+      "step": 18979
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.554184472079513e-06,
+      "loss": 0.0002,
+      "step": 18980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.551783742257647e-06,
+      "loss": 0.0002,
+      "step": 18981
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.549383012435781e-06,
+      "loss": 0.0002,
+      "step": 18982
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.546982282613915e-06,
+      "loss": 0.0002,
+      "step": 18983
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.544581552792049e-06,
+      "loss": 0.5196,
+      "step": 18984
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.542180822970183e-06,
+      "loss": 0.0003,
+      "step": 18985
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.539780093148317e-06,
+      "loss": 0.0002,
+      "step": 18986
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.537379363326451e-06,
+      "loss": 0.0002,
+      "step": 18987
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.534978633504586e-06,
+      "loss": 0.0002,
+      "step": 18988
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.53257790368272e-06,
+      "loss": 0.0003,
+      "step": 18989
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.530177173860854e-06,
+      "loss": 0.0003,
+      "step": 18990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.527776444038988e-06,
+      "loss": 0.0002,
+      "step": 18991
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.525375714217122e-06,
+      "loss": 0.0002,
+      "step": 18992
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5229749843952565e-06,
+      "loss": 0.0002,
+      "step": 18993
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.520574254573391e-06,
+      "loss": 0.0003,
+      "step": 18994
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.518173524751524e-06,
+      "loss": 0.0002,
+      "step": 18995
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.515772794929659e-06,
+      "loss": 0.0002,
+      "step": 18996
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.513372065107793e-06,
+      "loss": 0.0002,
+      "step": 18997
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.510971335285927e-06,
+      "loss": 0.0002,
+      "step": 18998
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.508570605464061e-06,
+      "loss": 0.0003,
+      "step": 18999
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.506169875642195e-06,
+      "loss": 0.0003,
+      "step": 19000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5037691458203296e-06,
+      "loss": 0.0003,
+      "step": 19001
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.501368415998464e-06,
+      "loss": 0.0002,
+      "step": 19002
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.498967686176598e-06,
+      "loss": 0.0002,
+      "step": 19003
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.496566956354733e-06,
+      "loss": 0.0003,
+      "step": 19004
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.494166226532866e-06,
+      "loss": 0.0002,
+      "step": 19005
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.491765496711e-06,
+      "loss": 0.0002,
+      "step": 19006
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.489364766889134e-06,
+      "loss": 0.0002,
+      "step": 19007
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.486964037067268e-06,
+      "loss": 0.0002,
+      "step": 19008
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.484563307245403e-06,
+      "loss": 0.0002,
+      "step": 19009
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.482162577423537e-06,
+      "loss": 0.0002,
+      "step": 19010
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.479761847601671e-06,
+      "loss": 0.0003,
+      "step": 19011
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.477361117779806e-06,
+      "loss": 0.0002,
+      "step": 19012
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.47496038795794e-06,
+      "loss": 0.0002,
+      "step": 19013
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.472559658136074e-06,
+      "loss": 0.0002,
+      "step": 19014
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.470158928314208e-06,
+      "loss": 0.0003,
+      "step": 19015
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.467758198492341e-06,
+      "loss": 0.0002,
+      "step": 19016
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.465357468670476e-06,
+      "loss": 0.0002,
+      "step": 19017
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.46295673884861e-06,
+      "loss": 0.0002,
+      "step": 19018
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.460556009026744e-06,
+      "loss": 0.0002,
+      "step": 19019
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.458155279204879e-06,
+      "loss": 0.0002,
+      "step": 19020
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.455754549383013e-06,
+      "loss": 0.0002,
+      "step": 19021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.453353819561147e-06,
+      "loss": 0.0002,
+      "step": 19022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.450953089739281e-06,
+      "loss": 0.0002,
+      "step": 19023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.448552359917415e-06,
+      "loss": 0.0002,
+      "step": 19024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4461516300955495e-06,
+      "loss": 0.0002,
+      "step": 19025
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.443750900273683e-06,
+      "loss": 0.0003,
+      "step": 19026
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.441350170451817e-06,
+      "loss": 0.0002,
+      "step": 19027
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.438949440629952e-06,
+      "loss": 0.0002,
+      "step": 19028
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.436548710808086e-06,
+      "loss": 0.0002,
+      "step": 19029
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.43414798098622e-06,
+      "loss": 0.0002,
+      "step": 19030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.431747251164354e-06,
+      "loss": 0.0002,
+      "step": 19031
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.429346521342488e-06,
+      "loss": 0.0002,
+      "step": 19032
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4269457915206226e-06,
+      "loss": 0.0003,
+      "step": 19033
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.424545061698757e-06,
+      "loss": 0.0003,
+      "step": 19034
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.422144331876891e-06,
+      "loss": 0.0003,
+      "step": 19035
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.419743602055025e-06,
+      "loss": 0.0002,
+      "step": 19036
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.417342872233159e-06,
+      "loss": 0.0002,
+      "step": 19037
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.414942142411293e-06,
+      "loss": 0.0002,
+      "step": 19038
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.412541412589427e-06,
+      "loss": 0.0003,
+      "step": 19039
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.410140682767561e-06,
+      "loss": 0.0002,
+      "step": 19040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.407739952945696e-06,
+      "loss": 0.0002,
+      "step": 19041
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.40533922312383e-06,
+      "loss": 0.0002,
+      "step": 19042
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.402938493301964e-06,
+      "loss": 0.0002,
+      "step": 19043
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.400537763480099e-06,
+      "loss": 0.0002,
+      "step": 19044
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.398137033658233e-06,
+      "loss": 0.0003,
+      "step": 19045
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3957363038363665e-06,
+      "loss": 0.0003,
+      "step": 19046
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3933355740145e-06,
+      "loss": 0.0002,
+      "step": 19047
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.390934844192634e-06,
+      "loss": 0.0002,
+      "step": 19048
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.388534114370769e-06,
+      "loss": 0.0002,
+      "step": 19049
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.386133384548903e-06,
+      "loss": 0.0002,
+      "step": 19050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.383732654727037e-06,
+      "loss": 0.0002,
+      "step": 19051
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.381331924905172e-06,
+      "loss": 0.0002,
+      "step": 19052
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.378931195083306e-06,
+      "loss": 0.9332,
+      "step": 19053
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.37653046526144e-06,
+      "loss": 1.017,
+      "step": 19054
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.374129735439574e-06,
+      "loss": 0.0002,
+      "step": 19055
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.371729005617708e-06,
+      "loss": 0.0003,
+      "step": 19056
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.369328275795842e-06,
+      "loss": 0.0003,
+      "step": 19057
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.366927545973976e-06,
+      "loss": 0.0002,
+      "step": 19058
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.36452681615211e-06,
+      "loss": 0.0002,
+      "step": 19059
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.362126086330245e-06,
+      "loss": 0.0002,
+      "step": 19060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.359725356508379e-06,
+      "loss": 0.0002,
+      "step": 19061
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.357324626686513e-06,
+      "loss": 0.0003,
+      "step": 19062
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.354923896864647e-06,
+      "loss": 0.0003,
+      "step": 19063
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.352523167042781e-06,
+      "loss": 0.0003,
+      "step": 19064
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3501224372209156e-06,
+      "loss": 0.0002,
+      "step": 19065
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.34772170739905e-06,
+      "loss": 0.0002,
+      "step": 19066
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.345320977577183e-06,
+      "loss": 0.0002,
+      "step": 19067
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.342920247755318e-06,
+      "loss": 0.0003,
+      "step": 19068
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.340519517933452e-06,
+      "loss": 0.0002,
+      "step": 19069
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.338118788111586e-06,
+      "loss": 0.0002,
+      "step": 19070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.33571805828972e-06,
+      "loss": 0.0002,
+      "step": 19071
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.333317328467854e-06,
+      "loss": 0.0003,
+      "step": 19072
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.330916598645989e-06,
+      "loss": 0.0003,
+      "step": 19073
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.328515868824123e-06,
+      "loss": 1.0331,
+      "step": 19074
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.326115139002257e-06,
+      "loss": 0.0003,
+      "step": 19075
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.323714409180392e-06,
+      "loss": 0.0003,
+      "step": 19076
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.321313679358525e-06,
+      "loss": 0.0008,
+      "step": 19077
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3189129495366595e-06,
+      "loss": 0.0007,
+      "step": 19078
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.316512219714793e-06,
+      "loss": 0.0006,
+      "step": 19079
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.314111489892927e-06,
+      "loss": 0.0002,
+      "step": 19080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.311710760071062e-06,
+      "loss": 0.0003,
+      "step": 19081
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.309310030249196e-06,
+      "loss": 0.0004,
+      "step": 19082
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.30690930042733e-06,
+      "loss": 0.0003,
+      "step": 19083
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.304508570605465e-06,
+      "loss": 0.0002,
+      "step": 19084
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.302107840783599e-06,
+      "loss": 0.0006,
+      "step": 19085
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2997071109617325e-06,
+      "loss": 0.0004,
+      "step": 19086
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.297306381139867e-06,
+      "loss": 1.0505,
+      "step": 19087
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.294905651318e-06,
+      "loss": 0.0003,
+      "step": 19088
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.292504921496135e-06,
+      "loss": 0.0002,
+      "step": 19089
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.290104191674269e-06,
+      "loss": 0.0003,
+      "step": 19090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.287703461852403e-06,
+      "loss": 0.0006,
+      "step": 19091
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.285302732030538e-06,
+      "loss": 0.0003,
+      "step": 19092
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.282902002208672e-06,
+      "loss": 0.0003,
+      "step": 19093
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.280501272386806e-06,
+      "loss": 0.0003,
+      "step": 19094
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.27810054256494e-06,
+      "loss": 0.0003,
+      "step": 19095
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.275699812743074e-06,
+      "loss": 0.0002,
+      "step": 19096
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2732990829212086e-06,
+      "loss": 0.0004,
+      "step": 19097
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.270898353099342e-06,
+      "loss": 0.0002,
+      "step": 19098
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.268497623277476e-06,
+      "loss": 0.0003,
+      "step": 19099
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.266096893455611e-06,
+      "loss": 0.0003,
+      "step": 19100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.263696163633745e-06,
+      "loss": 0.0003,
+      "step": 19101
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.261295433811879e-06,
+      "loss": 0.0004,
+      "step": 19102
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.258894703990013e-06,
+      "loss": 0.0003,
+      "step": 19103
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.256493974168147e-06,
+      "loss": 0.0002,
+      "step": 19104
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.254093244346282e-06,
+      "loss": 0.0002,
+      "step": 19105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.251692514524416e-06,
+      "loss": 0.0003,
+      "step": 19106
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.24929178470255e-06,
+      "loss": 0.0003,
+      "step": 19107
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.246891054880684e-06,
+      "loss": 0.0003,
+      "step": 19108
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.244490325058818e-06,
+      "loss": 0.0003,
+      "step": 19109
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2420895952369524e-06,
+      "loss": 0.0003,
+      "step": 19110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.239688865415086e-06,
+      "loss": 0.0004,
+      "step": 19111
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.23728813559322e-06,
+      "loss": 0.0004,
+      "step": 19112
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.234887405771355e-06,
+      "loss": 0.0002,
+      "step": 19113
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.232486675949489e-06,
+      "loss": 0.0004,
+      "step": 19114
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.230085946127623e-06,
+      "loss": 0.0003,
+      "step": 19115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.227685216305758e-06,
+      "loss": 0.0003,
+      "step": 19116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.225284486483891e-06,
+      "loss": 0.0003,
+      "step": 19117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2228837566620255e-06,
+      "loss": 0.0003,
+      "step": 19118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.220483026840159e-06,
+      "loss": 0.0002,
+      "step": 19119
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.218082297018293e-06,
+      "loss": 0.0003,
+      "step": 19120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.215681567196428e-06,
+      "loss": 0.0002,
+      "step": 19121
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.213280837374562e-06,
+      "loss": 0.0003,
+      "step": 19122
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.210880107552696e-06,
+      "loss": 0.0003,
+      "step": 19123
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.208479377730831e-06,
+      "loss": 0.4593,
+      "step": 19124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.206078647908965e-06,
+      "loss": 0.0002,
+      "step": 19125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.203677918087099e-06,
+      "loss": 0.0003,
+      "step": 19126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.201277188265233e-06,
+      "loss": 0.0002,
+      "step": 19127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.198876458443367e-06,
+      "loss": 0.0004,
+      "step": 19128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.196475728621501e-06,
+      "loss": 0.8246,
+      "step": 19129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.194074998799635e-06,
+      "loss": 0.0006,
+      "step": 19130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.191674268977769e-06,
+      "loss": 0.0004,
+      "step": 19131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.189273539155904e-06,
+      "loss": 0.0003,
+      "step": 19132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.186872809334038e-06,
+      "loss": 0.0003,
+      "step": 19133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.184472079512172e-06,
+      "loss": 0.0002,
+      "step": 19134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.182071349690306e-06,
+      "loss": 0.0003,
+      "step": 19135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.17967061986844e-06,
+      "loss": 0.0004,
+      "step": 19136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.177269890046575e-06,
+      "loss": 0.0003,
+      "step": 19137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.174869160224709e-06,
+      "loss": 0.0003,
+      "step": 19138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.172468430402842e-06,
+      "loss": 0.0009,
+      "step": 19139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.170067700580977e-06,
+      "loss": 0.0008,
+      "step": 19140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.167666970759111e-06,
+      "loss": 0.0003,
+      "step": 19141
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1652662409372454e-06,
+      "loss": 0.0003,
+      "step": 19142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.162865511115379e-06,
+      "loss": 0.0003,
+      "step": 19143
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.160464781293513e-06,
+      "loss": 0.0003,
+      "step": 19144
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.158064051471648e-06,
+      "loss": 0.0003,
+      "step": 19145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.155663321649782e-06,
+      "loss": 0.0003,
+      "step": 19146
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.153262591827916e-06,
+      "loss": 0.0013,
+      "step": 19147
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.15086186200605e-06,
+      "loss": 0.0003,
+      "step": 19148
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.148461132184184e-06,
+      "loss": 0.0003,
+      "step": 19149
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1460604023623185e-06,
+      "loss": 0.0006,
+      "step": 19150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.143659672540452e-06,
+      "loss": 0.0003,
+      "step": 19151
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.141258942718586e-06,
+      "loss": 0.0002,
+      "step": 19152
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.138858212896721e-06,
+      "loss": 0.0003,
+      "step": 19153
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.136457483074855e-06,
+      "loss": 0.039,
+      "step": 19154
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.134056753252989e-06,
+      "loss": 0.0002,
+      "step": 19155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.131656023431124e-06,
+      "loss": 0.0003,
+      "step": 19156
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.129255293609258e-06,
+      "loss": 0.0003,
+      "step": 19157
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1268545637873915e-06,
+      "loss": 0.0005,
+      "step": 19158
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.124453833965526e-06,
+      "loss": 0.0006,
+      "step": 19159
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.122053104143659e-06,
+      "loss": 0.0008,
+      "step": 19160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.119652374321794e-06,
+      "loss": 0.0002,
+      "step": 19161
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.117251644499928e-06,
+      "loss": 0.0003,
+      "step": 19162
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.114850914678062e-06,
+      "loss": 0.0002,
+      "step": 19163
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.112450184856197e-06,
+      "loss": 0.0003,
+      "step": 19164
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.110049455034331e-06,
+      "loss": 0.0009,
+      "step": 19165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.107648725212465e-06,
+      "loss": 0.0002,
+      "step": 19166
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.105247995390599e-06,
+      "loss": 0.0006,
+      "step": 19167
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.102847265568733e-06,
+      "loss": 0.0008,
+      "step": 19168
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.100446535746868e-06,
+      "loss": 0.1927,
+      "step": 19169
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.098045805925001e-06,
+      "loss": 0.0003,
+      "step": 19170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.095645076103135e-06,
+      "loss": 0.0004,
+      "step": 19171
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.09324434628127e-06,
+      "loss": 0.0002,
+      "step": 19172
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.090843616459404e-06,
+      "loss": 0.0003,
+      "step": 19173
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.088442886637538e-06,
+      "loss": 0.0009,
+      "step": 19174
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.086042156815672e-06,
+      "loss": 0.0004,
+      "step": 19175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.083641426993806e-06,
+      "loss": 0.0004,
+      "step": 19176
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.081240697171941e-06,
+      "loss": 0.0003,
+      "step": 19177
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.078839967350075e-06,
+      "loss": 0.0002,
+      "step": 19178
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0764392375282085e-06,
+      "loss": 0.0002,
+      "step": 19179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.074038507706343e-06,
+      "loss": 0.0002,
+      "step": 19180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.071637777884477e-06,
+      "loss": 0.0007,
+      "step": 19181
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.069237048062611e-06,
+      "loss": 0.0003,
+      "step": 19182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.066836318240745e-06,
+      "loss": 0.0003,
+      "step": 19183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.064435588418879e-06,
+      "loss": 0.0002,
+      "step": 19184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.062034858597014e-06,
+      "loss": 0.0003,
+      "step": 19185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.059634128775148e-06,
+      "loss": 0.0003,
+      "step": 19186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.057233398953282e-06,
+      "loss": 0.0002,
+      "step": 19187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.054832669131417e-06,
+      "loss": 0.0003,
+      "step": 19188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.05243193930955e-06,
+      "loss": 0.0002,
+      "step": 19189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0500312094876845e-06,
+      "loss": 0.0002,
+      "step": 19190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.047630479665818e-06,
+      "loss": 0.0002,
+      "step": 19191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.045229749843952e-06,
+      "loss": 0.0003,
+      "step": 19192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.042829020022087e-06,
+      "loss": 0.0002,
+      "step": 19193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.040428290200221e-06,
+      "loss": 0.0003,
+      "step": 19194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.038027560378355e-06,
+      "loss": 0.0004,
+      "step": 19195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.03562683055649e-06,
+      "loss": 0.0004,
+      "step": 19196
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.033226100734624e-06,
+      "loss": 0.0002,
+      "step": 19197
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0308253709127576e-06,
+      "loss": 0.0003,
+      "step": 19198
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.028424641090892e-06,
+      "loss": 0.0004,
+      "step": 19199
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.026023911269025e-06,
+      "loss": 0.0002,
+      "step": 19200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.02362318144716e-06,
+      "loss": 0.0002,
+      "step": 19201
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.021222451625294e-06,
+      "loss": 0.0003,
+      "step": 19202
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.018821721803428e-06,
+      "loss": 0.9755,
+      "step": 19203
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.016420991981563e-06,
+      "loss": 0.0003,
+      "step": 19204
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.014020262159697e-06,
+      "loss": 0.0004,
+      "step": 19205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.011619532337831e-06,
+      "loss": 0.0004,
+      "step": 19206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.009218802515965e-06,
+      "loss": 0.0003,
+      "step": 19207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.006818072694099e-06,
+      "loss": 0.0003,
+      "step": 19208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.004417342872234e-06,
+      "loss": 0.0004,
+      "step": 19209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.002016613050367e-06,
+      "loss": 0.0004,
+      "step": 19210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9996158832285014e-06,
+      "loss": 1.0546,
+      "step": 19211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.997215153406636e-06,
+      "loss": 0.0003,
+      "step": 19212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.99481442358477e-06,
+      "loss": 0.0004,
+      "step": 19213
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.992413693762904e-06,
+      "loss": 0.0002,
+      "step": 19214
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.990012963941038e-06,
+      "loss": 0.0003,
+      "step": 19215
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.987612234119172e-06,
+      "loss": 0.0003,
+      "step": 19216
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.985211504297307e-06,
+      "loss": 0.0003,
