diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14458 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 15.0,
+  "global_step": 2405,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 1.5611,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 1.6091,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.5721,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.591,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.5884,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.564,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.6268,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.5956,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.5867,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.5701,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.5979,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.6109,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.5324,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 1.5543,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 1.5795,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 1.5349,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.5529,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.5375,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 1.5317,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 1.5135,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.5551,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 1.5646,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 1.5566,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 1.5694,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 1.5738,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.5226,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 1.5462,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 1.5596,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.5353,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 1.5625,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 1.5176,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 1.5243,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 1.5416,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 1.5412,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 1.5527,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 1.5566,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 1.5423,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 1.5387,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 1.5725,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 1.5502,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 1.5143,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 1.5335,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 1.5432,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 1.5364,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 1.5493,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 1.537,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 1.5195,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 1.4999,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 1.5318,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 1.5244,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 1.5387,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 1.5433,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 1.5483,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 1.5202,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 1.5188,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 1.5481,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 1.557,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 1.5548,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 1.5516,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 1.5256,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 1.5522,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 1.5311,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 1.5492,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 1.5454,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 1.5548,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 1.5382,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 1.555,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 1.5314,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 1.5262,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 1.5353,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 1.5724,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 1.5203,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.532,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999092571131e-05,
+      "loss": 1.554,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963702861706e-05,
+      "loss": 1.5384,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999918331500595e-05,
+      "loss": 1.5481,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999854811710317e-05,
+      "loss": 1.5492,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999773143606153e-05,
+      "loss": 1.526,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999673327336323e-05,
+      "loss": 1.5555,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955536308197e-05,
+      "loss": 1.5492,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999941925105719e-05,
+      "loss": 1.5445,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999264991509007e-05,
+      "loss": 1.5637,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999092584717375e-05,
+      "loss": 1.5394,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999890203099519e-05,
+      "loss": 1.571,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998693330688283e-05,
+      "loss": 1.5714,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998466484175412e-05,
+      "loss": 1.5401,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998221491868274e-05,
+      "loss": 1.5385,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999795835421149e-05,
+      "loss": 1.5305,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997677071682623e-05,
+      "loss": 1.541,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999737764479216e-05,
+      "loss": 1.5606,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997060074083513e-05,
+      "loss": 1.5395,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996724360133032e-05,
+      "loss": 1.5701,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999637050354999e-05,
+      "loss": 1.5448,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995998504976585e-05,
+      "loss": 1.543,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.5286,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995200084592114e-05,
+      "loss": 1.5873,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994773664230064e-05,
+      "loss": 1.5431,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432910477569e-05,
+      "loss": 1.5696,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99938664070358e-05,
+      "loss": 1.5403,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993385571850126e-05,
+      "loss": 1.5371,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992886600091318e-05,
+      "loss": 1.5458,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992369492664937e-05,
+      "loss": 1.5388,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991834250509462e-05,
+      "loss": 1.539,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991280874596275e-05,
+      "loss": 1.54,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990709365929678e-05,
+      "loss": 1.5889,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999011972554688e-05,
+      "loss": 1.5483,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989511954517992e-05,
+      "loss": 1.5168,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988886053946034e-05,
+      "loss": 1.5139,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988242024966924e-05,
+      "loss": 1.5308,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987579868749484e-05,
+      "loss": 1.555,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986899586495432e-05,
+      "loss": 1.4917,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986201179439388e-05,
+      "loss": 1.5392,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985484648848854e-05,
+      "loss": 1.5301,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984749996024236e-05,
+      "loss": 1.5471,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998399722229883e-05,
+      "loss": 1.5147,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983226329038798e-05,
+      "loss": 1.5347,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 1.5505,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981630189544023e-05,
+      "loss": 1.5406,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980804946206036e-05,