+      "step": 19217
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.982810774475441e-06,
+      "loss": 0.0003,
+      "step": 19218
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.980410044653575e-06,
+      "loss": 0.0004,
+      "step": 19219
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.978009314831709e-06,
+      "loss": 0.0002,
+      "step": 19220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.975608585009843e-06,
+      "loss": 0.0002,
+      "step": 19221
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.973207855187977e-06,
+      "loss": 0.0004,
+      "step": 19222
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.970807125366111e-06,
+      "loss": 0.0003,
+      "step": 19223
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.968406395544245e-06,
+      "loss": 0.0003,
+      "step": 19224
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.96600566572238e-06,
+      "loss": 0.0003,
+      "step": 19225
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.963604935900514e-06,
+      "loss": 0.0003,
+      "step": 19226
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.961204206078648e-06,
+      "loss": 0.0003,
+      "step": 19227
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.958803476256783e-06,
+      "loss": 0.0003,
+      "step": 19228
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.956402746434917e-06,
+      "loss": 0.0003,
+      "step": 19229
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9540020166130505e-06,
+      "loss": 0.0003,
+      "step": 19230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.951601286791184e-06,
+      "loss": 0.0002,
+      "step": 19231
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.949200556969318e-06,
+      "loss": 0.0003,
+      "step": 19232
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.946799827147453e-06,
+      "loss": 0.0002,
+      "step": 19233
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.944399097325587e-06,
+      "loss": 0.0003,
+      "step": 19234
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.941998367503721e-06,
+      "loss": 0.0003,
+      "step": 19235
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.939597637681856e-06,
+      "loss": 0.0003,
+      "step": 19236
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.93719690785999e-06,
+      "loss": 0.0003,
+      "step": 19237
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.934796178038124e-06,
+      "loss": 0.0003,
+      "step": 19238
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.932395448216258e-06,
+      "loss": 0.0003,
+      "step": 19239
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.929994718394392e-06,
+      "loss": 0.0005,
+      "step": 19240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.927593988572526e-06,
+      "loss": 0.0004,
+      "step": 19241
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.92519325875066e-06,
+      "loss": 0.0004,
+      "step": 19242
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.9227925289287944e-06,
+      "loss": 0.0003,
+      "step": 19243
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.920391799106929e-06,
+      "loss": 0.0003,
+      "step": 19244
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.917991069285063e-06,
+      "loss": 0.0002,
+      "step": 19245
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.915590339463197e-06,
+      "loss": 0.0006,
+      "step": 19246
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.913189609641331e-06,
+      "loss": 0.0003,
+      "step": 19247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.910788879819465e-06,
+      "loss": 0.0003,
+      "step": 19248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.9083881499976e-06,
+      "loss": 0.0003,
+      "step": 19249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.905987420175734e-06,
+      "loss": 0.0002,
+      "step": 19250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.9035866903538675e-06,
+      "loss": 0.0004,
+      "step": 19251
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.901185960532002e-06,
+      "loss": 0.0002,
+      "step": 19252
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.898785230710136e-06,
+      "loss": 0.0004,
+      "step": 19253
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.89638450088827e-06,
+      "loss": 0.0003,
+      "step": 19254
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.893983771066404e-06,
+      "loss": 0.0004,
+      "step": 19255
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.891583041244538e-06,
+      "loss": 0.0003,
+      "step": 19256
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.889182311422673e-06,
+      "loss": 0.0004,
+      "step": 19257
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.886781581600807e-06,
+      "loss": 0.0003,
+      "step": 19258
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.884380851778941e-06,
+      "loss": 0.0003,
+      "step": 19259
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.881980121957076e-06,
+      "loss": 0.0004,
+      "step": 19260
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.879579392135209e-06,
+      "loss": 0.0002,
+      "step": 19261
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.877178662313343e-06,
+      "loss": 0.0003,
+      "step": 19262
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.874777932491477e-06,
+      "loss": 0.0003,
+      "step": 19263
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.872377202669611e-06,
+      "loss": 0.0002,
+      "step": 19264
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.869976472847746e-06,
+      "loss": 0.0003,
+      "step": 19265
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.86757574302588e-06,
+      "loss": 0.0003,
+      "step": 19266
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.865175013204014e-06,
+      "loss": 0.2405,
+      "step": 19267
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.862774283382149e-06,
+      "loss": 0.0003,
+      "step": 19268
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.860373553560283e-06,
+      "loss": 0.0003,
+      "step": 19269
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.857972823738417e-06,
+      "loss": 0.0003,
+      "step": 19270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.855572093916551e-06,
+      "loss": 0.0003,
+      "step": 19271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.853171364094684e-06,
+      "loss": 0.0003,
+      "step": 19272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.850770634272819e-06,
+      "loss": 0.0003,
+      "step": 19273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.848369904450953e-06,
+      "loss": 0.0003,
+      "step": 19274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8459691746290874e-06,
+      "loss": 0.0004,
+      "step": 19275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.843568444807222e-06,
+      "loss": 0.0003,
+      "step": 19276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.841167714985356e-06,
+      "loss": 0.0002,
+      "step": 19277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.83876698516349e-06,
+      "loss": 0.9208,
+      "step": 19278
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.836366255341624e-06,
+      "loss": 0.0002,
+      "step": 19279
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.833965525519758e-06,
+      "loss": 0.0004,
+      "step": 19280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.831564795697893e-06,
+      "loss": 0.0002,
+      "step": 19281
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.829164065876026e-06,
+      "loss": 0.0003,
+      "step": 19282
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8267633360541605e-06,
+      "loss": 0.0003,
+      "step": 19283
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.824362606232295e-06,
+      "loss": 0.0003,
+      "step": 19284
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.821961876410429e-06,
+      "loss": 0.0003,
+      "step": 19285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.819561146588563e-06,
+      "loss": 1.0766,
+      "step": 19286
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.817160416766697e-06,
+      "loss": 0.0003,
+      "step": 19287
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.814759686944831e-06,
+      "loss": 0.0002,
+      "step": 19288
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8123589571229657e-06,
+      "loss": 0.0003,
+      "step": 19289
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8099582273010996e-06,
+      "loss": 0.0003,
+      "step": 19290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.807557497479234e-06,
+      "loss": 0.0003,
+      "step": 19291
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8051567676573683e-06,
+      "loss": 0.0003,
+      "step": 19292
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8027560378355026e-06,
+      "loss": 0.0003,
+      "step": 19293
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.800355308013636e-06,
+      "loss": 0.0003,
+      "step": 19294
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7979545781917705e-06,
+      "loss": 0.0003,
+      "step": 19295
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7955538483699044e-06,
+      "loss": 0.0003,
+      "step": 19296
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7931531185480387e-06,
+      "loss": 0.0003,
+      "step": 19297
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.790752388726173e-06,
+      "loss": 0.0003,
+      "step": 19298
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7883516589043074e-06,
+      "loss": 0.0003,
+      "step": 19299
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7859509290824413e-06,
+      "loss": 0.0002,
+      "step": 19300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7835501992605757e-06,
+      "loss": 0.0002,
+      "step": 19301
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.781149469438709e-06,
+      "loss": 0.0003,
+      "step": 19302
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7787487396168435e-06,
+      "loss": 0.0003,
+      "step": 19303
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.776348009794978e-06,
+      "loss": 0.0005,
+      "step": 19304
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7739472799731118e-06,
+      "loss": 0.0022,
+      "step": 19305
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.771546550151246e-06,
+      "loss": 0.0003,
+      "step": 19306
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7691458203293804e-06,
+      "loss": 0.0002,
+      "step": 19307
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7667450905075148e-06,
+      "loss": 0.0012,
+      "step": 19308
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.764344360685649e-06,
+      "loss": 0.0003,
+      "step": 19309
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7619436308637826e-06,
+      "loss": 0.0003,
+      "step": 19310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7595429010419165e-06,
+      "loss": 0.0003,
+      "step": 19311
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.757142171220051e-06,
+      "loss": 0.0003,
+      "step": 19312
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7547414413981852e-06,
+      "loss": 0.0003,
+      "step": 19313
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7523407115763196e-06,
+      "loss": 0.0005,
+      "step": 19314
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7499399817544535e-06,
+      "loss": 0.4672,
+      "step": 19315
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.747539251932588e-06,
+      "loss": 0.4545,
+      "step": 19316
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.745138522110722e-06,
+      "loss": 0.0002,
+      "step": 19317
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7427377922888557e-06,
+      "loss": 0.0002,
+      "step": 19318
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.74033706246699e-06,
+      "loss": 0.0003,
+      "step": 19319
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7379363326451243e-06,
+      "loss": 0.0002,
+      "step": 19320
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7355356028232583e-06,
+      "loss": 0.0008,
+      "step": 19321
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7331348730013926e-06,
+      "loss": 0.0003,
+      "step": 19322
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.730734143179527e-06,
+      "loss": 0.0003,
+      "step": 19323
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7283334133576613e-06,
+      "loss": 0.0005,
+      "step": 19324
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.725932683535795e-06,
+      "loss": 0.0002,
+      "step": 19325
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.723531953713929e-06,
+      "loss": 0.0005,
+      "step": 19326
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.721131223892063e-06,
+      "loss": 0.0004,
+      "step": 19327
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7187304940701974e-06,
+      "loss": 0.5355,
+      "step": 19328
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7163297642483317e-06,
+      "loss": 0.0002,
+      "step": 19329
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.713929034426466e-06,
+      "loss": 0.0003,
+      "step": 19330
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7115283046046e-06,
+      "loss": 0.0005,
+      "step": 19331
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7091275747827343e-06,
+      "loss": 0.0003,
+      "step": 19332
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7067268449608687e-06,
+      "loss": 0.0006,
+      "step": 19333
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.704326115139002e-06,
+      "loss": 0.0003,
+      "step": 19334
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7019253853171365e-06,
+      "loss": 0.0005,
+      "step": 19335
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6995246554952704e-06,
+      "loss": 0.0003,
+      "step": 19336
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6971239256734048e-06,
+      "loss": 0.0005,
+      "step": 19337
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.694723195851539e-06,
+      "loss": 0.0026,
+      "step": 19338
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6923224660296734e-06,
+      "loss": 0.0003,
+      "step": 19339
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6899217362078078e-06,
+      "loss": 0.4518,
+      "step": 19340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6875210063859417e-06,
+      "loss": 0.0003,
+      "step": 19341
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.685120276564075e-06,
+      "loss": 0.0003,
+      "step": 19342
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6827195467422095e-06,
+      "loss": 0.0004,
+      "step": 19343
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.680318816920344e-06,
+      "loss": 0.0007,
+      "step": 19344
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.677918087098478e-06,
+      "loss": 0.0002,
+      "step": 19345
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.675517357276612e-06,
+      "loss": 0.0002,
+      "step": 19346
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6731166274547465e-06,
+      "loss": 0.0003,
+      "step": 19347
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.670715897632881e-06,
+      "loss": 0.0003,
+      "step": 19348
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.668315167811015e-06,
+      "loss": 0.0004,
+      "step": 19349
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6659144379891486e-06,
+      "loss": 0.0004,
+      "step": 19350
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.663513708167283e-06,
+      "loss": 0.9614,
+      "step": 19351
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.661112978345417e-06,
+      "loss": 0.0004,
+      "step": 19352
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6587122485235512e-06,
+      "loss": 0.0002,
+      "step": 19353
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6563115187016856e-06,
+      "loss": 0.0002,
+      "step": 19354
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.65391078887982e-06,
+      "loss": 0.0004,
+      "step": 19355
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.651510059057954e-06,
+      "loss": 0.0003,
+      "step": 19356
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.649109329236088e-06,
+      "loss": 0.0005,
+      "step": 19357
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6467085994142217e-06,
+      "loss": 0.0004,
+      "step": 19358
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.644307869592356e-06,
+      "loss": 0.0005,
+      "step": 19359
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6419071397704904e-06,
+      "loss": 0.0003,
+      "step": 19360
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6395064099486247e-06,
+      "loss": 0.0003,
+      "step": 19361
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6371056801267586e-06,
+      "loss": 0.0004,
+      "step": 19362
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.634704950304893e-06,
+      "loss": 0.0002,
+      "step": 19363
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6323042204830273e-06,
+      "loss": 0.0006,
+      "step": 19364
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6299034906611617e-06,
+      "loss": 0.0003,
+      "step": 19365
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.627502760839295e-06,
+      "loss": 0.0003,
+      "step": 19366
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.625102031017429e-06,
+      "loss": 0.0003,
+      "step": 19367
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6227013011955634e-06,
+      "loss": 0.0003,
+      "step": 19368
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6203005713736977e-06,
+      "loss": 0.0003,
+      "step": 19369
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.617899841551832e-06,
+      "loss": 0.0003,
+      "step": 19370
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6154991117299664e-06,
+      "loss": 0.0003,
+      "step": 19371
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6130983819081003e-06,
+      "loss": 0.0005,
+      "step": 19372
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6106976520862347e-06,
+      "loss": 0.0004,
+      "step": 19373
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.608296922264368e-06,
+      "loss": 0.0003,
+      "step": 19374
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6058961924425025e-06,
+      "loss": 0.0003,
+      "step": 19375
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.603495462620637e-06,
+      "loss": 0.0004,
+      "step": 19376
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6010947327987708e-06,
+      "loss": 0.0004,
+      "step": 19377
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.598694002976905e-06,
+      "loss": 0.0003,
+      "step": 19378
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.5962932731550395e-06,
+      "loss": 0.0004,
+      "step": 19379
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.593892543333174e-06,
+      "loss": 0.0003,
+      "step": 19380
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5914918135113077e-06,
+      "loss": 0.0003,
+      "step": 19381
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5890910836894416e-06,
+      "loss": 0.0003,
+      "step": 19382
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5866903538675756e-06,
+      "loss": 1.0134,
+      "step": 19383
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.58428962404571e-06,
+      "loss": 0.0003,
+      "step": 19384
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5818888942238442e-06,
+      "loss": 0.0003,
+      "step": 19385
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5794881644019786e-06,
+      "loss": 0.0003,
+      "step": 19386
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5770874345801125e-06,
+      "loss": 0.0004,
+      "step": 19387
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.574686704758247e-06,
+      "loss": 0.0004,
+      "step": 19388
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.572285974936381e-06,
+      "loss": 1.0221,
+      "step": 19389
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5698852451145147e-06,
+      "loss": 0.0003,
+      "step": 19390
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.567484515292649e-06,
+      "loss": 0.0003,
+      "step": 19391
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5650837854707834e-06,
+      "loss": 0.0004,
+      "step": 19392
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5626830556489173e-06,
+      "loss": 0.0003,
+      "step": 19393
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5602823258270516e-06,
+      "loss": 0.0005,
+      "step": 19394
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.557881596005186e-06,
+      "loss": 0.0005,
+      "step": 19395
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5554808661833203e-06,
+      "loss": 0.0003,
+      "step": 19396
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5530801363614542e-06,
+      "loss": 0.0003,
+      "step": 19397
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5506794065395877e-06,
+      "loss": 0.0004,
+      "step": 19398
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.548278676717722e-06,
+      "loss": 0.0003,
+      "step": 19399
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5458779468958564e-06,
+      "loss": 0.0005,
+      "step": 19400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5434772170739907e-06,
+      "loss": 0.0004,
+      "step": 19401
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.541076487252125e-06,
+      "loss": 0.0003,
+      "step": 19402
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.538675757430259e-06,
+      "loss": 0.0004,
+      "step": 19403
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5362750276083933e-06,
+      "loss": 0.0003,
+      "step": 19404
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5338742977865277e-06,
+      "loss": 0.0003,
+      "step": 19405
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.531473567964661e-06,
+      "loss": 0.0003,
+      "step": 19406
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5290728381427955e-06,
+      "loss": 0.0003,
+      "step": 19407
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5266721083209294e-06,
+      "loss": 0.0004,
+      "step": 19408
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5242713784990638e-06,
+      "loss": 0.0004,
+      "step": 19409
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.521870648677198e-06,
+      "loss": 0.0003,
+      "step": 19410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5194699188553325e-06,
+      "loss": 0.0003,
+      "step": 19411
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5170691890334664e-06,
+      "loss": 0.0003,
+      "step": 19412
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5146684592116007e-06,
+      "loss": 0.0004,
+      "step": 19413
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5122677293897342e-06,
+      "loss": 0.0003,
+      "step": 19414
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5098669995678686e-06,
+      "loss": 0.0003,
+      "step": 19415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.507466269746003e-06,
+      "loss": 0.0002,
+      "step": 19416
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5050655399241372e-06,
+      "loss": 0.0005,
+      "step": 19417
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.502664810102271e-06,
+      "loss": 0.0005,
+      "step": 19418
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5002640802804055e-06,
+      "loss": 0.0004,
+      "step": 19419
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.49786335045854e-06,
+      "loss": 0.0004,
+      "step": 19420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.495462620636674e-06,
+      "loss": 0.0003,
+      "step": 19421
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4930618908148077e-06,
+      "loss": 0.0003,
+      "step": 19422
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.490661160992942e-06,
+      "loss": 0.0006,
+      "step": 19423
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.488260431171076e-06,
+      "loss": 0.0003,
+      "step": 19424
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4858597013492103e-06,
+      "loss": 0.0002,
+      "step": 19425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4834589715273446e-06,
+      "loss": 1.0163,
+      "step": 19426
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.481058241705479e-06,
+      "loss": 0.0004,
+      "step": 19427
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.478657511883613e-06,
+      "loss": 0.005,
+      "step": 19428
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4762567820617472e-06,
+      "loss": 0.0004,
+      "step": 19429
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4738560522398807e-06,
+      "loss": 0.0003,
+      "step": 19430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.471455322418015e-06,
+      "loss": 0.0005,
+      "step": 19431
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4690545925961494e-06,
+      "loss": 0.0003,
+      "step": 19432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4666538627742837e-06,
+      "loss": 0.0004,
+      "step": 19433
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4642531329524177e-06,
+      "loss": 0.0004,
+      "step": 19434
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.461852403130552e-06,
+      "loss": 0.0003,
+      "step": 19435
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4594516733086863e-06,
+      "loss": 0.0004,
+      "step": 19436
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4570509434868207e-06,
+      "loss": 0.0004,
+      "step": 19437
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.454650213664954e-06,
+      "loss": 0.0004,
+      "step": 19438
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.452249483843088e-06,
+      "loss": 0.0003,
+      "step": 19439
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4498487540212224e-06,
+      "loss": 0.0004,
+      "step": 19440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4474480241993568e-06,
+      "loss": 0.0003,
+      "step": 19441
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.445047294377491e-06,
+      "loss": 0.0003,
+      "step": 19442
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.442646564555625e-06,
+      "loss": 0.0003,
+      "step": 19443
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4402458347337594e-06,
+      "loss": 0.0003,
+      "step": 19444
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4378451049118937e-06,
+      "loss": 0.0004,
+      "step": 19445
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4354443750900272e-06,
+      "loss": 0.0004,
+      "step": 19446
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4330436452681616e-06,
+      "loss": 0.0004,
+      "step": 19447
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.430642915446296e-06,
+      "loss": 0.0004,
+      "step": 19448
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.42824218562443e-06,
+      "loss": 0.0004,
+      "step": 19449
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.425841455802564e-06,
+      "loss": 0.0007,
+      "step": 19450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4234407259806985e-06,
+      "loss": 0.0004,
+      "step": 19451
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.421039996158833e-06,
+      "loss": 0.0004,
+      "step": 19452
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4186392663369668e-06,
+      "loss": 0.0003,
+      "step": 19453
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4162385365151007e-06,
+      "loss": 0.0004,
+      "step": 19454
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4138378066932346e-06,
+      "loss": 0.0003,
+      "step": 19455
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.411437076871369e-06,
+      "loss": 0.0003,
+      "step": 19456
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4090363470495033e-06,
+      "loss": 0.0004,
+      "step": 19457
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4066356172276376e-06,
+      "loss": 0.0003,
+      "step": 19458
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4042348874057715e-06,
+      "loss": 0.0003,
+      "step": 19459
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.401834157583906e-06,
+      "loss": 0.0004,
+      "step": 19460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3994334277620402e-06,
+      "loss": 0.0003,
+      "step": 19461
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3970326979401737e-06,
+      "loss": 0.0004,
+      "step": 19462
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.394631968118308e-06,
+      "loss": 0.0004,
+      "step": 19463
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3922312382964424e-06,
+      "loss": 0.0004,
+      "step": 19464
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3898305084745763e-06,
+      "loss": 0.0003,
+      "step": 19465
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3874297786527107e-06,
+      "loss": 0.0003,
+      "step": 19466
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.385029048830845e-06,
+      "loss": 0.0003,
+      "step": 19467
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3826283190089793e-06,
+      "loss": 0.0004,
+      "step": 19468
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3802275891871133e-06,
+      "loss": 0.0003,
+      "step": 19469
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3778268593652467e-06,
+      "loss": 0.0003,
+      "step": 19470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.375426129543381e-06,
+      "loss": 0.0004,
+      "step": 19471
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3730253997215154e-06,
+      "loss": 0.0003,
+      "step": 19472
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3706246698996498e-06,
+      "loss": 0.0003,
+      "step": 19473
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3682239400777837e-06,
+      "loss": 0.0002,
+      "step": 19474
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.365823210255918e-06,
+      "loss": 0.0004,
+      "step": 19475
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3634224804340524e-06,
+      "loss": 0.0004,
+      "step": 19476
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3610217506121867e-06,
+      "loss": 0.0003,
+      "step": 19477
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.35862102079032e-06,
+      "loss": 0.0003,
+      "step": 19478
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3562202909684546e-06,
+      "loss": 0.0004,
+      "step": 19479
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3538195611465885e-06,
+      "loss": 0.0003,
+      "step": 19480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.351418831324723e-06,
+      "loss": 0.0004,
+      "step": 19481
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.349018101502857e-06,
+      "loss": 0.0002,
+      "step": 19482
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3466173716809915e-06,
+      "loss": 0.0003,
+      "step": 19483
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3442166418591254e-06,
+      "loss": 0.0003,
+      "step": 19484
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3418159120372598e-06,
+      "loss": 0.0003,
+      "step": 19485
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3394151822153932e-06,
+      "loss": 0.0004,
+      "step": 19486
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3370144523935276e-06,
+      "loss": 0.0004,
+      "step": 19487
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.334613722571662e-06,
+      "loss": 0.0003,
+      "step": 19488
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3322129927497963e-06,
+      "loss": 0.0004,
+      "step": 19489
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.32981226292793e-06,
+      "loss": 0.0003,
+      "step": 19490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3274115331060645e-06,
+      "loss": 0.0003,
+      "step": 19491
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.325010803284199e-06,
+      "loss": 0.0004,
+      "step": 19492
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3226100734623332e-06,
+      "loss": 0.0005,
+      "step": 19493
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3202093436404667e-06,
+      "loss": 0.0003,
+      "step": 19494
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3178086138186006e-06,
+      "loss": 0.0003,
+      "step": 19495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.315407883996735e-06,
+      "loss": 0.0003,
+      "step": 19496
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3130071541748693e-06,
+      "loss": 0.0005,
+      "step": 19497
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3106064243530036e-06,
+      "loss": 0.0002,
+      "step": 19498
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.308205694531138e-06,
+      "loss": 0.0004,
+      "step": 19499
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.305804964709272e-06,
+      "loss": 0.0004,
+      "step": 19500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3034042348874062e-06,
+      "loss": 0.0003,
+      "step": 19501
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3010035050655397e-06,
+      "loss": 0.0003,
+      "step": 19502
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.298602775243674e-06,
+      "loss": 0.0003,
+      "step": 19503
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2962020454218084e-06,
+      "loss": 0.0004,
+      "step": 19504
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2938013155999423e-06,
+      "loss": 0.0003,
+      "step": 19505
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2914005857780767e-06,
+      "loss": 0.0004,
+      "step": 19506
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.288999855956211e-06,
+      "loss": 0.0003,
+      "step": 19507
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2865991261343454e-06,
+      "loss": 0.0002,
+      "step": 19508
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2841983963124793e-06,
+      "loss": 0.0003,
+      "step": 19509
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.281797666490613e-06,
+      "loss": 0.0005,
+      "step": 19510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.279396936668747e-06,
+      "loss": 0.0002,
+      "step": 19511
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2769962068468815e-06,
+      "loss": 0.0002,
+      "step": 19512
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.274595477025016e-06,
+      "loss": 0.0003,
+      "step": 19513
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.27219474720315e-06,
+      "loss": 0.0003,
+      "step": 19514
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.269794017381284e-06,
+      "loss": 0.0003,
+      "step": 19515
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2673932875594184e-06,
+      "loss": 0.0004,
+      "step": 19516
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.264992557737552e-06,
+      "loss": 0.0004,
+      "step": 19517
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2625918279156862e-06,
+      "loss": 0.0003,
+      "step": 19518
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2601910980938206e-06,
+      "loss": 0.0004,
+      "step": 19519
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.257790368271955e-06,
+      "loss": 0.0001,
+      "step": 19520
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.255389638450089e-06,
+      "loss": 0.0003,
+      "step": 19521
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.252988908628223e-06,
+      "loss": 0.0003,
+      "step": 19522
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2505881788063575e-06,
+      "loss": 0.0003,
+      "step": 19523
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.248187448984492e-06,
+      "loss": 0.0004,
+      "step": 19524
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2457867191626254e-06,
+      "loss": 0.0003,
+      "step": 19525
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2433859893407593e-06,
+      "loss": 0.0003,
+      "step": 19526
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2409852595188936e-06,
+      "loss": 0.0004,
+      "step": 19527
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.238584529697028e-06,
+      "loss": 0.0003,
+      "step": 19528
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2361837998751623e-06,
+      "loss": 0.0004,
+      "step": 19529
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2337830700532966e-06,
+      "loss": 0.0004,
+      "step": 19530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2313823402314306e-06,
+      "loss": 0.0003,
+      "step": 19531
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.228981610409565e-06,
+      "loss": 0.0002,
+      "step": 19532
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2265808805876984e-06,
+      "loss": 0.0003,
+      "step": 19533
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2241801507658327e-06,
+      "loss": 0.0003,
+      "step": 19534
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.221779420943967e-06,
+      "loss": 0.0003,
+      "step": 19535
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.219378691122101e-06,
+      "loss": 0.0002,
+      "step": 19536
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2169779613002353e-06,
+      "loss": 0.0004,
+      "step": 19537
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2145772314783697e-06,
+      "loss": 0.0002,
+      "step": 19538
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.212176501656504e-06,
+      "loss": 0.0003,
+      "step": 19539
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.209775771834638e-06,
+      "loss": 0.0004,
+      "step": 19540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.207375042012772e-06,
+      "loss": 0.0005,
+      "step": 19541
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2049743121909058e-06,
+      "loss": 0.8488,
+      "step": 19542
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.20257358236904e-06,
+      "loss": 0.0004,
+      "step": 19543
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2001728525471745e-06,
+      "loss": 0.0004,
+      "step": 19544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.197772122725309e-06,
+      "loss": 0.0003,
+      "step": 19545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1953713929034427e-06,
+      "loss": 0.0006,
+      "step": 19546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.192970663081577e-06,
+      "loss": 0.0003,
+      "step": 19547
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1905699332597114e-06,
+      "loss": 0.0002,
+      "step": 19548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.188169203437845e-06,
+      "loss": 0.0004,
+      "step": 19549
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1857684736159792e-06,
+      "loss": 0.0004,
+      "step": 19550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1833677437941136e-06,
+      "loss": 0.0004,
+      "step": 19551
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1809670139722475e-06,
+      "loss": 0.0003,
+      "step": 19552
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.178566284150382e-06,
+      "loss": 0.0003,
+      "step": 19553
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.176165554328516e-06,
+      "loss": 0.0003,
+      "step": 19554
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1737648245066505e-06,
+      "loss": 0.0004,
+      "step": 19555
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1713640946847844e-06,
+      "loss": 0.0003,
+      "step": 19556
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.168963364862918e-06,
+      "loss": 0.0002,
+      "step": 19557
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1665626350410523e-06,
+      "loss": 0.0003,
+      "step": 19558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1641619052191866e-06,
+      "loss": 0.0005,
+      "step": 19559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.161761175397321e-06,
+      "loss": 0.0003,
+      "step": 19560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1593604455754553e-06,
+      "loss": 0.0004,
+      "step": 19561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1569597157535892e-06,
+      "loss": 0.0005,
+      "step": 19562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1545589859317236e-06,
+      "loss": 0.0005,
+      "step": 19563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.152158256109858e-06,
+      "loss": 0.0003,
+      "step": 19564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1497575262879914e-06,
+      "loss": 0.0003,
+      "step": 19565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1473567964661257e-06,
+      "loss": 0.0004,
+      "step": 19566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1449560666442597e-06,
+      "loss": 0.0004,
+      "step": 19567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.142555336822394e-06,
+      "loss": 0.0003,
+      "step": 19568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1401546070005283e-06,
+      "loss": 0.0003,
+      "step": 19569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1377538771786627e-06,
+      "loss": 0.0004,
+      "step": 19570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1353531473567966e-06,
+      "loss": 0.0003,
+      "step": 19571
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.132952417534931e-06,
+      "loss": 0.0003,
+      "step": 19572
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1305516877130644e-06,
+      "loss": 0.0005,
+      "step": 19573
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1281509578911988e-06,
+      "loss": 0.0003,
+      "step": 19574
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.125750228069333e-06,
+      "loss": 0.0003,
+      "step": 19575
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1233494982474675e-06,
+      "loss": 0.0005,
+      "step": 19576
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1209487684256014e-06,
+      "loss": 0.0002,
+      "step": 19577
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1185480386037357e-06,
+      "loss": 0.0003,
+      "step": 19578
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1161473087818696e-06,
+      "loss": 0.0002,
+      "step": 19579
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.113746578960004e-06,
+      "loss": 0.0004,
+      "step": 19580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1113458491381383e-06,
+      "loss": 0.0003,
+      "step": 19581
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1089451193162722e-06,
+      "loss": 0.0003,
+      "step": 19582
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.106544389494406e-06,
+      "loss": 0.0003,
+      "step": 19583
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1041436596725405e-06,
+      "loss": 0.0003,
+      "step": 19584
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.101742929850675e-06,
+      "loss": 0.0004,
+      "step": 19585
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.099342200028809e-06,
+      "loss": 0.0005,
+      "step": 19586
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.096941470206943e-06,
+      "loss": 0.0003,
+      "step": 19587
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.094540740385077e-06,
+      "loss": 0.0004,
+      "step": 19588
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.0921400105632114e-06,
+      "loss": 0.0004,
+      "step": 19589
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0897392807413457e-06,
+      "loss": 0.0002,
+      "step": 19590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0873385509194796e-06,
+      "loss": 0.0002,
+      "step": 19591
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.084937821097614e-06,
+      "loss": 0.0002,
+      "step": 19592
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.082537091275748e-06,
+      "loss": 0.0004,
+      "step": 19593
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0801363614538822e-06,
+      "loss": 0.0004,
+      "step": 19594
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.077735631632016e-06,
+      "loss": 0.8256,
+      "step": 19595
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0753349018101505e-06,
+      "loss": 0.0004,
+      "step": 19596
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.072934171988285e-06,
+      "loss": 0.0004,
+      "step": 19597
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0705334421664187e-06,
+      "loss": 0.0004,
+      "step": 19598
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0681327123445527e-06,
+      "loss": 0.452,
+      "step": 19599
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.065731982522687e-06,
+      "loss": 0.0004,
+      "step": 19600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0633312527008213e-06,
+      "loss": 0.0003,
+      "step": 19601
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0609305228789553e-06,
+      "loss": 0.0003,
+      "step": 19602
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.058529793057089e-06,
+      "loss": 0.0003,
+      "step": 19603
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0561290632352235e-06,
+      "loss": 0.0003,
+      "step": 19604
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.053728333413358e-06,
+      "loss": 0.0003,
+      "step": 19605
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.051327603591492e-06,
+      "loss": 0.0003,
+      "step": 19606
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.048926873769626e-06,
+      "loss": 0.0002,
+      "step": 19607
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.04652614394776e-06,
+      "loss": 0.0003,
+      "step": 19608
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0441254141258944e-06,
+      "loss": 0.0003,
+      "step": 19609
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0417246843040287e-06,
+      "loss": 0.0004,
+      "step": 19610
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0393239544821626e-06,
+      "loss": 0.0003,
+      "step": 19611
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.036923224660297e-06,
+      "loss": 0.0003,
+      "step": 19612
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.034522494838431e-06,
+      "loss": 0.0006,
+      "step": 19613
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0321217650165652e-06,
+      "loss": 0.0006,
+      "step": 19614
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.029721035194699e-06,
+      "loss": 0.0002,
+      "step": 19615
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0273203053728335e-06,
+      "loss": 0.0003,
+      "step": 19616
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.024919575550968e-06,
+      "loss": 0.0003,
+      "step": 19617
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0225188457291017e-06,
+      "loss": 0.0003,
+      "step": 19618
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0201181159072357e-06,
+      "loss": 0.0003,
+      "step": 19619
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.01771738608537e-06,
+      "loss": 0.0003,
+      "step": 19620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0153166562635043e-06,
+      "loss": 0.0002,
+      "step": 19621
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0129159264416387e-06,
+      "loss": 0.0005,
+      "step": 19622
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0105151966197726e-06,
+      "loss": 0.0007,
+      "step": 19623
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0081144667979065e-06,
+      "loss": 0.0003,
+      "step": 19624
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.005713736976041e-06,
+      "loss": 0.0029,
+      "step": 19625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.003313007154175e-06,
+      "loss": 0.0002,
+      "step": 19626
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.000912277332309e-06,
+      "loss": 0.5464,
+      "step": 19627
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9985115475104435e-06,
+      "loss": 0.0007,
+      "step": 19628
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9961108176885774e-06,
+      "loss": 0.0003,
+      "step": 19629
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9937100878667117e-06,
+      "loss": 0.0028,
+      "step": 19630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9913093580448456e-06,
+      "loss": 0.0009,
+      "step": 19631
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.98890862822298e-06,
+      "loss": 0.0002,
+      "step": 19632
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.986507898401114e-06,
+      "loss": 0.0008,
+      "step": 19633
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9841071685792482e-06,
+      "loss": 1.0253,
+      "step": 19634
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.981706438757382e-06,
+      "loss": 0.0003,
+      "step": 19635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9793057089355165e-06,
+      "loss": 0.0006,
+      "step": 19636
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.976904979113651e-06,
+      "loss": 0.0007,
+      "step": 19637
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9745042492917848e-06,
+      "loss": 0.0002,
+      "step": 19638
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9721035194699187e-06,
+      "loss": 0.0003,
+      "step": 19639
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.969702789648053e-06,
+      "loss": 0.0003,
+      "step": 19640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9673020598261874e-06,
+      "loss": 0.0016,
+      "step": 19641
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9649013300043217e-06,
+      "loss": 0.0002,
+      "step": 19642
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9625006001824556e-06,
+      "loss": 0.0008,
+      "step": 19643
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9600998703605895e-06,
+      "loss": 0.0003,
+      "step": 19644
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.957699140538724e-06,
+      "loss": 0.0003,
+      "step": 19645
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9552984107168582e-06,
+      "loss": 0.0002,
+      "step": 19646
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.952897680894992e-06,
+      "loss": 0.0003,
+      "step": 19647
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9504969510731265e-06,
+      "loss": 0.0005,
+      "step": 19648
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9480962212512604e-06,
+      "loss": 0.0003,
+      "step": 19649
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9456954914293947e-06,
+      "loss": 0.0003,
+      "step": 19650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9432947616075287e-06,
+      "loss": 0.0005,
+      "step": 19651
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.940894031785663e-06,
+      "loss": 0.0004,
+      "step": 19652
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9384933019637973e-06,
+      "loss": 0.0002,
+      "step": 19653
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9360925721419313e-06,
+      "loss": 0.0003,
+      "step": 19654
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.933691842320065e-06,
+      "loss": 0.0004,
+      "step": 19655
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9312911124981995e-06,
+      "loss": 0.0003,
+      "step": 19656
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.928890382676334e-06,
+      "loss": 0.0004,
+      "step": 19657
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.926489652854468e-06,
+      "loss": 0.0003,
+      "step": 19658
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.924088923032602e-06,
+      "loss": 0.0004,
+      "step": 19659
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.921688193210736e-06,
+      "loss": 0.0003,
+      "step": 19660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9192874633888704e-06,
+      "loss": 0.0002,
+      "step": 19661
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9168867335670047e-06,
+      "loss": 0.0003,
+      "step": 19662
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9144860037451386e-06,
+      "loss": 0.0003,
+      "step": 19663
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9120852739232726e-06,
+      "loss": 0.0003,
+      "step": 19664
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.909684544101407e-06,
+      "loss": 0.0003,
+      "step": 19665
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9072838142795412e-06,
+      "loss": 0.0002,
+      "step": 19666
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.904883084457675e-06,
+      "loss": 0.0004,
+      "step": 19667
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9024823546358095e-06,
+      "loss": 0.0003,
+      "step": 19668
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9000816248139434e-06,
+      "loss": 0.0002,
+      "step": 19669
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8976808949920778e-06,
+      "loss": 0.0004,
+      "step": 19670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8952801651702117e-06,
+      "loss": 0.0002,
+      "step": 19671
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.892879435348346e-06,
+      "loss": 0.0005,
+      "step": 19672
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8904787055264804e-06,
+      "loss": 0.0004,
+      "step": 19673
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8880779757046143e-06,
+      "loss": 0.0004,
+      "step": 19674
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.885677245882748e-06,
+      "loss": 0.0006,
+      "step": 19675
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8832765160608825e-06,
+      "loss": 0.0003,
+      "step": 19676
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.880875786239017e-06,
+      "loss": 0.0004,
+      "step": 19677
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8784750564171512e-06,
+      "loss": 0.0002,
+      "step": 19678
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.876074326595285e-06,
+      "loss": 0.0002,
+      "step": 19679
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.873673596773419e-06,
+      "loss": 0.0005,
+      "step": 19680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8712728669515534e-06,
+      "loss": 0.0005,
+      "step": 19681
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8688721371296877e-06,
+      "loss": 0.0003,
+      "step": 19682
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8664714073078217e-06,
+      "loss": 0.0002,
+      "step": 19683
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.864070677485956e-06,
+      "loss": 0.0003,
+      "step": 19684
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.86166994766409e-06,
+      "loss": 0.0005,
+      "step": 19685
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8592692178422243e-06,
+      "loss": 0.0003,
+      "step": 19686
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.856868488020358e-06,
+      "loss": 0.0003,
+      "step": 19687
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8544677581984925e-06,
+      "loss": 0.0003,
+      "step": 19688
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.852067028376627e-06,
+      "loss": 0.0008,
+      "step": 19689
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8496662985547608e-06,
+      "loss": 0.0005,
+      "step": 19690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8472655687328947e-06,
+      "loss": 0.0003,
+      "step": 19691
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.844864838911029e-06,