+      "loss": 1.5221,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997996158912696e-05,
+      "loss": 1.5252,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997910011983737e-05,
+      "loss": 1.5309,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978220539900707e-05,
+      "loss": 1.547,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977322850913283e-05,
+      "loss": 1.5583,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976407054504276e-05,
+      "loss": 1.5346,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975473152335726e-05,
+      "loss": 1.5362,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974521146102535e-05,
+      "loss": 1.5886,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997355103753246e-05,
+      "loss": 1.5246,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99725628283861e-05,
+      "loss": 1.5588,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971556520456928e-05,
+      "loss": 1.5384,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997053211557124e-05,
+      "loss": 1.5415,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996948961558819e-05,
+      "loss": 1.5079,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968429022399762e-05,
+      "loss": 1.5547,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996735033793079e-05,
+      "loss": 1.5494,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966253564138922e-05,
+      "loss": 1.5676,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965138703014656e-05,
+      "loss": 1.5462,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99640057565813e-05,
+      "loss": 1.5461,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962854726894997e-05,
+      "loss": 1.5523,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961685616044696e-05,
+      "loss": 1.5252,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 1.5151,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959293159371998e-05,
+      "loss": 1.5499,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995806981789157e-05,
+      "loss": 1.5603,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956828403931078e-05,
+      "loss": 1.5403,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955568919743507e-05,
+      "loss": 1.546,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954291367614642e-05,
+      "loss": 1.5775,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995299574986306e-05,
+      "loss": 1.5319,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951682068840124e-05,
+      "loss": 1.5395,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995035032692998e-05,
+      "loss": 1.5207,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949000526549546e-05,
+      "loss": 1.5451,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947632670148517e-05,
+      "loss": 1.5312,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994624676020936e-05,
+      "loss": 1.5529,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944842799247306e-05,
+      "loss": 1.546,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943420789810343e-05,
+      "loss": 1.5294,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941980734479214e-05,
+      "loss": 1.5412,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994052263586742e-05,
+      "loss": 1.512,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939046496621196e-05,
+      "loss": 1.497,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937552319419525e-05,
+      "loss": 1.5711,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936040106974132e-05,
+      "loss": 1.5638,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993450986202946e-05,
+      "loss": 1.5176,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932961587362692e-05,
+      "loss": 1.5426,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931395285783726e-05,
+      "loss": 1.5307,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 1.5531,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928208613292358e-05,
+      "loss": 1.5481,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926588248163317e-05,
+      "loss": 1.538,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924949867688778e-05,
+      "loss": 1.5363,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923293474842175e-05,
+      "loss": 1.545,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921619072629617e-05,
+      "loss": 1.5522,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919926664089907e-05,
+      "loss": 1.5015,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918216252294532e-05,
+      "loss": 1.4942,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916487840347644e-05,
+      "loss": 1.5321,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914741431386063e-05,
+      "loss": 1.5171,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991297702857927e-05,
+      "loss": 1.5098,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991119463512941e-05,
+      "loss": 1.5524,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990939425427127e-05,
+      "loss": 1.5386,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990757588927229e-05,
+      "loss": 1.5803,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905739543432536e-05,
+      "loss": 1.4975,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903885220084723e-05,
+      "loss": 1.5393,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902012922594178e-05,
+      "loss": 1.5572,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900122654358857e-05,
+      "loss": 1.5225,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 1.5529,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989628821940877e-05,
+      "loss": 1.5202,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894344059652953e-05,
+      "loss": 1.5199,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892381943070255e-05,
+      "loss": 1.5553,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 1.5418,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888403853700654e-05,
+      "loss": 1.5307,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886387888133413e-05,
+      "loss": 1.5705,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988435398017861e-05,
+      "loss": 1.5527,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98823021335275e-05,
+      "loss": 1.5601,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988023235190389e-05,
+      "loss": 1.5464,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878144639064145e-05,
+      "loss": 1.5107,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987603899879716e-05,
+      "loss": 1.5627,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873915434924375e-05,
+      "loss": 1.5447,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871773951299757e-05,
+      "loss": 1.5423,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869614551809793e-05,
+      "loss": 1.5563,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986743724037349e-05,
+      "loss": 1.5335,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865242020942353e-05,
+      "loss": 1.5354,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863028897500396e-05,
+      "loss": 1.5299,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860797874064123e-05,
+      "loss": 1.4861,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858548954682518e-05,
+      "loss": 1.5467,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856282143437063e-05,
+      "loss": 1.509,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853997444441682e-05,
+      "loss": 1.5079,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851694861842795e-05,
+      "loss": 1.5117,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984937439981925e-05,
+      "loss": 1.5175,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984703606258236e-05,
+      "loss": 1.5113,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984467985437587e-05,
+      "loss": 1.5132,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 1.5048,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983991384219126e-05,
+      "loss": 1.5271,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837504046862777e-05,
+      "loss": 1.4897,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835076397863946e-05,
+      "loss": 1.5286,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832630899600607e-05,
+      "loss": 1.4938,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830167556511e-05,
+      "loss": 1.5135,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982768637306573e-05,
+      "loss": 1.5221,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98251873537678e-05,
+      "loss": 1.5111,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822670503152567e-05,
+      "loss": 1.5324,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820135825787764e-05,
+      "loss": 1.5298,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817583326273468e-05,
+      "loss": 1.5354,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815013009242103e-05,
+      "loss": 1.5319,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812424879358424e-05,
+      "loss": 1.5199,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809818941319526e-05,
+      "loss": 1.5015,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980719519985481e-05,
+      "loss": 1.5266,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804553659726e-05,
+      "loss": 1.5297,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801894325727104e-05,
+      "loss": 1.542,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799217202684445e-05,
+      "loss": 1.5446,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796522295456618e-05,
+      "loss": 1.5351,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793809608934498e-05,
+      "loss": 1.5184,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979107914804122e-05,
+      "loss": 1.516,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788330917732186e-05,
+      "loss": 1.5218,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 1.5374,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782781168849673e-05,
+      "loss": 1.5514,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97799796603482e-05,
+      "loss": 1.5319,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977716040257496e-05,
+      "loss": 1.5383,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977432340064651e-05,
+      "loss": 1.521,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771468659711595e-05,
+      "loss": 1.5351,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768596184951174e-05,
+      "loss": 1.5576,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765705981578374e-05,
+      "loss": 1.5229,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762798054838503e-05,
+      "loss": 1.5257,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759872410009036e-05,
+      "loss": 1.5182,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756929052399606e-05,
+      "loss": 1.5318,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975396798735198e-05,
+      "loss": 1.5017,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750989220240073e-05,
+      "loss": 1.5461,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747992756469925e-05,
+      "loss": 1.5449,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744978601479693e-05,
+      "loss": 1.4827,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974194676073964e-05,
+      "loss": 1.5038,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973889723975212e-05,
+      "loss": 1.4922,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735830044051582e-05,
+      "loss": 1.5477,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732745179204553e-05,
+      "loss": 1.5244,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972964265080962e-05,
+      "loss": 1.5171,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726522464497434e-05,
+      "loss": 1.5106,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723384625930685e-05,
+      "loss": 1.5265,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 1.4503,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971705601484446e-05,
+      "loss": 1.5026,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713865253810508e-05,
+      "loss": 1.5384,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971065686349303e-05,
+      "loss": 1.5434,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970743084971481e-05,
+      "loss": 1.502,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704187218330584e-05,
+      "loss": 1.5263,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700925975227096e-05,
+      "loss": 1.5156,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697647126323036e-05,
+      "loss": 1.5252,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694350677569043e-05,
+      "loss": 1.5311,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969103663494771e-05,
+      "loss": 1.5176,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687705004473545e-05,
+      "loss": 1.5416,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968435579219299e-05,
+      "loss": 1.5192,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680989004184383e-05,
+      "loss": 1.5193,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967760464655797e-05,
+      "loss": 1.5254,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674202725455877e-05,
+      "loss": 1.5176,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967078324705211e-05,
+      "loss": 1.5181,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667346217552528e-05,
+      "loss": 1.5496,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663891643194857e-05,
+      "loss": 1.5227,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660419530248657e-05,
+      "loss": 1.5301,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656929885015318e-05,
+      "loss": 1.5284,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965342271382805e-05,
+      "loss": 1.5079,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964989802305187e-05,
+      "loss": 1.5182,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 1.5381,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642796108351804e-05,
+      "loss": 1.4935,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639218897316885e-05,
+      "loss": 1.4931,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963562419247096e-05,
+      "loss": 1.5072,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963201200033791e-05,
+      "loss": 1.5259,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962838232747334e-05,
+      "loss": 1.5159,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624735180464602e-05,
+      "loss": 1.5187,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962107056593074e-05,
+      "loss": 1.5251,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617388490522516e-05,
+      "loss": 1.4696,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613688960922367e-05,
+      "loss": 1.4731,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609971983844412e-05,
+      "loss": 1.5347,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606237566034443e-05,
+      "loss": 1.5219,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960248571426989e-05,
+      "loss": 1.5023,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598716435359834e-05,
+      "loss": 1.5089,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 1.538,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959112562349764e-05,
+      "loss": 1.5376,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587304104321746e-05,
+      "loss": 1.5106,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583465185552814e-05,
+      "loss": 1.5099,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957960887415793e-05,
+      "loss": 1.5299,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957573517713575e-05,
+      "loss": 1.5394,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571844101516487e-05,
+      "loss": 1.5222,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567935654361888e-05,
+      "loss": 1.5246,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 1.5047,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560066673851294e-05,
+      "loss": 1.5271,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955610615477638e-05,
+      "loss": 1.4968,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552128292728266e-05,
+      "loss": 1.4749,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548133094926203e-05,
+      "loss": 1.5142,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544120568620905e-05,
+      "loss": 1.4825,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540090721094542e-05,
+      "loss": 1.5251,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536043559660708e-05,
+      "loss": 1.4998,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953197909166443e-05,
+      "loss": 1.509,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952789732448214e-05,
+      "loss": 1.535,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523798265521656e-05,
+      "loss": 1.5408,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519681922222195e-05,
+      "loss": 1.5243,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515548302054335e-05,
+      "loss": 1.4963,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511397412520003e-05,
+      "loss": 1.497,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507229261152476e-05,