+      "loss": 0.0002,
+      "step": 19692
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8424641090891634e-06,
+      "loss": 0.0002,
+      "step": 19693
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8400633792672977e-06,
+      "loss": 0.0002,
+      "step": 19694
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8376626494454312e-06,
+      "loss": 0.0005,
+      "step": 19695
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8352619196235656e-06,
+      "loss": 0.0003,
+      "step": 19696
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8328611898017e-06,
+      "loss": 0.0003,
+      "step": 19697
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8304604599798342e-06,
+      "loss": 0.0006,
+      "step": 19698
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.828059730157968e-06,
+      "loss": 0.0003,
+      "step": 19699
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.825659000336102e-06,
+      "loss": 0.0002,
+      "step": 19700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8232582705142364e-06,
+      "loss": 0.0003,
+      "step": 19701
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8208575406923708e-06,
+      "loss": 0.0003,
+      "step": 19702
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8184568108705047e-06,
+      "loss": 0.0003,
+      "step": 19703
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.816056081048639e-06,
+      "loss": 0.0005,
+      "step": 19704
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.813655351226773e-06,
+      "loss": 0.0002,
+      "step": 19705
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8112546214049073e-06,
+      "loss": 0.0004,
+      "step": 19706
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.808853891583041e-06,
+      "loss": 0.0004,
+      "step": 19707
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8064531617611755e-06,
+      "loss": 0.0004,
+      "step": 19708
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.80405243193931e-06,
+      "loss": 0.0002,
+      "step": 19709
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.801651702117444e-06,
+      "loss": 0.0003,
+      "step": 19710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7992509722955777e-06,
+      "loss": 0.0004,
+      "step": 19711
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.796850242473712e-06,
+      "loss": 0.0003,
+      "step": 19712
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7944495126518464e-06,
+      "loss": 0.0004,
+      "step": 19713
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7920487828299807e-06,
+      "loss": 0.0004,
+      "step": 19714
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7896480530081147e-06,
+      "loss": 0.0004,
+      "step": 19715
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7872473231862486e-06,
+      "loss": 0.0003,
+      "step": 19716
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.784846593364383e-06,
+      "loss": 0.6347,
+      "step": 19717
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7824458635425173e-06,
+      "loss": 0.0004,
+      "step": 19718
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.780045133720651e-06,
+      "loss": 0.0003,
+      "step": 19719
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7776444038987855e-06,
+      "loss": 0.0003,
+      "step": 19720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7752436740769194e-06,
+      "loss": 0.0002,
+      "step": 19721
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7728429442550538e-06,
+      "loss": 0.0006,
+      "step": 19722
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7704422144331877e-06,
+      "loss": 0.0002,
+      "step": 19723
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.768041484611322e-06,
+      "loss": 0.0008,
+      "step": 19724
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7656407547894564e-06,
+      "loss": 0.0003,
+      "step": 19725
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7632400249675903e-06,
+      "loss": 0.0004,
+      "step": 19726
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.760839295145724e-06,
+      "loss": 0.0003,
+      "step": 19727
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.7584385653238586e-06,
+      "loss": 0.0003,
+      "step": 19728
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.756037835501993e-06,
+      "loss": 0.0005,
+      "step": 19729
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7536371056801272e-06,
+      "loss": 0.0003,
+      "step": 19730
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7512363758582607e-06,
+      "loss": 0.0009,
+      "step": 19731
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.748835646036395e-06,
+      "loss": 0.0004,
+      "step": 19732
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7464349162145294e-06,
+      "loss": 0.0003,
+      "step": 19733
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7440341863926638e-06,
+      "loss": 0.0004,
+      "step": 19734
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7416334565707977e-06,
+      "loss": 0.0002,
+      "step": 19735
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7392327267489316e-06,
+      "loss": 0.0003,
+      "step": 19736
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.736831996927066e-06,
+      "loss": 0.0003,
+      "step": 19737
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7344312671052003e-06,
+      "loss": 0.0003,
+      "step": 19738
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.732030537283334e-06,
+      "loss": 0.0003,
+      "step": 19739
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7296298074614685e-06,
+      "loss": 0.0006,
+      "step": 19740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7272290776396025e-06,
+      "loss": 0.0004,
+      "step": 19741
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.724828347817737e-06,
+      "loss": 0.0002,
+      "step": 19742
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7224276179958707e-06,
+      "loss": 0.0002,
+      "step": 19743
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.720026888174005e-06,
+      "loss": 0.0003,
+      "step": 19744
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7176261583521394e-06,
+      "loss": 0.0003,
+      "step": 19745
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7152254285302733e-06,
+      "loss": 0.0006,
+      "step": 19746
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7128246987084072e-06,
+      "loss": 0.0002,
+      "step": 19747
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7104239688865416e-06,
+      "loss": 0.0003,
+      "step": 19748
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.708023239064676e-06,
+      "loss": 0.0002,
+      "step": 19749
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7056225092428103e-06,
+      "loss": 0.0003,
+      "step": 19750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.703221779420944e-06,
+      "loss": 0.0003,
+      "step": 19751
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.700821049599078e-06,
+      "loss": 0.0003,
+      "step": 19752
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6984203197772124e-06,
+      "loss": 0.0007,
+      "step": 19753
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6960195899553468e-06,
+      "loss": 0.0003,
+      "step": 19754
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6936188601334807e-06,
+      "loss": 0.0002,
+      "step": 19755
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.691218130311615e-06,
+      "loss": 0.0002,
+      "step": 19756
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.688817400489749e-06,
+      "loss": 0.0003,
+      "step": 19757
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6864166706678833e-06,
+      "loss": 0.0003,
+      "step": 19758
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.684015940846017e-06,
+      "loss": 0.0004,
+      "step": 19759
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6816152110241515e-06,
+      "loss": 0.0002,
+      "step": 19760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.679214481202286e-06,
+      "loss": 0.0002,
+      "step": 19761
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.67681375138042e-06,
+      "loss": 0.0003,
+      "step": 19762
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6744130215585537e-06,
+      "loss": 0.0003,
+      "step": 19763
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.672012291736688e-06,
+      "loss": 0.23,
+      "step": 19764
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6696115619148224e-06,
+      "loss": 0.0003,
+      "step": 19765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6672108320929563e-06,
+      "loss": 0.0002,
+      "step": 19766
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6648101022710902e-06,
+      "loss": 0.0003,
+      "step": 19767
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6624093724492246e-06,
+      "loss": 0.0003,
+      "step": 19768
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.660008642627359e-06,
+      "loss": 0.0006,
+      "step": 19769
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6576079128054933e-06,
+      "loss": 0.0005,
+      "step": 19770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.655207182983627e-06,
+      "loss": 0.0002,
+      "step": 19771
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.652806453161761e-06,
+      "loss": 0.0004,
+      "step": 19772
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6504057233398954e-06,
+      "loss": 0.0016,
+      "step": 19773
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6480049935180298e-06,
+      "loss": 0.0006,
+      "step": 19774
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6456042636961637e-06,
+      "loss": 0.0004,
+      "step": 19775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.643203533874298e-06,
+      "loss": 0.0005,
+      "step": 19776
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.640802804052432e-06,
+      "loss": 0.0003,
+      "step": 19777
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6384020742305663e-06,
+      "loss": 0.0002,
+      "step": 19778
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6360013444087002e-06,
+      "loss": 0.0004,
+      "step": 19779
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6336006145868346e-06,
+      "loss": 0.0003,
+      "step": 19780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.631199884764969e-06,
+      "loss": 0.0003,
+      "step": 19781
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.628799154943103e-06,
+      "loss": 0.0003,
+      "step": 19782
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6263984251212367e-06,
+      "loss": 0.0003,
+      "step": 19783
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.623997695299371e-06,
+      "loss": 0.0006,
+      "step": 19784
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6215969654775054e-06,
+      "loss": 0.0002,
+      "step": 19785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6191962356556398e-06,
+      "loss": 0.0004,
+      "step": 19786
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6167955058337737e-06,
+      "loss": 0.0004,
+      "step": 19787
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6143947760119076e-06,
+      "loss": 0.0004,
+      "step": 19788
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.611994046190042e-06,
+      "loss": 0.0004,
+      "step": 19789
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6095933163681763e-06,
+      "loss": 0.0003,
+      "step": 19790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.60719258654631e-06,
+      "loss": 0.0003,
+      "step": 19791
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6047918567244445e-06,
+      "loss": 0.0003,
+      "step": 19792
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6023911269025785e-06,
+      "loss": 0.0002,
+      "step": 19793
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.599990397080713e-06,
+      "loss": 0.0004,
+      "step": 19794
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5975896672588467e-06,
+      "loss": 0.0003,
+      "step": 19795
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.595188937436981e-06,
+      "loss": 0.0003,
+      "step": 19796
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.592788207615115e-06,
+      "loss": 0.0003,
+      "step": 19797
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5903874777932493e-06,
+      "loss": 0.0002,
+      "step": 19798
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5879867479713832e-06,
+      "loss": 0.0037,
+      "step": 19799
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5855860181495176e-06,
+      "loss": 0.0009,
+      "step": 19800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.583185288327652e-06,
+      "loss": 0.0003,
+      "step": 19801
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.580784558505786e-06,
+      "loss": 0.0003,
+      "step": 19802
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5783838286839198e-06,
+      "loss": 0.0003,
+      "step": 19803
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.575983098862054e-06,
+      "loss": 0.0004,
+      "step": 19804
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5735823690401884e-06,
+      "loss": 0.0004,
+      "step": 19805
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5711816392183228e-06,
+      "loss": 0.0003,
+      "step": 19806
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5687809093964567e-06,
+      "loss": 0.0004,
+      "step": 19807
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5663801795745906e-06,
+      "loss": 0.0006,
+      "step": 19808
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.563979449752725e-06,
+      "loss": 0.0005,
+      "step": 19809
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5615787199308593e-06,
+      "loss": 0.0003,
+      "step": 19810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5591779901089932e-06,
+      "loss": 0.0003,
+      "step": 19811
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5567772602871276e-06,
+      "loss": 0.0003,
+      "step": 19812
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5543765304652615e-06,
+      "loss": 0.0003,
+      "step": 19813
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.551975800643396e-06,
+      "loss": 0.0002,
+      "step": 19814
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5495750708215297e-06,
+      "loss": 0.0003,
+      "step": 19815
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.547174340999664e-06,
+      "loss": 0.0004,
+      "step": 19816
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5447736111777984e-06,
+      "loss": 0.0003,
+      "step": 19817
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5423728813559323e-06,
+      "loss": 0.0003,
+      "step": 19818
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5399721515340663e-06,
+      "loss": 0.0002,
+      "step": 19819
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5375714217122006e-06,
+      "loss": 0.0002,
+      "step": 19820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.535170691890335e-06,
+      "loss": 0.0003,
+      "step": 19821
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5327699620684693e-06,
+      "loss": 0.0004,
+      "step": 19822
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.530369232246603e-06,
+      "loss": 0.0007,
+      "step": 19823
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.527968502424737e-06,
+      "loss": 0.0003,
+      "step": 19824
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5255677726028715e-06,
+      "loss": 0.0002,
+      "step": 19825
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.523167042781006e-06,
+      "loss": 0.0004,
+      "step": 19826
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5207663129591397e-06,
+      "loss": 0.0003,
+      "step": 19827
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5183655831372736e-06,
+      "loss": 0.0003,
+      "step": 19828
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.515964853315408e-06,
+      "loss": 0.0003,
+      "step": 19829
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5135641234935423e-06,
+      "loss": 0.0003,
+      "step": 19830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5111633936716762e-06,
+      "loss": 0.0004,
+      "step": 19831
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5087626638498106e-06,
+      "loss": 0.0005,
+      "step": 19832
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5063619340279445e-06,
+      "loss": 0.0002,
+      "step": 19833
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.503961204206079e-06,
+      "loss": 0.0003,
+      "step": 19834
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5015604743842128e-06,
+      "loss": 0.0004,
+      "step": 19835
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.499159744562347e-06,
+      "loss": 0.0003,
+      "step": 19836
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4967590147404814e-06,
+      "loss": 0.0002,
+      "step": 19837
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4943582849186154e-06,
+      "loss": 0.0002,
+      "step": 19838
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4919575550967493e-06,
+      "loss": 0.0003,
+      "step": 19839
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4895568252748836e-06,
+      "loss": 0.0002,
+      "step": 19840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.487156095453018e-06,
+      "loss": 0.0004,
+      "step": 19841
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4847553656311523e-06,
+      "loss": 0.0002,
+      "step": 19842
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4823546358092862e-06,
+      "loss": 0.0003,
+      "step": 19843
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.47995390598742e-06,
+      "loss": 0.0003,
+      "step": 19844
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4775531761655545e-06,
+      "loss": 0.0004,
+      "step": 19845
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.475152446343689e-06,
+      "loss": 0.0003,
+      "step": 19846
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4727517165218227e-06,
+      "loss": 0.0003,
+      "step": 19847
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.470350986699957e-06,
+      "loss": 0.0003,
+      "step": 19848
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.467950256878091e-06,
+      "loss": 0.0003,
+      "step": 19849
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4655495270562253e-06,
+      "loss": 0.0003,
+      "step": 19850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4631487972343593e-06,
+      "loss": 0.0003,
+      "step": 19851
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4607480674124936e-06,
+      "loss": 0.0003,
+      "step": 19852
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.458347337590628e-06,
+      "loss": 0.0004,
+      "step": 19853
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.455946607768762e-06,
+      "loss": 0.0003,
+      "step": 19854
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4535458779468958e-06,
+      "loss": 0.0003,
+      "step": 19855
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.45114514812503e-06,
+      "loss": 0.0004,
+      "step": 19856
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4487444183031645e-06,
+      "loss": 0.0003,
+      "step": 19857
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4463436884812984e-06,
+      "loss": 0.0003,
+      "step": 19858
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4439429586594323e-06,
+      "loss": 0.0004,
+      "step": 19859
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4415422288375666e-06,
+      "loss": 0.0005,
+      "step": 19860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.439141499015701e-06,
+      "loss": 0.0003,
+      "step": 19861
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.436740769193835e-06,
+      "loss": 0.0003,
+      "step": 19862
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4343400393719692e-06,
+      "loss": 0.0002,
+      "step": 19863
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.431939309550103e-06,
+      "loss": 0.0004,
+      "step": 19864
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4295385797282375e-06,
+      "loss": 0.0002,
+      "step": 19865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4271378499063714e-06,
+      "loss": 0.0003,
+      "step": 19866
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4247371200845058e-06,
+      "loss": 0.0003,
+      "step": 19867
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.42233639026264e-06,
+      "loss": 0.0003,
+      "step": 19868
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.419935660440774e-06,
+      "loss": 0.0175,
+      "step": 19869
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.417534930618908e-06,
+      "loss": 0.0002,
+      "step": 19870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4151342007970423e-06,
+      "loss": 0.0002,
+      "step": 19871
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4127334709751766e-06,
+      "loss": 0.0004,
+      "step": 19872
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.410332741153311e-06,
+      "loss": 0.0004,
+      "step": 19873
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.407932011331445e-06,
+      "loss": 0.0003,
+      "step": 19874
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.405531281509579e-06,
+      "loss": 0.0002,
+      "step": 19875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.403130551687713e-06,
+      "loss": 0.0003,
+      "step": 19876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4007298218658475e-06,
+      "loss": 0.0003,
+      "step": 19877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3983290920439814e-06,
+      "loss": 0.0002,
+      "step": 19878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3959283622221157e-06,
+      "loss": 0.0003,
+      "step": 19879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3935276324002496e-06,
+      "loss": 0.0003,
+      "step": 19880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.391126902578384e-06,
+      "loss": 0.0003,
+      "step": 19881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.388726172756518e-06,
+      "loss": 0.0004,
+      "step": 19882
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3863254429346522e-06,
+      "loss": 0.0003,
+      "step": 19883
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3839247131127866e-06,
+      "loss": 0.0003,
+      "step": 19884
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3815239832909205e-06,
+      "loss": 0.0003,
+      "step": 19885
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3791232534690544e-06,
+      "loss": 0.0003,
+      "step": 19886
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3767225236471888e-06,
+      "loss": 0.0003,
+      "step": 19887
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.374321793825323e-06,
+      "loss": 0.0002,
+      "step": 19888
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3719210640034575e-06,
+      "loss": 0.0003,
+      "step": 19889
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.369520334181591e-06,
+      "loss": 0.0002,
+      "step": 19890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3671196043597253e-06,
+      "loss": 0.0003,
+      "step": 19891
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3647188745378596e-06,
+      "loss": 0.0002,
+      "step": 19892
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.362318144715994e-06,
+      "loss": 0.0003,
+      "step": 19893
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.359917414894128e-06,
+      "loss": 0.0003,
+      "step": 19894
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.357516685072262e-06,
+      "loss": 0.0003,
+      "step": 19895
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.355115955250396e-06,
+      "loss": 0.0004,
+      "step": 19896
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3527152254285305e-06,
+      "loss": 0.0003,
+      "step": 19897
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3503144956066644e-06,
+      "loss": 0.0003,
+      "step": 19898
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3479137657847987e-06,
+      "loss": 0.0003,
+      "step": 19899
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3455130359629327e-06,
+      "loss": 0.0005,
+      "step": 19900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.343112306141067e-06,
+      "loss": 0.0002,
+      "step": 19901
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.340711576319201e-06,
+      "loss": 0.0003,
+      "step": 19902
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3383108464973353e-06,
+      "loss": 0.0002,
+      "step": 19903
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3359101166754696e-06,
+      "loss": 0.0002,
+      "step": 19904
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3335093868536035e-06,
+      "loss": 0.0002,
+      "step": 19905
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3311086570317374e-06,
+      "loss": 0.0003,
+      "step": 19906
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3287079272098718e-06,
+      "loss": 0.0002,
+      "step": 19907
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.326307197388006e-06,
+      "loss": 0.0002,
+      "step": 19908
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3239064675661405e-06,
+      "loss": 0.0003,
+      "step": 19909
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3215057377442744e-06,
+      "loss": 0.0005,
+      "step": 19910
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3191050079224083e-06,
+      "loss": 0.0002,
+      "step": 19911
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3167042781005426e-06,
+      "loss": 0.0003,
+      "step": 19912
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.314303548278677e-06,
+      "loss": 0.0004,
+      "step": 19913
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.311902818456811e-06,
+      "loss": 0.0002,
+      "step": 19914
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3095020886349452e-06,
+      "loss": 0.0353,
+      "step": 19915
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.307101358813079e-06,
+      "loss": 0.0003,
+      "step": 19916
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3047006289912135e-06,
+      "loss": 0.0003,
+      "step": 19917
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3022998991693474e-06,
+      "loss": 0.0004,
+      "step": 19918
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2998991693474818e-06,
+      "loss": 0.0003,
+      "step": 19919
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.297498439525616e-06,
+      "loss": 0.0003,
+      "step": 19920
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.29509770970375e-06,
+      "loss": 0.0003,
+      "step": 19921
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.292696979881884e-06,
+      "loss": 0.0003,
+      "step": 19922
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2902962500600183e-06,
+      "loss": 0.0002,
+      "step": 19923
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2878955202381526e-06,
+      "loss": 0.0004,
+      "step": 19924
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.285494790416287e-06,
+      "loss": 0.0025,
+      "step": 19925
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2830940605944205e-06,
+      "loss": 0.0002,
+      "step": 19926
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.280693330772555e-06,
+      "loss": 0.0003,
+      "step": 19927
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.278292600950689e-06,
+      "loss": 0.0003,
+      "step": 19928
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2758918711288235e-06,
+      "loss": 0.0002,
+      "step": 19929