+      "loss": 1.5132,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950304385551636e-05,
+      "loss": 1.5178,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949884120320756e-05,
+      "loss": 1.5233,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94946213118533e-05,
+      "loss": 1.512,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490384189112084e-05,
+      "loss": 1.5489,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948612984267368e-05,
+      "loss": 1.5427,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948185828025913e-05,
+      "loss": 1.5063,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477569509620703e-05,
+      "loss": 1.4969,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 1.5069,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946894037483749e-05,
+      "loss": 1.4935,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946460002635335e-05,
+      "loss": 1.4938,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460242500966614e-05,
+      "loss": 1.4989,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945586780658557e-05,
+      "loss": 1.4804,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451475951149665e-05,
+      "loss": 1.4927,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447066942629495e-05,
+      "loss": 1.5206,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442640789026778e-05,
+      "loss": 1.514,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438197498374357e-05,
+      "loss": 1.5198,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433737078736172e-05,
+      "loss": 1.4828,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942925953820725e-05,
+      "loss": 1.5218,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942476488491369e-05,
+      "loss": 1.517,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420253127012647e-05,
+      "loss": 1.5074,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941572427269232e-05,
+      "loss": 1.4696,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941117833017194e-05,
+      "loss": 1.4995,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406615307701736e-05,
+      "loss": 1.4958,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402035213562956e-05,
+      "loss": 1.5187,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939743805606781e-05,
+      "loss": 1.5193,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939282384355949e-05,
+      "loss": 1.5165,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388192584412137e-05,
+      "loss": 1.5508,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938354428703082e-05,
+      "loss": 1.4897,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937887895985154e-05,
+      "loss": 1.542,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 1.5254,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936949724999762e-05,
+      "loss": 1.4993,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936478088434944e-05,
+      "loss": 1.5238,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360047522956207e-05,
+      "loss": 1.5357,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355297174408298e-05,
+      "loss": 1.5329,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935052984732692e-05,
+      "loss": 1.4913,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934574555036409e-05,
+      "loss": 1.5288,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340944292202626e-05,
+      "loss": 1.4972,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336126081556134e-05,
+      "loss": 1.4849,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933129092716898e-05,
+      "loss": 1.4895,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326438837816276e-05,
+      "loss": 1.5201,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932156982230388e-05,
+      "loss": 1.5317,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931668388946836e-05,
+      "loss": 1.5085,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311781048176986e-05,
+      "loss": 1.4817,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306861307327724e-05,
+      "loss": 1.5007,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301924675849196e-05,
+      "loss": 1.499,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296971162700696e-05,
+      "loss": 1.5352,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292000776872135e-05,
+      "loss": 1.5264,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928701352738406e-05,
+      "loss": 1.491,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282009423287623e-05,
+      "loss": 1.492,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276988473664557e-05,
+      "loss": 1.5071,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271950687627175e-05,
+      "loss": 1.5214,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 1.5038,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926182464291145e-05,
+      "loss": 1.5128,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256736402610437e-05,
+      "loss": 1.4921,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251631362649735e-05,
+      "loss": 1.4869,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246509532294265e-05,
+      "loss": 1.5023,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924137092083942e-05,
+      "loss": 1.5225,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236215537611044e-05,
+      "loss": 1.4925,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231043391965435e-05,
+      "loss": 1.5027,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225854493289287e-05,
+      "loss": 1.4967,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220648850999723e-05,
+      "loss": 1.5037,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215426474544242e-05,
+      "loss": 1.5041,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921018737340071e-05,
+      "loss": 1.5277,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204931557077355e-05,
+      "loss": 1.5401,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199659035112734e-05,
+      "loss": 1.4946,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9194369817075725e-05,
+      "loss": 1.526,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189063912565506e-05,
+      "loss": 1.5142,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183741331211535e-05,
+      "loss": 1.5411,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178402082673548e-05,
+      "loss": 1.4767,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173046176641515e-05,
+      "loss": 1.4983,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167673622835644e-05,
+      "loss": 1.5178,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916228443100636e-05,
+      "loss": 1.4953,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156878610934276e-05,
+      "loss": 1.5096,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.4909,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146017125335044e-05,
+      "loss": 1.5186,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140561479519955e-05,
+      "loss": 1.5213,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135089244886128e-05,
+      "loss": 1.4865,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91296004313649e-05,
+      "loss": 1.496,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912409504891768e-05,
+      "loss": 1.4767,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118573107535955e-05,
+      "loss": 1.5148,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113034617241265e-05,
+      "loss": 1.4965,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107479588085182e-05,
+      "loss": 1.4777,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101908030149294e-05,
+      "loss": 1.5009,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 1.5233,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909071536841442e-05,
+      "loss": 1.5651,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908509428492852e-05,
+      "loss": 1.5008,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079456713288958e-05,
+      "loss": 1.4773,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907380266372712e-05,
+      "loss": 1.4999,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90681321465043e-05,
+      "loss": 1.504,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9062445171911688e-05,
+      "loss": 1.526,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056741750270328e-05,
+      "loss": 1.5257,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905102189193112e-05,
+      "loss": 1.5182,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045285607274787e-05,
+      "loss": 1.4915,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903953290671188e-05,