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2734911413069574e-06,
+      "loss": 0.0005,
+      "step": 19930
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2710904114850913e-06,
+      "loss": 0.0003,
+      "step": 19931
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2686896816632257e-06,
+      "loss": 0.9912,
+      "step": 19932
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.26628895184136e-06,
+      "loss": 0.0004,
+      "step": 19933
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.263888222019494e-06,
+      "loss": 0.0003,
+      "step": 19934
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2614874921976283e-06,
+      "loss": 0.0004,
+      "step": 19935
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.259086762375762e-06,
+      "loss": 0.0003,
+      "step": 19936
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2566860325538965e-06,
+      "loss": 0.0002,
+      "step": 19937
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2542853027320304e-06,
+      "loss": 0.0004,
+      "step": 19938
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2518845729101648e-06,
+      "loss": 0.0002,
+      "step": 19939
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.249483843088299e-06,
+      "loss": 0.0003,
+      "step": 19940
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.247083113266433e-06,
+      "loss": 0.0002,
+      "step": 19941
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.244682383444567e-06,
+      "loss": 0.0002,
+      "step": 19942
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2422816536227013e-06,
+      "loss": 0.0002,
+      "step": 19943
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2398809238008356e-06,
+      "loss": 0.0003,
+      "step": 19944
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.23748019397897e-06,
+      "loss": 0.0003,
+      "step": 19945
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.235079464157104e-06,
+      "loss": 0.0002,
+      "step": 19946
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.232678734335238e-06,
+      "loss": 0.0003,
+      "step": 19947
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.230278004513372e-06,
+      "loss": 0.0003,
+      "step": 19948
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2278772746915065e-06,
+      "loss": 0.0005,
+      "step": 19949
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2254765448696404e-06,
+      "loss": 0.0003,
+      "step": 19950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2230758150477748e-06,
+      "loss": 0.0002,
+      "step": 19951
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2206750852259087e-06,
+      "loss": 0.0003,
+      "step": 19952
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.218274355404043e-06,
+      "loss": 0.0003,
+      "step": 19953
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.215873625582177e-06,
+      "loss": 0.0002,
+      "step": 19954
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2134728957603113e-06,
+      "loss": 0.0003,
+      "step": 19955
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2110721659384456e-06,
+      "loss": 0.0003,
+      "step": 19956
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2086714361165795e-06,
+      "loss": 0.0003,
+      "step": 19957
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2062707062947135e-06,
+      "loss": 0.0004,
+      "step": 19958
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.203869976472848e-06,
+      "loss": 0.0004,
+      "step": 19959
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.201469246650982e-06,
+      "loss": 0.0002,
+      "step": 19960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1990685168291165e-06,
+      "loss": 0.0004,
+      "step": 19961
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.19666778700725e-06,
+      "loss": 0.0003,
+      "step": 19962
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1942670571853843e-06,
+      "loss": 0.0004,
+      "step": 19963
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1918663273635187e-06,
+      "loss": 0.0003,
+      "step": 19964
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.189465597541653e-06,
+      "loss": 0.0004,
+      "step": 19965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.187064867719787e-06,
+      "loss": 0.0003,
+      "step": 19966
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.184664137897921e-06,
+      "loss": 0.0002,
+      "step": 19967
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.182263408076055e-06,
+      "loss": 0.0002,
+      "step": 19968
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1798626782541895e-06,
+      "loss": 0.0003,
+      "step": 19969
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1774619484323234e-06,
+      "loss": 0.0002,
+      "step": 19970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1750612186104578e-06,
+      "loss": 0.0003,
+      "step": 19971
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1726604887885917e-06,
+      "loss": 0.0003,
+      "step": 19972
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.170259758966726e-06,
+      "loss": 0.0049,
+      "step": 19973
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.16785902914486e-06,
+      "loss": 0.0002,
+      "step": 19974
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1654582993229943e-06,
+      "loss": 0.0002,
+      "step": 19975
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1630575695011286e-06,
+      "loss": 0.0003,
+      "step": 19976
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1606568396792626e-06,
+      "loss": 0.0001,
+      "step": 19977
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1582561098573965e-06,
+      "loss": 0.0003,
+      "step": 19978
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.155855380035531e-06,
+      "loss": 0.0004,
+      "step": 19979
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.153454650213665e-06,
+      "loss": 1.0354,
+      "step": 19980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1510539203917995e-06,
+      "loss": 0.0002,
+      "step": 19981
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1486531905699334e-06,
+      "loss": 0.0003,
+      "step": 19982
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1462524607480673e-06,
+      "loss": 0.0004,
+      "step": 19983
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1438517309262017e-06,
+      "loss": 0.0002,
+      "step": 19984
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.141451001104336e-06,
+      "loss": 0.0006,
+      "step": 19985
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.13905027128247e-06,
+      "loss": 0.0004,
+      "step": 19986
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1366495414606043e-06,
+      "loss": 0.0002,
+      "step": 19987
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.134248811638738e-06,
+      "loss": 0.0003,
+      "step": 19988
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1318480818168725e-06,
+      "loss": 0.0003,
+      "step": 19989
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1294473519950065e-06,
+      "loss": 0.0003,
+      "step": 19990
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.127046622173141e-06,
+      "loss": 0.0002,
+      "step": 19991
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.124645892351275e-06,
+      "loss": 0.0003,
+      "step": 19992
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.122245162529409e-06,
+      "loss": 0.0002,
+      "step": 19993
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.119844432707543e-06,
+      "loss": 0.0003,
+      "step": 19994
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1174437028856773e-06,
+      "loss": 0.0002,
+      "step": 19995
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1150429730638117e-06,
+      "loss": 0.0003,
+      "step": 19996
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1126422432419456e-06,
+      "loss": 0.0003,
+      "step": 19997
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1102415134200795e-06,
+      "loss": 0.0003,
+      "step": 19998
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.107840783598214e-06,
+      "loss": 0.0002,
+      "step": 19999
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.105440053776348e-06,
+      "loss": 0.0004,
+      "step": 20000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1030393239544825e-06,
+      "loss": 0.0003,
+      "step": 20001
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1006385941326164e-06,
+      "loss": 0.0003,
+      "step": 20002
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0982378643107503e-06,
+      "loss": 0.0002,
+      "step": 20003
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0958371344888847e-06,
+      "loss": 0.0004,
+      "step": 20004
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.093436404667019e-06,
+      "loss": 0.0003,
+      "step": 20005
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.091035674845153e-06,
+      "loss": 0.0003,
+      "step": 20006
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0886349450232873e-06,
+      "loss": 0.0003,
+      "step": 20007
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.086234215201421e-06,
+      "loss": 0.0003,
+      "step": 20008
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0838334853795556e-06,
+      "loss": 0.0004,
+      "step": 20009
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0814327555576895e-06,
+      "loss": 0.0002,
+      "step": 20010
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.079032025735824e-06,
+      "loss": 0.0004,
+      "step": 20011
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.076631295913958e-06,
+      "loss": 0.0002,
+      "step": 20012
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.074230566092092e-06,
+      "loss": 0.0003,
+      "step": 20013
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.071829836270226e-06,
+      "loss": 0.0004,
+      "step": 20014
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0694291064483603e-06,
+      "loss": 0.0002,
+      "step": 20015
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0670283766264947e-06,
+      "loss": 0.0003,
+      "step": 20016
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.064627646804629e-06,
+      "loss": 0.0003,
+      "step": 20017
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.062226916982763e-06,
+      "loss": 0.0002,
+      "step": 20018
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.059826187160897e-06,
+      "loss": 0.0003,
+      "step": 20019
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.057425457339031e-06,
+      "loss": 0.0003,
+      "step": 20020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0550247275171655e-06,
+      "loss": 0.0003,
+      "step": 20021
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0526239976952994e-06,
+      "loss": 0.0002,
+      "step": 20022
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.050223267873434e-06,
+      "loss": 0.0003,
+      "step": 20023
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0478225380515677e-06,
+      "loss": 0.0004,
+      "step": 20024
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.045421808229702e-06,
+      "loss": 0.0002,
+      "step": 20025
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.043021078407836e-06,
+      "loss": 0.0003,
+      "step": 20026
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0406203485859703e-06,
+      "loss": 0.0003,
+      "step": 20027
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0382196187641042e-06,
+      "loss": 0.0002,
+      "step": 20028
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0358188889422386e-06,
+      "loss": 0.0003,
+      "step": 20029
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0334181591203725e-06,
+      "loss": 0.0002,
+      "step": 20030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.031017429298507e-06,
+      "loss": 0.0002,
+      "step": 20031
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.028616699476641e-06,
+      "loss": 0.0008,
+      "step": 20032
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.026215969654775e-06,
+      "loss": 0.0002,
+      "step": 20033
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.023815239832909e-06,
+      "loss": 0.0002,
+      "step": 20034
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0214145100110433e-06,
+      "loss": 0.0003,
+      "step": 20035
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0190137801891777e-06,
+      "loss": 0.0002,
+      "step": 20036
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.016613050367312e-06,
+      "loss": 0.0003,
+      "step": 20037
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.014212320545446e-06,
+      "loss": 0.0003,
+      "step": 20038
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.01181159072358e-06,
+      "loss": 0.0003,
+      "step": 20039
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.009410860901714e-06,
+      "loss": 0.0002,
+      "step": 20040
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0070101310798485e-06,
+      "loss": 0.0004,
+      "step": 20041
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0046094012579825e-06,
+      "loss": 0.0003,
+      "step": 20042
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.002208671436117e-06,
+      "loss": 0.0003,
+      "step": 20043
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9998079416142507e-06,
+      "loss": 0.0003,
+      "step": 20044
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.997407211792385e-06,
+      "loss": 0.0004,
+      "step": 20045
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.995006481970519e-06,
+      "loss": 0.0002,
+      "step": 20046
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9926057521486533e-06,
+      "loss": 0.0003,
+      "step": 20047
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9902050223267877e-06,
+      "loss": 0.0005,
+      "step": 20048
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9878042925049216e-06,
+      "loss": 0.0003,
+      "step": 20049
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9854035626830555e-06,
+      "loss": 0.0003,
+      "step": 20050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.98300283286119e-06,
+      "loss": 0.0004,
+      "step": 20051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.980602103039324e-06,
+      "loss": 0.0004,
+      "step": 20052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9782013732174585e-06,
+      "loss": 0.0002,
+      "step": 20053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.975800643395592e-06,
+      "loss": 0.0004,
+      "step": 20054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9733999135737264e-06,
+      "loss": 0.0003,
+      "step": 20055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9709991837518607e-06,
+      "loss": 0.0002,
+      "step": 20056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.968598453929995e-06,
+      "loss": 0.0003,
+      "step": 20057
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.966197724108129e-06,
+      "loss": 0.0003,
+      "step": 20058
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.963796994286263e-06,
+      "loss": 0.0004,
+      "step": 20059
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9613962644643972e-06,
+      "loss": 0.0003,
+      "step": 20060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9589955346425316e-06,
+      "loss": 0.0004,
+      "step": 20061
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9565948048206655e-06,
+      "loss": 0.0005,
+      "step": 20062
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9541940749988e-06,
+      "loss": 0.0003,
+      "step": 20063
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9517933451769337e-06,
+      "loss": 1.1464,
+      "step": 20064
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.949392615355068e-06,
+      "loss": 0.0004,
+      "step": 20065
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.946991885533202e-06,
+      "loss": 0.0003,
+      "step": 20066
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9445911557113363e-06,
+      "loss": 0.0002,
+      "step": 20067
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9421904258894707e-06,
+      "loss": 0.0003,
+      "step": 20068
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9397896960676046e-06,
+      "loss": 0.0004,
+      "step": 20069
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9373889662457385e-06,
+      "loss": 0.0003,
+      "step": 20070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.934988236423873e-06,
+      "loss": 0.0004,
+      "step": 20071
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.932587506602007e-06,
+      "loss": 0.0002,
+      "step": 20072
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9301867767801415e-06,
+      "loss": 0.0002,
+      "step": 20073
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9277860469582755e-06,
+      "loss": 0.0003,
+      "step": 20074
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9253853171364094e-06,
+      "loss": 0.0002,
+      "step": 20075
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9229845873145437e-06,
+      "loss": 0.0004,
+      "step": 20076
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.920583857492678e-06,
+      "loss": 0.0004,
+      "step": 20077
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.918183127670812e-06,
+      "loss": 0.0003,
+      "step": 20078
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9157823978489463e-06,
+      "loss": 0.0005,
+      "step": 20079
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9133816680270802e-06,
+      "loss": 0.0004,
+      "step": 20080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9109809382052146e-06,
+      "loss": 0.0003,
+      "step": 20081
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9085802083833485e-06,
+      "loss": 0.0002,
+      "step": 20082
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9061794785614828e-06,
+      "loss": 0.0003,
+      "step": 20083
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.903778748739617e-06,
+      "loss": 0.0003,
+      "step": 20084
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9013780189177513e-06,
+      "loss": 0.0003,
+      "step": 20085
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8989772890958852e-06,
+      "loss": 0.0003,
+      "step": 20086
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8965765592740194e-06,
+      "loss": 0.0003,
+      "step": 20087
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8941758294521537e-06,
+      "loss": 0.0003,
+      "step": 20088
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8917750996302878e-06,
+      "loss": 0.0002,
+      "step": 20089
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8893743698084217e-06,
+      "loss": 0.0003,
+      "step": 20090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8869736399865559e-06,
+      "loss": 0.0002,
+      "step": 20091
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8845729101646902e-06,
+      "loss": 0.0003,
+      "step": 20092
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8821721803428246e-06,
+      "loss": 0.0004,
+      "step": 20093
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8797714505209583e-06,
+      "loss": 0.0002,
+      "step": 20094
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8773707206990926e-06,
+      "loss": 0.0003,
+      "step": 20095
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8749699908772267e-06,
+      "loss": 0.0004,
+      "step": 20096
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.872569261055361e-06,
+      "loss": 0.0005,
+      "step": 20097
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.870168531233495e-06,
+      "loss": 0.0003,
+      "step": 20098
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8677678014116291e-06,
+      "loss": 0.0003,
+      "step": 20099
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8653670715897635e-06,
+      "loss": 0.0003,
+      "step": 20100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8629663417678976e-06,
+      "loss": 0.0004,
+      "step": 20101
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8605656119460315e-06,
+      "loss": 0.0002,
+      "step": 20102
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8581648821241659e-06,
+      "loss": 0.0003,
+      "step": 20103
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8557641523023e-06,
+      "loss": 0.0003,
+      "step": 20104
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8533634224804343e-06,
+      "loss": 0.0002,
+      "step": 20105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8509626926585682e-06,
+      "loss": 0.0002,
+      "step": 20106
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8485619628367024e-06,
+      "loss": 0.0003,
+      "step": 20107
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8461612330148367e-06,
+      "loss": 0.0003,
+      "step": 20108
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8437605031929708e-06,
+      "loss": 0.0003,
+      "step": 20109
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8413597733711048e-06,
+      "loss": 0.0003,
+      "step": 20110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.838959043549239e-06,
+      "loss": 0.0003,
+      "step": 20111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8365583137273732e-06,
+      "loss": 0.0003,
+      "step": 20112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8341575839055076e-06,
+      "loss": 0.0003,
+      "step": 20113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8317568540836415e-06,
+      "loss": 0.0002,
+      "step": 20114
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8293561242617756e-06,
+      "loss": 0.0003,
+      "step": 20115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.82695539443991e-06,
+      "loss": 0.0003,
+      "step": 20116
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.824554664618044e-06,
+      "loss": 0.0002,
+      "step": 20117
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.822153934796178e-06,
+      "loss": 0.0002,
+      "step": 20118
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8197532049743124e-06,
+      "loss": 0.0002,
+      "step": 20119
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8173524751524465e-06,
+      "loss": 0.0004,
+      "step": 20120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8149517453305808e-06,
+      "loss": 0.0003,
+      "step": 20121
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8125510155087145e-06,
+      "loss": 0.0003,
+      "step": 20122
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8101502856868489e-06,
+      "loss": 0.0004,
+      "step": 20123
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8077495558649832e-06,
+      "loss": 0.0003,
+      "step": 20124
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8053488260431173e-06,
+      "loss": 0.0003,
+      "step": 20125
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8029480962212513e-06,
+      "loss": 0.0004,
+      "step": 20126
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8005473663993854e-06,
+      "loss": 0.0002,
+      "step": 20127
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7981466365775197e-06,
+      "loss": 0.0004,
+      "step": 20128
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7957459067556539e-06,
+      "loss": 0.0003,
+      "step": 20129
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7933451769337878e-06,
+      "loss": 0.0003,
+      "step": 20130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7909444471119221e-06,
+      "loss": 0.0004,
+      "step": 20131
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7885437172900563e-06,
+      "loss": 0.0003,
+      "step": 20132
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7861429874681906e-06,
+      "loss": 0.0004,
+      "step": 20133
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7837422576463245e-06,
+      "loss": 0.0003,
+      "step": 20134
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7813415278244586e-06,
+      "loss": 0.0002,
+      "step": 20135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.778940798002593e-06,
+      "loss": 0.0003,
+      "step": 20136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7765400681807271e-06,
+      "loss": 0.0003,
+      "step": 20137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.774139338358861e-06,
+      "loss": 0.0003,
+      "step": 20138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7717386085369954e-06,
+      "loss": 0.0002,
+      "step": 20139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7693378787151295e-06,
+      "loss": 0.0003,
+      "step": 20140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7669371488932638e-06,
+      "loss": 0.0004,
+      "step": 20141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7645364190713978e-06,
+      "loss": 0.0003,
+      "step": 20142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7621356892495319e-06,
+      "loss": 0.0004,
+      "step": 20143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7597349594276662e-06,
+      "loss": 0.0002,
+      "step": 20144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7573342296058004e-06,
+      "loss": 0.0002,
+      "step": 20145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7549334997839343e-06,
+      "loss": 0.0003,
+      "step": 20146
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7525327699620686e-06,
+      "loss": 0.0002,
+      "step": 20147
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7501320401402028e-06,
+      "loss": 0.0002,
+      "step": 20148
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.747731310318337e-06,
+      "loss": 0.0002,
+      "step": 20149
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.745330580496471e-06,
+      "loss": 0.507,
+      "step": 20150
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7429298506746051e-06,
+      "loss": 0.0003,
+      "step": 20151
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7405291208527395e-06,
+      "loss": 0.0006,
+      "step": 20152
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7381283910308736e-06,
+      "loss": 0.0002,
+      "step": 20153
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7357276612090075e-06,
+      "loss": 0.0003,
+      "step": 20154
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7333269313871419e-06,
+      "loss": 0.0002,
+      "step": 20155
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.730926201565276e-06,
+      "loss": 0.0003,
+      "step": 20156
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7285254717434103e-06,
+      "loss": 0.0003,
+      "step": 20157
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.726124741921544e-06,
+      "loss": 0.0002,
+      "step": 20158
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7237240120996784e-06,
+      "loss": 0.0003,
+      "step": 20159
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7213232822778125e-06,
+      "loss": 0.0002,
+      "step": 20160
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7189225524559469e-06,
+      "loss": 0.0002,
+      "step": 20161
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7165218226340808e-06,
+      "loss": 0.0003,
+      "step": 20162
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.714121092812215e-06,
+      "loss": 0.0003,
+      "step": 20163
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7117203629903492e-06,
+      "loss": 0.0004,
+      "step": 20164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7093196331684834e-06,
+      "loss": 0.0003,
+      "step": 20165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7069189033466173e-06,
+      "loss": 0.1547,
+      "step": 20166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7045181735247516e-06,