+      "loss": 1.4876,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033763800682728e-05,
+      "loss": 1.4999,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 1.4795,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022176414135868e-05,
+      "loss": 1.5195,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016358154647618e-05,
+      "loss": 1.5151,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010523531752004e-05,
+      "loss": 1.4991,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900467255603803e-05,
+      "loss": 1.4903,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998805238124386e-05,
+      "loss": 1.5068,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992921588659424e-05,
+      "loss": 1.476,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898702161832113e-05,
+      "loss": 1.4889,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981105337817106e-05,
+      "loss": 1.5114,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897517275788456e-05,
+      "loss": 1.4981,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969223889290283e-05,
+      "loss": 1.5237,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896325874283063e-05,
+      "loss": 1.5009,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957277329331485e-05,
+      "loss": 1.4785,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951279659648265e-05,
+      "loss": 1.5215,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945265744665886e-05,
+      "loss": 1.4927,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939235595298756e-05,
+      "loss": 1.4965,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933189222490727e-05,
+      "loss": 1.5055,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927126637215112e-05,
+      "loss": 1.4722,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8921047850474645e-05,
+      "loss": 1.5099,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914952873301448e-05,
+      "loss": 1.4822,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908841716757044e-05,
+      "loss": 1.5267,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902714391932312e-05,
+      "loss": 1.4929,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 1.5126,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889041128195208e-05,
+      "loss": 1.5027,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884235519124972e-05,
+      "loss": 1.4759,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878043632674284e-05,
+      "loss": 1.5109,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887183563383741e-05,
+      "loss": 1.514,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886561153388099e-05,
+      "loss": 1.5173,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859371344100865e-05,
+      "loss": 1.4943,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853115075822105e-05,
+      "loss": 1.4818,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884684274039894e-05,
+      "loss": 1.5424,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840554349214772e-05,
+      "loss": 1.5225,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834249913682132e-05,
+      "loss": 1.4969,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827929445242678e-05,
+      "loss": 1.4889,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821592955367154e-05,
+      "loss": 1.5042,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881524045555539e-05,
+      "loss": 1.4959,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808871957336277e-05,
+      "loss": 1.5029,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802487472267725e-05,
+      "loss": 1.5143,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796087011936665e-05,
+      "loss": 1.4875,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789670587959032e-05,
+      "loss": 1.5095,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878323821197971e-05,
+      "loss": 1.5034,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776789895672557e-05,
+      "loss": 1.5005,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8770325650740347e-05,
+      "loss": 1.4934,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763845488914757e-05,
+      "loss": 1.4748,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 1.4859,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750837461654618e-05,
+      "loss": 1.4546,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874430961982778e-05,
+      "loss": 1.4927,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737765908322967e-05,
+      "loss": 1.4715,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731206339016078e-05,
+      "loss": 1.4685,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87246309238118e-05,
+      "loss": 1.5098,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871803967464358e-05,
+      "loss": 1.4678,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711432603473593e-05,
+      "loss": 1.4952,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870480972229274e-05,
+      "loss": 1.5088,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698171043120595e-05,
+      "loss": 1.4646,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691516578005426e-05,
+      "loss": 1.4969,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684846339024145e-05,
+      "loss": 1.4352,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678160338282273e-05,
+      "loss": 1.4942,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671458587913962e-05,
+      "loss": 1.5237,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866474110008193e-05,
+      "loss": 1.5337,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865800788697746e-05,
+      "loss": 1.499,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651258960820383e-05,
+      "loss": 1.4919,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644494333859037e-05,
+      "loss": 1.4942,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637714018370255e-05,
+      "loss": 1.51,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630918026659346e-05,
+      "loss": 1.5016,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862410637106007e-05,
+      "loss": 1.493,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617279063934607e-05,
+      "loss": 1.4642,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 1.517,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603577544695887e-05,
+      "loss": 1.4785,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596703357448936e-05,
+      "loss": 1.5066,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589813568408373e-05,
+      "loss": 1.4875,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582908190078184e-05,
+      "loss": 1.4779,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575987234990654e-05,
+      "loss": 1.4597,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569050715706325e-05,
+      "loss": 1.4898,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562098644813996e-05,
+      "loss": 1.521,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555131034930686e-05,
+      "loss": 1.475,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548147898701616e-05,
+      "loss": 1.4987,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541149248800185e-05,
+      "loss": 1.4786,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853413509792795e-05,
+      "loss": 1.5336,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852710545881459e-05,
+      "loss": 1.4976,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520060344217903e-05,
+      "loss": 1.4654,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512999766923774e-05,
+      "loss": 1.4665,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850592373974614e-05,
+      "loss": 1.5047,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498832275526988e-05,
+      "loss": 1.5131,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491725387136313e-05,
+      "loss": 1.5047,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848460308747211e-05,
+      "loss": 1.4896,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847746538946034e-05,
+      "loss": 1.5024,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470312306054903e-05,
+      "loss": 1.5,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463143850237636e-05,
+      "loss": 1.4943,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 1.5345,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448760873434387e-05,
+      "loss": 1.4889,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441546378551457e-05,
+      "loss": 1.503,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434316563462764e-05,
+      "loss": 1.5042,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842707144128939e-05,