+      "loss": 0.0004,
+      "step": 20167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7021174437028858e-06,
+      "loss": 0.0003,
+      "step": 20168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6997167138810201e-06,
+      "loss": 0.0002,
+      "step": 20169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.697315984059154e-06,
+      "loss": 0.0004,
+      "step": 20170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6949152542372882e-06,
+      "loss": 0.0003,
+      "step": 20171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6925145244154225e-06,
+      "loss": 0.0003,
+      "step": 20172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6901137945935566e-06,
+      "loss": 0.0003,
+      "step": 20173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6877130647716905e-06,
+      "loss": 0.0004,
+      "step": 20174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6853123349498249e-06,
+      "loss": 0.0002,
+      "step": 20175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.682911605127959e-06,
+      "loss": 0.0005,
+      "step": 20176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6805108753060934e-06,
+      "loss": 0.0003,
+      "step": 20177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6781101454842273e-06,
+      "loss": 0.0003,
+      "step": 20178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6757094156623614e-06,
+      "loss": 0.0002,
+      "step": 20179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6733086858404957e-06,
+      "loss": 0.0003,
+      "step": 20180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6709079560186299e-06,
+      "loss": 0.0004,
+      "step": 20181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6685072261967638e-06,
+      "loss": 0.0004,
+      "step": 20182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6661064963748981e-06,
+      "loss": 0.0003,
+      "step": 20183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6637057665530323e-06,
+      "loss": 0.0004,
+      "step": 20184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6613050367311666e-06,
+      "loss": 0.0003,
+      "step": 20185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6589043069093003e-06,
+      "loss": 0.0003,
+      "step": 20186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6565035770874347e-06,
+      "loss": 0.0003,
+      "step": 20187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.654102847265569e-06,
+      "loss": 0.0002,
+      "step": 20188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6517021174437031e-06,
+      "loss": 0.0002,
+      "step": 20189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.649301387621837e-06,
+      "loss": 0.0003,
+      "step": 20190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6469006577999712e-06,
+      "loss": 0.0003,
+      "step": 20191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6444999279781055e-06,
+      "loss": 0.0004,
+      "step": 20192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6420991981562396e-06,
+      "loss": 0.0004,
+      "step": 20193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6396984683343736e-06,
+      "loss": 0.0003,
+      "step": 20194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.637297738512508e-06,
+      "loss": 0.0003,
+      "step": 20195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.634897008690642e-06,
+      "loss": 0.0003,
+      "step": 20196
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.632496278868776e-06,
+      "loss": 0.0003,
+      "step": 20197
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6300955490469103e-06,
+      "loss": 0.0003,
+      "step": 20198
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6276948192250444e-06,
+      "loss": 0.0002,
+      "step": 20199
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6252940894031788e-06,
+      "loss": 0.0004,
+      "step": 20200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6228933595813127e-06,
+      "loss": 0.0002,
+      "step": 20201
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6204926297594468e-06,
+      "loss": 0.0002,
+      "step": 20202
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6180918999375812e-06,
+      "loss": 0.0002,
+      "step": 20203
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6156911701157153e-06,
+      "loss": 0.0002,
+      "step": 20204
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6132904402938492e-06,
+      "loss": 0.0002,
+      "step": 20205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6108897104719835e-06,
+      "loss": 0.0003,
+      "step": 20206
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6084889806501177e-06,
+      "loss": 0.0002,
+      "step": 20207
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.606088250828252e-06,
+      "loss": 0.0002,
+      "step": 20208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.603687521006386e-06,
+      "loss": 0.0003,
+      "step": 20209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.60128679118452e-06,
+      "loss": 0.0003,
+      "step": 20210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5988860613626544e-06,
+      "loss": 0.0003,
+      "step": 20211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5964853315407885e-06,
+      "loss": 0.0002,
+      "step": 20212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5940846017189224e-06,
+      "loss": 0.0003,
+      "step": 20213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5916838718970568e-06,
+      "loss": 0.0002,
+      "step": 20214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.589283142075191e-06,
+      "loss": 0.0414,
+      "step": 20215
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5868824122533253e-06,
+      "loss": 0.0006,
+      "step": 20216
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.584481682431459e-06,
+      "loss": 0.0003,
+      "step": 20217
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5820809526095933e-06,
+      "loss": 0.0003,
+      "step": 20218
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5796802227877276e-06,
+      "loss": 0.0002,
+      "step": 20219
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5772794929658618e-06,
+      "loss": 0.0002,
+      "step": 20220
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5748787631439957e-06,
+      "loss": 0.0002,
+      "step": 20221
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5724780333221298e-06,
+      "loss": 0.0003,
+      "step": 20222
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5700773035002642e-06,
+      "loss": 0.0003,
+      "step": 20223
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5676765736783983e-06,
+      "loss": 0.0003,
+      "step": 20224
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5652758438565322e-06,
+      "loss": 0.0002,
+      "step": 20225
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5628751140346666e-06,
+      "loss": 0.0004,
+      "step": 20226
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5604743842128007e-06,
+      "loss": 0.0003,
+      "step": 20227
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5580736543909348e-06,
+      "loss": 0.0002,
+      "step": 20228
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5556729245690692e-06,
+      "loss": 0.0002,
+      "step": 20229
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.553272194747203e-06,
+      "loss": 0.8743,
+      "step": 20230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5508714649253374e-06,
+      "loss": 0.0003,
+      "step": 20231
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5484707351034715e-06,
+      "loss": 0.0003,
+      "step": 20232
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5460700052816057e-06,
+      "loss": 0.0003,
+      "step": 20233
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5436692754597398e-06,
+      "loss": 0.0002,
+      "step": 20234
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.541268545637874e-06,
+      "loss": 0.0003,
+      "step": 20235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.538867815816008e-06,
+      "loss": 0.0003,
+      "step": 20236
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5364670859941424e-06,
+      "loss": 0.0004,
+      "step": 20237
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5340663561722763e-06,
+      "loss": 0.0002,
+      "step": 20238
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5316656263504107e-06,
+      "loss": 0.0003,
+      "step": 20239
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5292648965285446e-06,
+      "loss": 0.0002,
+      "step": 20240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.526864166706679e-06,
+      "loss": 0.0004,
+      "step": 20241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.524463436884813e-06,
+      "loss": 0.0003,
+      "step": 20242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5220627070629472e-06,
+      "loss": 0.0002,
+      "step": 20243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5196619772410813e-06,
+      "loss": 0.0005,
+      "step": 20244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5172612474192154e-06,
+      "loss": 0.0003,
+      "step": 20245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5148605175973496e-06,
+      "loss": 0.0003,
+      "step": 20246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.512459787775484e-06,
+      "loss": 0.0004,
+      "step": 20247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5100590579536178e-06,
+      "loss": 0.0002,
+      "step": 20248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5076583281317522e-06,
+      "loss": 0.0002,
+      "step": 20249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5052575983098863e-06,
+      "loss": 0.0003,
+      "step": 20250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5028568684880204e-06,
+      "loss": 0.0003,
+      "step": 20251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5004561386661546e-06,
+      "loss": 0.0004,
+      "step": 20252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4980554088442887e-06,
+      "loss": 0.0003,
+      "step": 20253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4956546790224228e-06,
+      "loss": 0.0003,
+      "step": 20254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.493253949200557e-06,
+      "loss": 0.0003,
+      "step": 20255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.490853219378691e-06,
+      "loss": 0.0002,
+      "step": 20256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4884524895568254e-06,
+      "loss": 0.0006,
+      "step": 20257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4860517597349593e-06,
+      "loss": 0.9779,
+      "step": 20258
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4836510299130937e-06,
+      "loss": 0.0003,
+      "step": 20259
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4812503000912278e-06,
+      "loss": 0.0002,
+      "step": 20260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.478849570269362e-06,
+      "loss": 0.0003,
+      "step": 20261
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.476448840447496e-06,
+      "loss": 0.0002,
+      "step": 20262
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4740481106256302e-06,
+      "loss": 0.0003,
+      "step": 20263
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4716473808037643e-06,
+      "loss": 0.0004,
+      "step": 20264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4692466509818987e-06,
+      "loss": 0.0003,
+      "step": 20265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4668459211600326e-06,
+      "loss": 0.0003,
+      "step": 20266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.464445191338167e-06,
+      "loss": 0.0002,
+      "step": 20267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.462044461516301e-06,
+      "loss": 0.0002,
+      "step": 20268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4596437316944352e-06,
+      "loss": 0.0003,
+      "step": 20269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4572430018725693e-06,
+      "loss": 0.0004,
+      "step": 20270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4548422720507035e-06,
+      "loss": 0.0004,
+      "step": 20271
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4524415422288376e-06,
+      "loss": 0.0003,
+      "step": 20272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4500408124069717e-06,
+      "loss": 0.0002,
+      "step": 20273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4476400825851058e-06,
+      "loss": 0.0003,
+      "step": 20274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4452393527632402e-06,
+      "loss": 0.0002,
+      "step": 20275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.442838622941374e-06,
+      "loss": 0.0003,
+      "step": 20276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4404378931195084e-06,
+      "loss": 0.0004,
+      "step": 20277
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4380371632976426e-06,
+      "loss": 0.0002,
+      "step": 20278
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4356364334757767e-06,
+      "loss": 0.0003,
+      "step": 20279
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4332357036539108e-06,
+      "loss": 0.0003,
+      "step": 20280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.430834973832045e-06,
+      "loss": 0.0251,
+      "step": 20281
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.428434244010179e-06,
+      "loss": 0.0003,
+      "step": 20282
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4260335141883134e-06,
+      "loss": 0.0003,
+      "step": 20283
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4236327843664473e-06,
+      "loss": 1.0004,
+      "step": 20284
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4212320545445817e-06,
+      "loss": 0.0003,
+      "step": 20285
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4188313247227156e-06,
+      "loss": 0.0002,
+      "step": 20286
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.41643059490085e-06,
+      "loss": 0.0003,
+      "step": 20287
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.414029865078984e-06,
+      "loss": 0.0002,
+      "step": 20288
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4116291352571182e-06,
+      "loss": 0.0004,
+      "step": 20289
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4092284054352523e-06,
+      "loss": 0.0002,
+      "step": 20290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4068276756133865e-06,
+      "loss": 0.0003,
+      "step": 20291
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4044269457915206e-06,
+      "loss": 0.0003,
+      "step": 20292
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.402026215969655e-06,
+      "loss": 0.0003,
+      "step": 20293
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3996254861477889e-06,
+      "loss": 0.0003,
+      "step": 20294
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3972247563259232e-06,
+      "loss": 0.425,
+      "step": 20295
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3948240265040573e-06,
+      "loss": 0.0004,
+      "step": 20296
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3924232966821915e-06,
+      "loss": 0.0003,
+      "step": 20297
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3900225668603256e-06,
+      "loss": 0.0003,
+      "step": 20298
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3876218370384597e-06,
+      "loss": 0.0003,
+      "step": 20299
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3852211072165938e-06,
+      "loss": 0.0003,
+      "step": 20300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3828203773947282e-06,
+      "loss": 0.0002,
+      "step": 20301
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.380419647572862e-06,
+      "loss": 0.0004,
+      "step": 20302
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3780189177509964e-06,
+      "loss": 0.0002,
+      "step": 20303
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3756181879291304e-06,
+      "loss": 0.3884,
+      "step": 20304
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3732174581072647e-06,
+      "loss": 0.0004,
+      "step": 20305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3708167282853988e-06,
+      "loss": 0.0002,
+      "step": 20306
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.368415998463533e-06,
+      "loss": 0.0002,
+      "step": 20307
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.366015268641667e-06,
+      "loss": 0.0004,
+      "step": 20308
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3636145388198012e-06,
+      "loss": 0.0004,
+      "step": 20309
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3612138089979354e-06,
+      "loss": 0.0004,
+      "step": 20310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3588130791760697e-06,
+      "loss": 0.0003,
+      "step": 20311
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3564123493542036e-06,
+      "loss": 0.0005,
+      "step": 20312
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.354011619532338e-06,
+      "loss": 0.0003,
+      "step": 20313
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.351610889710472e-06,
+      "loss": 0.0002,
+      "step": 20314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3492101598886062e-06,
+      "loss": 0.0003,
+      "step": 20315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3468094300667403e-06,
+      "loss": 0.0003,
+      "step": 20316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3444087002448745e-06,
+      "loss": 0.0003,
+      "step": 20317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3420079704230086e-06,
+      "loss": 0.0003,
+      "step": 20318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.339607240601143e-06,
+      "loss": 0.0004,
+      "step": 20319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3372065107792769e-06,
+      "loss": 0.0002,
+      "step": 20320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3348057809574112e-06,
+      "loss": 0.0003,
+      "step": 20321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3324050511355451e-06,
+      "loss": 0.0003,
+      "step": 20322
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3300043213136795e-06,
+      "loss": 0.0002,
+      "step": 20323
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3276035914918136e-06,
+      "loss": 0.0002,
+      "step": 20324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3252028616699477e-06,
+      "loss": 0.0002,
+      "step": 20325
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3228021318480819e-06,
+      "loss": 0.0003,
+      "step": 20326
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.320401402026216e-06,
+      "loss": 0.0002,
+      "step": 20327
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3180006722043501e-06,
+      "loss": 0.0003,
+      "step": 20328
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3155999423824845e-06,
+      "loss": 0.0004,
+      "step": 20329
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3131992125606184e-06,
+      "loss": 0.0004,
+      "step": 20330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3107984827387527e-06,
+      "loss": 0.0003,
+      "step": 20331
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3083977529168868e-06,
+      "loss": 0.0002,
+      "step": 20332
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.305997023095021e-06,
+      "loss": 0.0002,
+      "step": 20333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.303596293273155e-06,
+      "loss": 0.0003,
+      "step": 20334
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3011955634512892e-06,
+      "loss": 0.0003,
+      "step": 20335
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2987948336294234e-06,
+      "loss": 0.0004,
+      "step": 20336
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2963941038075575e-06,
+      "loss": 0.0004,
+      "step": 20337
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2939933739856916e-06,
+      "loss": 0.0003,
+      "step": 20338
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.291592644163826e-06,
+      "loss": 0.0005,
+      "step": 20339
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2891919143419599e-06,
+      "loss": 0.0003,
+      "step": 20340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2867911845200942e-06,
+      "loss": 0.0002,
+      "step": 20341
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2843904546982284e-06,
+      "loss": 0.0004,
+      "step": 20342
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2819897248763625e-06,
+      "loss": 0.0003,
+      "step": 20343
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2795889950544966e-06,
+      "loss": 0.0003,
+      "step": 20344
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2771882652326307e-06,
+      "loss": 0.0004,
+      "step": 20345
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2747875354107649e-06,
+      "loss": 0.0002,
+      "step": 20346
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2723868055888992e-06,
+      "loss": 0.0003,
+      "step": 20347
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2699860757670331e-06,
+      "loss": 0.0004,
+      "step": 20348
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2675853459451675e-06,
+      "loss": 0.0003,
+      "step": 20349
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2651846161233016e-06,
+      "loss": 0.0004,
+      "step": 20350
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2627838863014357e-06,
+      "loss": 0.0002,
+      "step": 20351
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2603831564795699e-06,
+      "loss": 0.0003,
+      "step": 20352
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.257982426657704e-06,
+      "loss": 0.0002,
+      "step": 20353
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2555816968358381e-06,
+      "loss": 0.0003,
+      "step": 20354
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2531809670139722e-06,
+      "loss": 0.0004,
+      "step": 20355
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2507802371921064e-06,
+      "loss": 0.0002,
+      "step": 20356
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2483795073702407e-06,
+      "loss": 0.0003,
+      "step": 20357
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2459787775483746e-06,
+      "loss": 0.0004,
+      "step": 20358
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.243578047726509e-06,
+      "loss": 0.0037,
+      "step": 20359
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2411773179046431e-06,
+      "loss": 0.0003,
+      "step": 20360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2387765880827772e-06,
+      "loss": 0.0002,
+      "step": 20361
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2363758582609114e-06,
+      "loss": 0.0003,
+      "step": 20362
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2339751284390455e-06,
+      "loss": 0.0005,
+      "step": 20363
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2315743986171796e-06,
+      "loss": 0.0003,
+      "step": 20364
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.229173668795314e-06,
+      "loss": 0.0004,
+      "step": 20365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2267729389734479e-06,
+      "loss": 0.0002,
+      "step": 20366
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2243722091515822e-06,
+      "loss": 0.0004,
+      "step": 20367
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2219714793297161e-06,
+      "loss": 0.0003,
+      "step": 20368
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2195707495078505e-06,
+      "loss": 0.0002,
+      "step": 20369
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2171700196859846e-06,
+      "loss": 0.0002,
+      "step": 20370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2147692898641187e-06,
+      "loss": 0.0002,
+      "step": 20371
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2123685600422529e-06,
+      "loss": 0.0003,
+      "step": 20372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.209967830220387e-06,
+      "loss": 0.0003,
+      "step": 20373
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2075671003985211e-06,
+      "loss": 0.0004,
+      "step": 20374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2051663705766555e-06,
+      "loss": 0.0005,
+      "step": 20375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2027656407547894e-06,
+      "loss": 0.0002,
+      "step": 20376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2003649109329237e-06,
+      "loss": 0.0003,
+      "step": 20377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1979641811110579e-06,
+      "loss": 0.0003,
+      "step": 20378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.195563451289192e-06,
+      "loss": 0.0002,
+      "step": 20379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1931627214673261e-06,
+      "loss": 0.0002,
+      "step": 20380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1907619916454603e-06,
+      "loss": 0.0004,
+      "step": 20381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1883612618235944e-06,
+      "loss": 0.0003,
+      "step": 20382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1859605320017287e-06,
+      "loss": 0.0003,
+      "step": 20383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1835598021798626e-06,
+      "loss": 0.0003,
+      "step": 20384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.181159072357997e-06,
+      "loss": 0.0002,
+      "step": 20385
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.178758342536131e-06,
+      "loss": 0.0003,
+      "step": 20386
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1763576127142652e-06,
+      "loss": 0.0003,
+      "step": 20387
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1739568828923994e-06,
+      "loss": 0.0003,
+      "step": 20388
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1715561530705335e-06,
+      "loss": 0.0002,
+      "step": 20389
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1691554232486676e-06,
+      "loss": 0.0002,
+      "step": 20390
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1667546934268018e-06,
+      "loss": 0.0005,
+      "step": 20391
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1643539636049359e-06,
+      "loss": 0.0001,
+      "step": 20392
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1619532337830702e-06,
+      "loss": 0.0002,
+      "step": 20393
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1595525039612042e-06,
+      "loss": 0.0003,
+      "step": 20394
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1571517741393385e-06,
+      "loss": 0.0003,
+      "step": 20395
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1547510443174726e-06,
+      "loss": 0.0002,
+      "step": 20396
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1523503144956068e-06,
+      "loss": 0.0002,
+      "step": 20397
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1499495846737409e-06,
+      "loss": 0.0003,
+      "step": 20398
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.147548854851875e-06,
+      "loss": 0.0002,
+      "step": 20399
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1451481250300091e-06,
+      "loss": 0.0003,
+      "step": 20400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1427473952081435e-06,
+      "loss": 0.0002,
+      "step": 20401
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1403466653862774e-06,
+      "loss": 0.0003,
+      "step": 20402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1379459355644117e-06,
+      "loss": 0.0002,
+      "step": 20403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1355452057425457e-06,
+      "loss": 0.0002,
+      "step": 20404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.13314447592068e-06,
+      "loss": 0.0003,
+      "step": 20405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1307437460988141e-06,
+      "loss": 0.0004,
+      "step": 20406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1283430162769483e-06,
+      "loss": 0.0003,
+      "step": 20407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1259422864550824e-06,
+      "loss": 0.0003,
+      "step": 20408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1235415566332165e-06,
+      "loss": 0.0003,
+      "step": 20409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1211408268113506e-06,
+      "loss": 0.0003,
+      "step": 20410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.118740096989485e-06,
+      "loss": 0.0004,
+      "step": 20411
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.116339367167619e-06,
+      "loss": 0.0003,
+      "step": 20412
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1139386373457533e-06,
+      "loss": 0.0002,
+      "step": 20413
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1115379075238874e-06,
+      "loss": 0.0004,
+      "step": 20414
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1091371777020215e-06,
+      "loss": 0.0001,
+      "step": 20415
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1067364478801556e-06,
+      "loss": 0.0002,
+      "step": 20416
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1043357180582898e-06,
+      "loss": 0.0003,
+      "step": 20417
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.101934988236424e-06,
+      "loss": 0.0005,
+      "step": 20418
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0995342584145582e-06,
+      "loss": 0.0003,
+      "step": 20419
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0971335285926922e-06,
+      "loss": 0.0002,
+      "step": 20420
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0947327987708265e-06,
+      "loss": 0.0002,
+      "step": 20421
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0923320689489604e-06,
+      "loss": 0.0004,
+      "step": 20422
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0899313391270948e-06,
+      "loss": 0.0003,
+      "step": 20423
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0875306093052289e-06,
+      "loss": 0.0002,
+      "step": 20424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.085129879483363e-06,
+      "loss": 0.0002,
+      "step": 20425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0827291496614971e-06,
+      "loss": 0.0003,
+      "step": 20426
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0803284198396313e-06,
+      "loss": 0.0003,
+      "step": 20427
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0779276900177654e-06,
+      "loss": 0.0003,
+      "step": 20428
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0755269601958997e-06,
+      "loss": 0.0003,
+      "step": 20429
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0731262303740337e-06,
+      "loss": 0.0002,
+      "step": 20430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.070725500552168e-06,
+      "loss": 0.0002,
+      "step": 20431
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0683247707303021e-06,
+      "loss": 0.0003,
+      "step": 20432
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0659240409084363e-06,
+      "loss": 0.0003,
+      "step": 20433
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0635233110865704e-06,
+      "loss": 0.0002,
+      "step": 20434
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0611225812647045e-06,
+      "loss": 0.0003,
+      "step": 20435
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0587218514428387e-06,
+      "loss": 0.0003,
+      "step": 20436
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0563211216209728e-06,
+      "loss": 0.0002,
+      "step": 20437
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.053920391799107e-06,
+      "loss": 0.0003,
+      "step": 20438
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0515196619772413e-06,
+      "loss": 0.0003,
+      "step": 20439
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0491189321553752e-06,
+      "loss": 0.0002,
+      "step": 20440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0467182023335095e-06,
+      "loss": 0.0002,
+      "step": 20441
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0443174725116436e-06,
+      "loss": 0.0003,
+      "step": 20442
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0419167426897778e-06,
+      "loss": 0.0002,
+      "step": 20443
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.039516012867912e-06,
+      "loss": 0.0003,
+      "step": 20444
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.037115283046046e-06,
+      "loss": 0.0003,
+      "step": 20445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0347145532241802e-06,
+      "loss": 0.0003,
+      "step": 20446
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0323138234023145e-06,
+      "loss": 0.0002,
+      "step": 20447
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0299130935804484e-06,
+      "loss": 0.0003,
+      "step": 20448
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0275123637585828e-06,
+      "loss": 0.0005,
+      "step": 20449
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.025111633936717e-06,
+      "loss": 0.0003,
+      "step": 20450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.022710904114851e-06,
+      "loss": 0.0002,
+      "step": 20451
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0203101742929852e-06,
+      "loss": 0.0002,
+      "step": 20452
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0179094444711193e-06,
+      "loss": 0.0004,
+      "step": 20453
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0155087146492534e-06,
+      "loss": 0.0003,
+      "step": 20454
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0131079848273875e-06,
+      "loss": 0.0002,
+      "step": 20455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0107072550055217e-06,
+      "loss": 0.0001,
+      "step": 20456
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.008306525183656e-06,
+      "loss": 0.0002,
+      "step": 20457
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.00590579536179e-06,
+      "loss": 0.0002,
+      "step": 20458
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0035050655399243e-06,
+      "loss": 0.0002,
+      "step": 20459
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0011043357180584e-06,
+      "loss": 0.0003,
+      "step": 20460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.987036058961925e-07,
+      "loss": 0.0002,
+      "step": 20461
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.963028760743267e-07,
+      "loss": 0.0002,
+      "step": 20462
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.939021462524608e-07,
+      "loss": 0.0003,
+      "step": 20463
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.91501416430595e-07,
+      "loss": 0.0003,
+      "step": 20464
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.891006866087293e-07,
+      "loss": 0.0003,
+      "step": 20465
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.866999567868632e-07,
+      "loss": 0.0003,
+      "step": 20466
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.842992269649975e-07,
+      "loss": 0.0003,
+      "step": 20467
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.818984971431314e-07,
+      "loss": 0.0004,
+      "step": 20468
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.794977673212658e-07,
+      "loss": 0.0002,
+      "step": 20469
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.770970374994e-07,
+      "loss": 0.0003,
+      "step": 20470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.74696307677534e-07,
+      "loss": 0.0003,
+      "step": 20471
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.722955778556682e-07,
+      "loss": 0.0002,
+      "step": 20472
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.698948480338023e-07,
+      "loss": 0.0002,
+      "step": 20473
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.674941182119364e-07,
+      "loss": 0.0003,
+      "step": 20474
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.650933883900708e-07,
+      "loss": 0.0003,
+      "step": 20475
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.626926585682047e-07,
+      "loss": 0.0002,
+      "step": 20476
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.60291928746339e-07,
+      "loss": 0.0003,
+      "step": 20477
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.578911989244732e-07,
+      "loss": 0.0004,
+      "step": 20478
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.554904691026073e-07,
+      "loss": 0.0002,
+      "step": 20479
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.530897392807414e-07,
+      "loss": 0.0003,
+      "step": 20480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.506890094588757e-07,
+      "loss": 0.0003,
+      "step": 20481
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.482882796370097e-07,
+      "loss": 0.0003,
+      "step": 20482
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.458875498151439e-07,
+      "loss": 0.0003,
+      "step": 20483
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.434868199932779e-07,
+      "loss": 0.0002,
+      "step": 20484
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.410860901714123e-07,
+      "loss": 0.0002,
+      "step": 20485
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.386853603495463e-07,
+      "loss": 0.0004,
+      "step": 20486
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.362846305276805e-07,
+      "loss": 0.0004,
+      "step": 20487
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.338839007058146e-07,
+      "loss": 0.0003,
+      "step": 20488
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.314831708839488e-07,
+      "loss": 0.0003,
+      "step": 20489
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.290824410620829e-07,
+      "loss": 0.0003,
+      "step": 20490
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.266817112402172e-07,
+      "loss": 0.0002,
+      "step": 20491
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.242809814183512e-07,
+      "loss": 0.0003,
+      "step": 20492
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.218802515964854e-07,
+      "loss": 0.0002,
+      "step": 20493
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.194795217746196e-07,
+      "loss": 0.0002,
+      "step": 20494
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.170787919527538e-07,
+      "loss": 0.0002,
+      "step": 20495
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.146780621308878e-07,
+      "loss": 0.0002,
+      "step": 20496
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.12277332309022e-07,
+      "loss": 0.0002,
+      "step": 20497
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.098766024871562e-07,
+      "loss": 0.0004,
+      "step": 20498
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.074758726652904e-07,
+      "loss": 0.0002,
+      "step": 20499
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.050751428434244e-07,
+      "loss": 0.0003,
+      "step": 20500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.026744130215587e-07,
+      "loss": 0.0002,
+      "step": 20501
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.002736831996927e-07,
+      "loss": 0.0002,
+      "step": 20502
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.978729533778269e-07,
+      "loss": 0.0004,
+      "step": 20503
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.954722235559611e-07,
+      "loss": 0.0003,
+      "step": 20504
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.930714937340953e-07,
+      "loss": 0.0002,
+      "step": 20505
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.906707639122293e-07,
+      "loss": 0.0002,
+      "step": 20506
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.882700340903636e-07,
+      "loss": 0.0003,
+      "step": 20507
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.858693042684977e-07,
+      "loss": 0.0004,
+      "step": 20508
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.834685744466319e-07,
+      "loss": 0.0003,
+      "step": 20509
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.810678446247659e-07,
+      "loss": 0.0002,
+      "step": 20510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.786671148029002e-07,
+      "loss": 0.0002,
+      "step": 20511
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.762663849810343e-07,
+      "loss": 0.0003,
+      "step": 20512
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.738656551591685e-07,
+      "loss": 0.0005,
+      "step": 20513
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.714649253373026e-07,
+      "loss": 0.0002,
+      "step": 20514
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.690641955154368e-07,
+      "loss": 0.0002,
+      "step": 20515
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.666634656935709e-07,
+      "loss": 0.0002,
+      "step": 20516
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.642627358717052e-07,
+      "loss": 0.0002,
+      "step": 20517
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.618620060498392e-07,
+      "loss": 0.0002,
+      "step": 20518
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.594612762279734e-07,
+      "loss": 0.0002,
+      "step": 20519
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.570605464061075e-07,
+      "loss": 0.0002,
+      "step": 20520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.546598165842417e-07,
+      "loss": 0.0002,
+      "step": 20521
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.522590867623758e-07,
+      "loss": 0.0002,
+      "step": 20522
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.498583569405101e-07,
+      "loss": 0.0002,
+      "step": 20523
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.474576271186441e-07,
+      "loss": 0.0003,
+      "step": 20524
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.450568972967783e-07,
+      "loss": 0.0004,
+      "step": 20525
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.426561674749124e-07,
+      "loss": 0.0002,
+      "step": 20526
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.402554376530467e-07,
+      "loss": 0.0002,
+      "step": 20527
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.378547078311807e-07,
+      "loss": 0.0002,
+      "step": 20528
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.354539780093149e-07,
+      "loss": 0.0003,
+      "step": 20529
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.330532481874491e-07,
+      "loss": 0.0002,
+      "step": 20530
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.306525183655833e-07,
+      "loss": 0.0003,
+      "step": 20531
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.282517885437173e-07,
+      "loss": 0.0003,
+      "step": 20532
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.258510587218516e-07,
+      "loss": 0.0003,
+      "step": 20533
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.234503288999856e-07,
+      "loss": 0.0002,
+      "step": 20534
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.210495990781198e-07,
+      "loss": 0.0002,
+      "step": 20535
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.18648869256254e-07,
+      "loss": 0.0004,
+      "step": 20536
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.16248139434388e-07,
+      "loss": 0.0003,
+      "step": 20537
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.138474096125222e-07,
+      "loss": 0.0004,
+      "step": 20538
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.114466797906563e-07,
+      "loss": 0.0003,
+      "step": 20539
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.090459499687906e-07,
+      "loss": 0.0002,
+      "step": 20540
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.066452201469246e-07,
+      "loss": 0.0003,
+      "step": 20541
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.042444903250588e-07,
+      "loss": 0.0003,
+      "step": 20542
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.01843760503193e-07,
+      "loss": 0.0004,
+      "step": 20543
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.994430306813272e-07,
+      "loss": 0.0003,
+      "step": 20544
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.970423008594612e-07,
+      "loss": 0.0002,
+      "step": 20545
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.946415710375955e-07,
+      "loss": 0.0003,
+      "step": 20546
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.922408412157295e-07,
+      "loss": 0.0003,
+      "step": 20547
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.898401113938638e-07,
+      "loss": 0.0003,
+      "step": 20548
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.874393815719978e-07,
+      "loss": 0.0004,
+      "step": 20549
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.850386517501321e-07,
+      "loss": 0.0005,
+      "step": 20550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.826379219282661e-07,
+      "loss": 0.0003,
+      "step": 20551
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.802371921064003e-07,
+      "loss": 0.0004,
+      "step": 20552
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.778364622845346e-07,
+      "loss": 0.0002,
+      "step": 20553
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.754357324626687e-07,
+      "loss": 0.0001,
+      "step": 20554
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.730350026408028e-07,
+      "loss": 0.0004,
+      "step": 20555
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.70634272818937e-07,
+      "loss": 0.0002,
+      "step": 20556
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.682335429970712e-07,
+      "loss": 0.0002,
+      "step": 20557
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.658328131752053e-07,
+      "loss": 0.0004,
+      "step": 20558
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.634320833533395e-07,
+      "loss": 0.0002,
+      "step": 20559
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.610313535314736e-07,
+      "loss": 0.0003,
+      "step": 20560
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586306237096077e-07,
+      "loss": 0.0002,
+      "step": 20561
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.56229893887742e-07,
+      "loss": 0.0002,
+      "step": 20562
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.538291640658761e-07,
+      "loss": 0.0003,
+      "step": 20563
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.514284342440102e-07,
+      "loss": 0.0006,
+      "step": 20564
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.490277044221443e-07,
+      "loss": 0.0002,
+      "step": 20565
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.466269746002785e-07,
+      "loss": 0.0003,
+      "step": 20566
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.442262447784127e-07,
+      "loss": 0.0003,
+      "step": 20567
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.418255149565468e-07,
+      "loss": 0.0003,
+      "step": 20568
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.39424785134681e-07,
+      "loss": 0.0003,
+      "step": 20569
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.370240553128151e-07,
+      "loss": 0.0002,
+      "step": 20570
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.346233254909493e-07,
+      "loss": 0.0002,
+      "step": 20571
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.322225956690835e-07,
+      "loss": 0.0003,
+      "step": 20572
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.298218658472176e-07,
+      "loss": 0.0002,
+      "step": 20573
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.274211360253517e-07,
+      "loss": 0.0003,
+      "step": 20574
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.250204062034859e-07,
+      "loss": 0.0003,
+      "step": 20575
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.226196763816201e-07,
+      "loss": 0.0002,
+      "step": 20576
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.202189465597542e-07,
+      "loss": 0.0002,
+      "step": 20577
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.178182167378884e-07,
+      "loss": 0.0003,
+      "step": 20578
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.154174869160225e-07,
+      "loss": 0.0002,
+      "step": 20579
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.130167570941567e-07,
+      "loss": 0.0003,
+      "step": 20580
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.106160272722908e-07,
+      "loss": 0.0003,
+      "step": 20581
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.08215297450425e-07,
+      "loss": 0.0002,
+      "step": 20582
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.058145676285591e-07,
+      "loss": 0.0001,
+      "step": 20583
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.034138378066932e-07,
+      "loss": 0.0003,
+      "step": 20584
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.010131079848275e-07,
+      "loss": 0.0002,
+      "step": 20585
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.986123781629616e-07,
+      "loss": 0.0003,
+      "step": 20586
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.962116483410957e-07,
+      "loss": 0.0004,
+      "step": 20587
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.938109185192299e-07,
+      "loss": 0.0002,
+      "step": 20588
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.914101886973641e-07,
+      "loss": 0.0003,
+      "step": 20589
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.890094588754982e-07,
+      "loss": 0.0003,
+      "step": 20590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.866087290536324e-07,
+      "loss": 0.0003,
+      "step": 20591
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.842079992317665e-07,
+      "loss": 0.0002,
+      "step": 20592
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.818072694099006e-07,
+      "loss": 0.0003,
+      "step": 20593
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.794065395880348e-07,
+      "loss": 0.0002,
+      "step": 20594
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.77005809766169e-07,
+      "loss": 0.0003,
+      "step": 20595
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.746050799443031e-07,
+      "loss": 0.0004,
+      "step": 20596
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.722043501224372e-07,
+      "loss": 0.0004,
+      "step": 20597
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.698036203005715e-07,
+      "loss": 0.0004,
+      "step": 20598
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.674028904787056e-07,
+      "loss": 0.0002,
+      "step": 20599
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.650021606568397e-07,
+      "loss": 0.0004,
+      "step": 20600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.626014308349739e-07,
+      "loss": 0.0004,
+      "step": 20601
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.60200701013108e-07,
+      "loss": 0.0003,
+      "step": 20602
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.577999711912422e-07,
+      "loss": 0.0002,
+      "step": 20603
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.553992413693764e-07,
+      "loss": 0.0002,
+      "step": 20604
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.529985115475105e-07,
+      "loss": 0.0002,
+      "step": 20605
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.505977817256446e-07,
+      "loss": 0.0008,
+      "step": 20606
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.481970519037787e-07,
+      "loss": 0.0006,
+      "step": 20607
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.45796322081913e-07,
+      "loss": 0.0002,
+      "step": 20608
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.433955922600471e-07,
+      "loss": 0.0003,
+      "step": 20609
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.409948624381812e-07,
+      "loss": 0.0004,
+      "step": 20610
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.385941326163154e-07,
+      "loss": 0.0003,
+      "step": 20611
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.361934027944496e-07,
+      "loss": 0.0003,
+      "step": 20612
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.337926729725837e-07,
+      "loss": 0.0002,
+      "step": 20613
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.313919431507179e-07,
+      "loss": 0.0003,
+      "step": 20614
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.28991213328852e-07,
+      "loss": 0.0003,
+      "step": 20615
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.265904835069861e-07,
+      "loss": 0.0003,
+      "step": 20616
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.241897536851204e-07,
+      "loss": 0.0002,
+      "step": 20617
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.217890238632545e-07,
+      "loss": 0.0003,
+      "step": 20618
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.193882940413886e-07,
+      "loss": 0.0003,
+      "step": 20619
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.169875642195227e-07,
+      "loss": 0.0002,
+      "step": 20620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.14586834397657e-07,
+      "loss": 0.0004,
+      "step": 20621
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.121861045757911e-07,
+      "loss": 0.0004,
+      "step": 20622
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.097853747539252e-07,
+      "loss": 0.0002,
+      "step": 20623
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.073846449320594e-07,
+      "loss": 0.0002,
+      "step": 20624
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.049839151101935e-07,
+      "loss": 0.0003,
+      "step": 20625
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.025831852883277e-07,
+      "loss": 0.0003,
+      "step": 20626
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.001824554664619e-07,
+      "loss": 0.0004,
+      "step": 20627
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.97781725644596e-07,
+      "loss": 0.0002,
+      "step": 20628
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.953809958227301e-07,
+      "loss": 0.0002,
+      "step": 20629
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.929802660008644e-07,
+      "loss": 0.0002,
+      "step": 20630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.905795361789985e-07,
+      "loss": 0.0003,
+      "step": 20631
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.881788063571326e-07,
+      "loss": 0.0005,
+      "step": 20632
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.857780765352668e-07,
+      "loss": 0.0004,
+      "step": 20633
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.833773467134009e-07,
+      "loss": 0.0003,
+      "step": 20634
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.809766168915351e-07,
+      "loss": 0.0002,
+      "step": 20635
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.785758870696692e-07,
+      "loss": 0.0004,
+      "step": 20636
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.761751572478034e-07,
+      "loss": 0.0002,
+      "step": 20637
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.737744274259375e-07,
+      "loss": 0.0002,
+      "step": 20638
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.713736976040717e-07,
+      "loss": 0.0003,
+      "step": 20639
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.689729677822059e-07,
+      "loss": 0.0003,
+      "step": 20640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6657223796034e-07,
+      "loss": 0.0056,
+      "step": 20641
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.641715081384741e-07,
+      "loss": 0.0002,
+      "step": 20642