+      "loss": 1.469,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419811025180195e-05,
+      "loss": 1.4341,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 1.5246,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840524436388859e-05,
+      "loss": 1.5092,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839793814514259e-05,
+      "loss": 1.4743,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390616685333565e-05,
+      "loss": 1.4905,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838327999774892e-05,
+      "loss": 1.4845,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375928095703703e-05,
+      "loss": 1.499,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368560992540563e-05,
+      "loss": 1.5105,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836117870162975e-05,
+      "loss": 1.5368,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353781236369065e-05,
+      "loss": 1.4705,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346368610183863e-05,
+      "loss": 1.5125,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338940836527005e-05,
+      "loss": 1.4811,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833149792887884e-05,
+      "loss": 1.5038,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832403990074719e-05,
+      "loss": 1.4883,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831656676566731e-05,
+      "loss": 1.4814,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309078537201882e-05,
+      "loss": 1.493,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301575228940974e-05,
+      "loss": 1.4884,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 1.4724,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286523427529802e-05,
+      "loss": 1.4938,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278974961696422e-05,
+      "loss": 1.4623,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271411470701267e-05,
+      "loss": 1.4735,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263832968271e-05,
+      "loss": 1.5092,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825623946815952e-05,
+      "loss": 1.4916,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248630984147956e-05,
+      "loss": 1.5028,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241007530044622e-05,
+      "loss": 1.4989,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233369119685e-05,
+      "loss": 1.4672,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822571576693172e-05,
+      "loss": 1.4649,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218047485674522e-05,
+      "loss": 1.4636,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821036428983026e-05,
+      "loss": 1.5131,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202666193342834e-05,
+      "loss": 1.4984,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194953210183185e-05,
+      "loss": 1.4863,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187225354349297e-05,
+      "loss": 1.5016,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179482639866115e-05,
+      "loss": 1.4745,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817172508078557e-05,
+      "loss": 1.4645,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163952691186528e-05,
+      "loss": 1.4684,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815616548517477e-05,
+      "loss": 1.4952,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814836347688297e-05,
+      "loss": 1.4644,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814054668047066e-05,
+      "loss": 1.5216,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813271511012421e-05,
+      "loss": 1.4792,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 1.5186,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811700770450844e-05,
+      "loss": 1.466,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109131897745823e-05,
+      "loss": 1.4832,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101241374062425e-05,
+      "loss": 1.4635,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093336147778437e-05,
+      "loss": 1.4872,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085416233240712e-05,
+      "loss": 1.4878,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807748164482277e-05,
+      "loss": 1.4749,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806953239692476e-05,
+      "loss": 1.4599,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061568503973435e-05,
+      "loss": 1.4739,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053589980422126e-05,
+      "loss": 1.4706,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045596840750722e-05,
+      "loss": 1.497,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037589099465637e-05,
+      "loss": 1.4992,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029566771099775e-05,
+      "loss": 1.4909,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021529870212525e-05,
+      "loss": 1.474,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801347841138972e-05,
+      "loss": 1.5063,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005412409243604e-05,
+      "loss": 1.5457,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997331878412837e-05,
+      "loss": 1.4755,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798923683356242e-05,
+      "loss": 1.478,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981127289383718e-05,
+      "loss": 1.4772,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797300326059439e-05,
+      "loss": 1.4686,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964864761938404e-05,
+      "loss": 1.4459,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956711808185967e-05,
+      "loss": 1.486,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 1.4997,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940362594603767e-05,
+      "loss": 1.4772,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932166364445498e-05,
+      "loss": 1.4757,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792395573853372e-05,
+      "loss": 1.5406,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915730731769558e-05,
+      "loss": 1.5183,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907491359080224e-05,
+      "loss": 1.4959,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899237635419002e-05,
+      "loss": 1.486,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890969575765232e-05,
+      "loss": 1.4966,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788268719512427e-05,
+      "loss": 1.4715,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874390508527454e-05,
+      "loss": 1.5097,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866079531032088e-05,
+      "loss": 1.4767,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857754277721417e-05,
+      "loss": 1.5086,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849414763704587e-05,
+      "loss": 1.4804,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841061004116635e-05,
+      "loss": 1.4478,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832693014118447e-05,
+      "loss": 1.4905,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824310808896728e-05,
+      "loss": 1.4967,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781591440366399e-05,
+      "loss": 1.4728,
+      "step": 573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807503813658518e-05,
+      "loss": 1.4827,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799079054144334e-05,
+      "loss": 1.4852,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790640140411176e-05,
+      "loss": 1.5133,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778218708777448e-05,
+      "loss": 1.4902,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777371991157532e-05,
+      "loss": 1.4588,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 1.4646,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775674324998212e-05,
+      "loss": 1.5052,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748233795398308e-05,
+      "loss": 1.479,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739710278872428e-05,
+      "loss": 1.5095,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773117271587346e-05,
+      "loss": 1.488,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722621121895867e-05,
+      "loss": 1.4669,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771405551245957e-05,
+      "loss": 1.5104,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770547590310993e-05,
+      "loss": 1.4857,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769688230941772e-05,
+      "loss": 1.4911,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688274746979098e-05,
+      "loss": 1.4681,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767965323141555e-05,
+      "loss": 1.4724,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671017778373914e-05,
+      "loss": 1.464,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76623684035263e-05,
+      "loss": 1.5086,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76537051225701e-05,
+      "loss": 1.4581,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764502795122793e-05,
+      "loss": 1.4599,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636336905247625e-05,
+      "loss": 1.4764,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627632000402193e-05,
+      "loss": 1.5184,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618913252489802e-05,
+      "loss": 1.4761,
+      "step": 596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761018067733374e-05,
+      "loss": 1.4876,
+      "step": 597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601434290782386e-05,
+      "loss": 1.4424,
+      "step": 598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759267410870919e-05,
+      "loss": 1.4611,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758390014701263e-05,
+      "loss": 1.4658,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 1.4616,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756631094846838e-05,
+      "loss": 1.4668,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 1.4926,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548666822837157e-05,
+      "loss": 1.4493,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753982420237533e-05,
+      "loss": 1.4887,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530967898205208e-05,
+      "loss": 1.4823,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522097926399722e-05,
+      "loss": 1.4748,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513214303056603e-05,
+      "loss": 1.4671,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750431704429837e-05,
+      "loss": 1.4819,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495406166272278e-05,
+      "loss": 1.4825,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74864816851503e-05,
+      "loss": 1.4799,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747754361712911e-05,
+      "loss": 1.4866,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468591978430024e-05,
+      "loss": 1.4816,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745962678529899e-05,
+      "loss": 1.4725,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745064805400656e-05,
+      "loss": 1.4709,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441655800847855e-05,
+      "loss": 1.5013,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432650042142535e-05,
+      "loss": 1.4521,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423630794234774e-05,
+      "loss": 1.4621,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414598073493216e-05,
+      "loss": 1.416,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405551896310974e-05,
+      "loss": 1.4468,
+      "step": 620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396492279105562e-05,
+      "loss": 1.4918,
+      "step": 621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387419238318904e-05,
+      "loss": 1.4664,
+      "step": 622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 1.4637,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369232951891286e-05,
+      "loss": 1.4608,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736011973925585e-05,
+      "loss": 1.4623,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350993169050152e-05,
+      "loss": 1.4872,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734185325783762e-05,
+      "loss": 1.5107,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332700022205888e-05,
+      "loss": 1.5041,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323533478766777e-05,
+      "loss": 1.4822,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314353644156264e-05,
+      "loss": 1.4931,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305160535034438e-05,
+      "loss": 1.4771,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729595416808549e-05,
+      "loss": 1.4807,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728673456001766e-05,
+      "loss": 1.4647,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277501727563227e-05,
+      "loss": 1.4782,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726825568747847e-05,
+      "loss": 1.4867,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258996456543634e-05,
+      "loss": 1.4539,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249724051562905e-05,
+      "loss": 1.4867,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240438489364387e-05,
+      "loss": 1.4818,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231139786800043e-05,
+      "loss": 1.45,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221827960745706e-05,
+      "loss": 1.4687,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212503028101012e-05,
+      "loss": 1.4844,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203165005789378e-05,
+      "loss": 1.474,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193813910758e-05,
+      "loss": 1.46,
+      "step": 643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184449759977776e-05,
+      "loss": 1.4853,
+      "step": 644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 1.4988,
+      "step": 645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165682359172874e-05,
+      "loss": 1.4975,
+      "step": 646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715627914320835e-05,
+      "loss": 1.4664,
+      "step": 647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714686293961525e-05,
+      "loss": 1.4742,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137433765482644e-05,
+      "loss": 1.4874,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127991637923133e-05,
+      "loss": 1.4704,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711853657407284e-05,
+      "loss": 1.4687,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109068591091366e-05,
+      "loss": 1.4836,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709958770616174e-05,
+      "loss": 1.4703,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090093936490433e-05,
+      "loss": 1.4693,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080587299307283e-05,
+      "loss": 1.48,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.4521,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061535491441538e-05,
+      "loss": 1.4666,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051990355335266e-05,
+      "loss": 1.4825,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704243242086973e-05,
+      "loss": 1.4585,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032861705391218e-05,
+      "loss": 1.4783,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023278226269222e-05,
+      "loss": 1.4611,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013682000896388e-05,
+      "loss": 1.4952,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004073046688498e-05,
+      "loss": 1.4818,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994451381084443e-05,
+      "loss": 1.4613,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984817021546177e-05,
+      "loss": 1.4641,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975169985558686e-05,
+      "loss": 1.4595,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 1.4648,
+      "step": 667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955837954291007e-05,
+      "loss": 1.487,
+      "step": 668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946152994095705e-05,
+      "loss": 1.4873,
+      "step": 669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936455427620894e-05,
+      "loss": 1.4681,
+      "step": 670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692674527246627e-05,
+      "loss": 1.4941,
+      "step": 671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917022546254388e-05,
+      "loss": 1.4822,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907287266630614e-05,
+      "loss": 1.5096,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897539451263093e-05,
+      "loss": 1.4441,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887779117842727e-05,
+      "loss": 1.4785,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687800628408313e-05,
+      "loss": 1.4655,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868220967720604e-05,
+      "loss": 1.4751,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858423186514108e-05,
+      "loss": 1.4815,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848612958245215e-05,
+      "loss": 1.4831,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838790300718107e-05,
+      "loss": 1.4598,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828955231759495e-05,