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.617707783166083e-07,
+      "loss": 0.0003,
+      "step": 20643
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.593700484947425e-07,
+      "loss": 0.0003,
+      "step": 20644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.569693186728766e-07,
+      "loss": 0.0002,
+      "step": 20645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.545685888510108e-07,
+      "loss": 0.0003,
+      "step": 20646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.521678590291449e-07,
+      "loss": 0.0002,
+      "step": 20647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.497671292072791e-07,
+      "loss": 0.0002,
+      "step": 20648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.473663993854132e-07,
+      "loss": 0.0003,
+      "step": 20649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.449656695635474e-07,
+      "loss": 0.0003,
+      "step": 20650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.425649397416815e-07,
+      "loss": 0.0004,
+      "step": 20651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.401642099198156e-07,
+      "loss": 0.0003,
+      "step": 20652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.377634800979499e-07,
+      "loss": 0.0003,
+      "step": 20653
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.35362750276084e-07,
+      "loss": 0.0004,
+      "step": 20654
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.329620204542181e-07,
+      "loss": 0.0004,
+      "step": 20655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.305612906323523e-07,
+      "loss": 0.0002,
+      "step": 20656
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.281605608104864e-07,
+      "loss": 0.0002,
+      "step": 20657
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.257598309886206e-07,
+      "loss": 0.0003,
+      "step": 20658
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.233591011667548e-07,
+      "loss": 0.0002,
+      "step": 20659
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.209583713448889e-07,
+      "loss": 0.0003,
+      "step": 20660
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.18557641523023e-07,
+      "loss": 0.0003,
+      "step": 20661
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.161569117011573e-07,
+      "loss": 0.0003,
+      "step": 20662
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.137561818792914e-07,
+      "loss": 0.0002,
+      "step": 20663
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.113554520574255e-07,
+      "loss": 0.0003,
+      "step": 20664
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.089547222355596e-07,
+      "loss": 0.0003,
+      "step": 20665
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.065539924136938e-07,
+      "loss": 0.0003,
+      "step": 20666
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.04153262591828e-07,
+      "loss": 0.0042,
+      "step": 20667
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.017525327699621e-07,
+      "loss": 0.0002,
+      "step": 20668
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.993518029480963e-07,
+      "loss": 0.0003,
+      "step": 20669
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.969510731262304e-07,
+      "loss": 0.0003,
+      "step": 20670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.945503433043646e-07,
+      "loss": 0.0002,
+      "step": 20671
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.921496134824988e-07,
+      "loss": 0.0002,
+      "step": 20672
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.897488836606329e-07,
+      "loss": 0.0003,
+      "step": 20673
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.87348153838767e-07,
+      "loss": 0.0003,
+      "step": 20674
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.849474240169012e-07,
+      "loss": 0.0003,
+      "step": 20675
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.825466941950354e-07,
+      "loss": 0.0003,
+      "step": 20676
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.801459643731695e-07,
+      "loss": 0.0031,
+      "step": 20677
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.777452345513036e-07,
+      "loss": 0.0001,
+      "step": 20678
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.7534450472943783e-07,
+      "loss": 0.0001,
+      "step": 20679
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.7294377490757196e-07,
+      "loss": 0.0003,
+      "step": 20680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.7054304508570614e-07,
+      "loss": 0.0002,
+      "step": 20681
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6814231526384027e-07,
+      "loss": 0.0002,
+      "step": 20682
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.657415854419744e-07,
+      "loss": 0.0003,
+      "step": 20683
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.633408556201086e-07,
+      "loss": 0.0002,
+      "step": 20684
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.609401257982427e-07,
+      "loss": 0.0003,
+      "step": 20685
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.585393959763769e-07,
+      "loss": 0.0003,
+      "step": 20686
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.56138666154511e-07,
+      "loss": 0.0002,
+      "step": 20687
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.537379363326452e-07,
+      "loss": 0.0003,
+      "step": 20688
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.5133720651077934e-07,
+      "loss": 0.0002,
+      "step": 20689
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4893647668891347e-07,
+      "loss": 0.0003,
+      "step": 20690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4653574686704765e-07,
+      "loss": 0.0004,
+      "step": 20691
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.441350170451818e-07,
+      "loss": 0.0004,
+      "step": 20692
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4173428722331596e-07,
+      "loss": 0.0003,
+      "step": 20693
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.393335574014501e-07,
+      "loss": 0.0002,
+      "step": 20694
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3693282757958427e-07,
+      "loss": 0.0003,
+      "step": 20695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.345320977577184e-07,
+      "loss": 0.0002,
+      "step": 20696
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.321313679358526e-07,
+      "loss": 0.0003,
+      "step": 20697
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.297306381139867e-07,
+      "loss": 0.0004,
+      "step": 20698
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2732990829212084e-07,
+      "loss": 0.0004,
+      "step": 20699
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2492917847025503e-07,
+      "loss": 0.0003,
+      "step": 20700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2252844864838916e-07,
+      "loss": 0.0004,
+      "step": 20701
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2012771882652334e-07,
+      "loss": 0.0003,
+      "step": 20702
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1772698900465747e-07,
+      "loss": 0.0002,
+      "step": 20703
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.1532625918279165e-07,
+      "loss": 0.0001,
+      "step": 20704
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.129255293609258e-07,
+      "loss": 0.0004,
+      "step": 20705
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.105247995390599e-07,
+      "loss": 0.0004,
+      "step": 20706
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.08124069717194e-07,
+      "loss": 0.0002,
+      "step": 20707
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0572333989532817e-07,
+      "loss": 0.0003,
+      "step": 20708
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.033226100734623e-07,
+      "loss": 0.0002,
+      "step": 20709
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.009218802515965e-07,
+      "loss": 0.0003,
+      "step": 20710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.985211504297306e-07,
+      "loss": 0.0004,
+      "step": 20711
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9612042060786474e-07,
+      "loss": 0.0004,
+      "step": 20712
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.937196907859989e-07,
+      "loss": 0.0003,
+      "step": 20713
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9131896096413305e-07,
+      "loss": 0.0003,
+      "step": 20714
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.889182311422673e-07,
+      "loss": 0.0002,
+      "step": 20715
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.865175013204014e-07,
+      "loss": 0.0003,
+      "step": 20716
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.841167714985356e-07,
+      "loss": 0.0004,
+      "step": 20717
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8171604167666973e-07,
+      "loss": 0.0002,
+      "step": 20718
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7931531185480386e-07,
+      "loss": 0.0003,
+      "step": 20719
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7691458203293804e-07,
+      "loss": 0.0023,
+      "step": 20720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7451385221107217e-07,
+      "loss": 0.0002,
+      "step": 20721
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7211312238920636e-07,
+      "loss": 0.0003,
+      "step": 20722
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.697123925673405e-07,
+      "loss": 0.0003,
+      "step": 20723
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6731166274547467e-07,
+      "loss": 0.0002,
+      "step": 20724
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.649109329236088e-07,
+      "loss": 0.0002,
+      "step": 20725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6251020310174293e-07,
+      "loss": 0.0004,
+      "step": 20726
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.601094732798771e-07,
+      "loss": 0.4383,
+      "step": 20727
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5770874345801124e-07,
+      "loss": 0.0004,
+      "step": 20728
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.553080136361454e-07,
+      "loss": 0.0002,
+      "step": 20729
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5290728381427955e-07,
+      "loss": 0.0004,
+      "step": 20730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5050655399241373e-07,
+      "loss": 0.0002,
+      "step": 20731
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4810582417054786e-07,
+      "loss": 0.0002,
+      "step": 20732
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4570509434868205e-07,
+      "loss": 0.0002,
+      "step": 20733
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.433043645268162e-07,
+      "loss": 0.0002,
+      "step": 20734
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.409036347049503e-07,
+      "loss": 0.0002,
+      "step": 20735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.385029048830845e-07,
+      "loss": 0.0003,
+      "step": 20736
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.361021750612186e-07,
+      "loss": 0.0002,
+      "step": 20737
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.337014452393528e-07,
+      "loss": 0.0008,
+      "step": 20738
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3130071541748693e-07,
+      "loss": 0.0003,
+      "step": 20739
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.288999855956211e-07,
+      "loss": 0.0002,
+      "step": 20740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2649925577375524e-07,
+      "loss": 0.0003,
+      "step": 20741
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2409852595188937e-07,
+      "loss": 0.0003,
+      "step": 20742
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2169779613002356e-07,
+      "loss": 0.0003,
+      "step": 20743
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.192970663081577e-07,
+      "loss": 0.0002,
+      "step": 20744
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1689633648629187e-07,
+      "loss": 0.0002,
+      "step": 20745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.14495606664426e-07,
+      "loss": 0.0004,
+      "step": 20746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.120948768425602e-07,
+      "loss": 0.0003,
+      "step": 20747
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.096941470206943e-07,
+      "loss": 0.0003,
+      "step": 20748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.072934171988285e-07,
+      "loss": 0.0002,
+      "step": 20749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.048926873769626e-07,
+      "loss": 0.0002,
+      "step": 20750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0249195755509675e-07,
+      "loss": 0.0003,
+      "step": 20751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0009122773323093e-07,
+      "loss": 0.0024,
+      "step": 20752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9769049791136506e-07,
+      "loss": 0.0003,
+      "step": 20753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9528976808949925e-07,
+      "loss": 0.0003,
+      "step": 20754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.928890382676334e-07,
+      "loss": 0.0002,
+      "step": 20755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9048830844576756e-07,
+      "loss": 0.0003,
+      "step": 20756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.880875786239017e-07,
+      "loss": 0.0002,
+      "step": 20757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8568684880203587e-07,
+      "loss": 0.0002,
+      "step": 20758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8328611898017e-07,
+      "loss": 0.0003,
+      "step": 20759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8088538915830413e-07,
+      "loss": 0.0003,
+      "step": 20760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.784846593364383e-07,
+      "loss": 0.9142,
+      "step": 20761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7608392951457244e-07,
+      "loss": 0.0003,
+      "step": 20762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.736831996927066e-07,
+      "loss": 0.0002,
+      "step": 20763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7128246987084075e-07,
+      "loss": 0.0005,
+      "step": 20764
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6888174004897494e-07,
+      "loss": 0.0002,
+      "step": 20765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6648101022710907e-07,
+      "loss": 0.0005,
+      "step": 20766
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.640802804052432e-07,
+      "loss": 0.0005,
+      "step": 20767
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.616795505833774e-07,
+      "loss": 0.0003,
+      "step": 20768
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.592788207615115e-07,
+      "loss": 0.0004,
+      "step": 20769
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.568780909396457e-07,
+      "loss": 0.0002,
+      "step": 20770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.544773611177798e-07,
+      "loss": 0.0003,
+      "step": 20771
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.52076631295914e-07,
+      "loss": 0.0002,
+      "step": 20772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4967590147404813e-07,
+      "loss": 0.0003,
+      "step": 20773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.472751716521823e-07,
+      "loss": 0.0004,
+      "step": 20774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4487444183031645e-07,
+      "loss": 0.0002,
+      "step": 20775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.424737120084506e-07,
+      "loss": 0.0002,
+      "step": 20776
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4007298218658476e-07,
+      "loss": 0.0002,
+      "step": 20777
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3767225236471891e-07,
+      "loss": 0.0002,
+      "step": 20778
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3527152254285307e-07,
+      "loss": 0.0003,
+      "step": 20779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.328707927209872e-07,
+      "loss": 0.0002,
+      "step": 20780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3047006289912136e-07,
+      "loss": 0.0002,
+      "step": 20781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.280693330772555e-07,
+      "loss": 0.0003,
+      "step": 20782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2566860325538967e-07,
+      "loss": 0.0003,
+      "step": 20783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2326787343352382e-07,
+      "loss": 0.7876,
+      "step": 20784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2086714361165798e-07,
+      "loss": 0.0002,
+      "step": 20785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1846641378979214e-07,
+      "loss": 0.0004,
+      "step": 20786
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.160656839679263e-07,
+      "loss": 0.0004,
+      "step": 20787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1366495414606042e-07,
+      "loss": 0.9991,
+      "step": 20788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1126422432419458e-07,
+      "loss": 0.0003,
+      "step": 20789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0886349450232873e-07,
+      "loss": 0.0003,
+      "step": 20790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.064627646804629e-07,
+      "loss": 0.0002,
+      "step": 20791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.04062034858597e-07,
+      "loss": 0.8782,
+      "step": 20792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0166130503673115e-07,
+      "loss": 0.0004,
+      "step": 20793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.992605752148653e-07,
+      "loss": 0.0002,
+      "step": 20794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9685984539299946e-07,
+      "loss": 0.0002,
+      "step": 20795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9445911557113364e-07,
+      "loss": 0.0003,
+      "step": 20796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.920583857492678e-07,
+      "loss": 0.0003,
+      "step": 20797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8965765592740193e-07,
+      "loss": 0.0002,
+      "step": 20798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8725692610553609e-07,
+      "loss": 0.0003,
+      "step": 20799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8485619628367024e-07,
+      "loss": 0.0004,
+      "step": 20800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.824554664618044e-07,
+      "loss": 0.0002,
+      "step": 20801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8005473663993856e-07,
+      "loss": 0.0003,
+      "step": 20802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.776540068180727e-07,
+      "loss": 0.0003,
+      "step": 20803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7525327699620687e-07,
+      "loss": 0.0002,
+      "step": 20804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7285254717434102e-07,
+      "loss": 0.0002,
+      "step": 20805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7045181735247515e-07,
+      "loss": 0.0003,
+      "step": 20806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.680510875306093e-07,
+      "loss": 0.0003,
+      "step": 20807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6565035770874347e-07,
+      "loss": 0.0003,
+      "step": 20808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6324962788687762e-07,
+      "loss": 0.0003,
+      "step": 20809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6084889806501178e-07,
+      "loss": 0.0004,
+      "step": 20810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5844816824314593e-07,
+      "loss": 0.0003,
+      "step": 20811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.560474384212801e-07,
+      "loss": 0.0003,
+      "step": 20812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5364670859941425e-07,
+      "loss": 0.0002,
+      "step": 20813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5124597877754838e-07,
+      "loss": 0.0003,
+      "step": 20814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4884524895568253e-07,
+      "loss": 0.0002,
+      "step": 20815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.464445191338167e-07,
+      "loss": 0.0002,
+      "step": 20816
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4404378931195084e-07,
+      "loss": 0.0002,
+      "step": 20817
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.41643059490085e-07,
+      "loss": 0.0003,
+      "step": 20818
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3924232966821916e-07,
+      "loss": 0.0005,
+      "step": 20819
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.368415998463533e-07,
+      "loss": 0.0003,
+      "step": 20820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3444087002448747e-07,
+      "loss": 0.0004,
+      "step": 20821
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.320401402026216e-07,
+      "loss": 0.0006,
+      "step": 20822
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2963941038075575e-07,
+      "loss": 0.0003,
+      "step": 20823
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.272386805588899e-07,
+      "loss": 0.0004,
+      "step": 20824
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2483795073702407e-07,
+      "loss": 0.0002,
+      "step": 20825
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2243722091515822e-07,
+      "loss": 0.0001,
+      "step": 20826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2003649109329238e-07,
+      "loss": 0.0002,
+      "step": 20827
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1763576127142654e-07,
+      "loss": 0.0003,
+      "step": 20828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1523503144956068e-07,
+      "loss": 0.0002,
+      "step": 20829
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1283430162769483e-07,
+      "loss": 0.0003,
+      "step": 20830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1043357180582899e-07,
+      "loss": 0.0003,
+      "step": 20831
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0803284198396315e-07,
+      "loss": 0.0003,
+      "step": 20832
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0563211216209729e-07,
+      "loss": 0.0004,
+      "step": 20833
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0323138234023145e-07,
+      "loss": 0.0003,
+      "step": 20834
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0083065251836557e-07,
+      "loss": 0.0002,
+      "step": 20835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.842992269649973e-08,
+      "loss": 0.0002,
+      "step": 20836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.60291928746339e-08,
+      "loss": 0.0003,
+      "step": 20837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.362846305276804e-08,
+      "loss": 0.0002,
+      "step": 20838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.12277332309022e-08,
+      "loss": 0.0003,
+      "step": 20839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.882700340903636e-08,
+      "loss": 0.0003,
+      "step": 20840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.642627358717051e-08,
+      "loss": 0.0003,
+      "step": 20841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.402554376530465e-08,
+      "loss": 0.0003,
+      "step": 20842
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.162481394343881e-08,
+      "loss": 0.0002,
+      "step": 20843
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.922408412157297e-08,
+      "loss": 0.0002,
+      "step": 20844
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.682335429970712e-08,
+      "loss": 0.0002,
+      "step": 20845
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.442262447784127e-08,
+      "loss": 0.0003,
+      "step": 20846
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.202189465597542e-08,
+      "loss": 0.0003,
+      "step": 20847
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.962116483410958e-08,
+      "loss": 0.0003,
+      "step": 20848
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.722043501224373e-08,
+      "loss": 0.0003,
+      "step": 20849
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.481970519037788e-08,
+      "loss": 0.0004,
+      "step": 20850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.241897536851203e-08,
+      "loss": 0.0003,
+      "step": 20851
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.001824554664619e-08,
+      "loss": 0.0002,
+      "step": 20852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.761751572478034e-08,
+      "loss": 0.0001,
+      "step": 20853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.5216785902914495e-08,
+      "loss": 0.0002,
+      "step": 20854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.2816056081048645e-08,
+      "loss": 0.0003,
+      "step": 20855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.041532625918279e-08,
+      "loss": 0.0003,
+      "step": 20856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.801459643731695e-08,
+      "loss": 0.0002,
+      "step": 20857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.56138666154511e-08,
+      "loss": 0.0003,
+      "step": 20858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.3213136793585256e-08,
+      "loss": 0.0003,
+      "step": 20859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0812406971719405e-08,
+      "loss": 0.0003,
+      "step": 20860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.841167714985356e-08,
+      "loss": 0.0004,
+      "step": 20861
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.601094732798771e-08,
+      "loss": 0.0003,
+      "step": 20862
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.361021750612187e-08,
+      "loss": 0.8909,
+      "step": 20863
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1209487684256017e-08,
+      "loss": 0.0002,
+      "step": 20864
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.880875786239017e-08,
+      "loss": 0.0003,
+      "step": 20865
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.6408028040524322e-08,
+      "loss": 0.0003,
+      "step": 20866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.4007298218658475e-08,
+      "loss": 0.0003,
+      "step": 20867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1606568396792628e-08,
+      "loss": 0.0004,
+      "step": 20868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.920583857492678e-08,
+      "loss": 0.0004,
+      "step": 20869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6805108753060934e-08,
+      "loss": 0.0003,
+      "step": 20870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4404378931195085e-08,
+      "loss": 0.0003,
+      "step": 20871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2003649109329238e-08,
+      "loss": 0.0001,
+      "step": 20872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.60291928746339e-09,
+      "loss": 0.0003,
+      "step": 20873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.202189465597542e-09,
+      "loss": 0.0004,
+      "step": 20874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.801459643731695e-09,
+      "loss": 0.0003,
+      "step": 20875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.4007298218658476e-09,
+      "loss": 0.0002,
+      "step": 20876
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0004,
+      "step": 20877
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9943234892577423,
+      "eval_loss": 0.03778468817472458,
+      "eval_runtime": 265.0103,
+      "eval_samples_per_second": 52.515,
+      "eval_steps_per_second": 1.641,
+      "step": 20877
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 20877,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 4.393288162153218e+16,
+  "trial_name": null,
+  "trial_params": null
+}