+      "loss": 1.489,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681910776921864e-05,
+      "loss": 1.476,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809247930967282e-05,
+      "loss": 1.4994,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799375734899624e-05,
+      "loss": 1.449,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789491198932302e-05,
+      "loss": 1.4366,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779594341004336e-05,
+      "loss": 1.4463,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676968517907712e-05,
+      "loss": 1.458,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675976373113437e-05,
+      "loss": 1.4497,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 1.4867,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739884049248608e-05,
+      "loss": 1.4862,
+      "step": 690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729925851384385e-05,
+      "loss": 1.4601,
+      "step": 691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719955439662158e-05,
+      "loss": 1.4711,
+      "step": 692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709972832176797e-05,
+      "loss": 1.4748,
+      "step": 693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669997804704532e-05,
+      "loss": 1.492,
+      "step": 694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668997110240684e-05,
+      "loss": 1.4713,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679952016422537e-05,
+      "loss": 1.4833,
+      "step": 696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669920807275622e-05,
+      "loss": 1.4848,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665987749317132e-05,
+      "loss": 1.5077,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649822092336815e-05,
+      "loss": 1.4744,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639754623021227e-05,
+      "loss": 1.455,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662967510349558e-05,
+      "loss": 1.4693,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661958355205277e-05,
+      "loss": 1.4909,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609479987007527e-05,
+      "loss": 1.4983,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599364426696385e-05,
+      "loss": 1.4741,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658923688947765e-05,
+      "loss": 1.4674,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657909739373135e-05,
+      "loss": 1.5006,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568945957859236e-05,
+      "loss": 1.4073,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6558782600284724e-05,
+      "loss": 1.478,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 1.4387,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653842019383028e-05,
+      "loss": 1.4319,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 1.467,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651801032218743e-05,
+      "loss": 1.4244,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507787633208173e-05,
+      "loss": 1.4474,
+      "step": 713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649755313352017e-05,
+      "loss": 1.4683,
+      "step": 714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648730684169758e-05,
+      "loss": 1.4559,
+      "step": 715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477048776335972e-05,
+      "loss": 1.4429,
+      "step": 716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646677895605227e-05,
+      "loss": 1.4464,
+      "step": 717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645649739948474e-05,
+      "loss": 1.4708,
+      "step": 718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644620412529294e-05,
+      "loss": 1.456,
+      "step": 719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435899152157706e-05,
+      "loss": 1.4702,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642558249878109e-05,
+      "loss": 1.4708,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415254183886356e-05,
+      "loss": 1.4153,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404914226217924e-05,
+      "loss": 1.4749,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394562644541342e-05,
+      "loss": 1.4503,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384199457643264e-05,
+      "loss": 1.478,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373824684331397e-05,
+      "loss": 1.4802,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363438343434483e-05,
+      "loss": 1.4289,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635304045380225e-05,
+      "loss": 1.4849,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6342631034305386e-05,
+      "loss": 1.4443,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633221010383551e-05,
+      "loss": 1.4706,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321777681305128e-05,
+      "loss": 1.4649,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63113337856476e-05,
+      "loss": 1.4251,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 1.4394,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.629041165078864e-05,
+      "loss": 1.4639,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279933449557905e-05,
+      "loss": 1.4727,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269443851141357e-05,
+      "loss": 1.4578,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258942874576117e-05,
+      "loss": 1.4702,
+      "step": 737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624843053891997e-05,
+      "loss": 1.4208,
+      "step": 738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237906863251313e-05,
+      "loss": 1.437,
+      "step": 739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622737186666911e-05,
+      "loss": 1.4597,
+      "step": 740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6216825568292885e-05,
+      "loss": 1.4573,
+      "step": 741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620626798726267e-05,
+      "loss": 1.4292,
+      "step": 742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195699142738976e-05,
+      "loss": 1.4402,
+      "step": 743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185119053902748e-05,
+      "loss": 1.447,
+      "step": 744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174527739955345e-05,
+      "loss": 1.5101,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163925220118494e-05,
+      "loss": 1.4307,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153311513634257e-05,
+      "loss": 1.466,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142686639765e-05,
+      "loss": 1.4651,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613205061779337e-05,
+      "loss": 1.4325,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6121403467022223e-05,
+      "loss": 1.461,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611074520677462e-05,
+      "loss": 1.4951,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61000758563938e-05,
+      "loss": 1.4639,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089395435243105e-05,
+      "loss": 1.4782,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078703962705983e-05,
+      "loss": 1.4276,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 1.4884,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057287941106484e-05,
+      "loss": 1.428,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6046563430911148e-05,
+      "loss": 1.4274,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6035827947063375e-05,
+      "loss": 1.4644,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025081509046545e-05,
+      "loss": 1.4729,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014324136363915e-05,
+      "loss": 1.4975,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003555848538586e-05,
+      "loss": 1.474,
+      "step": 761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599277666511347e-05,
+      "loss": 1.4805,
+      "step": 762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598198660565125e-05,
+      "loss": 1.4321,
+      "step": 763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597118568973435e-05,
+      "loss": 1.4763,
+      "step": 764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596037393696489e-05,
+      "loss": 1.4587,
+      "step": 765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949551366964675e-05,
+      "loss": 1.4667,
+      "step": 766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593871799937512e-05,
+      "loss": 1.4956,
+      "step": 767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927873853857248e-05,
+      "loss": 1.4835,
+      "step": 768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917018950091642e-05,
+      "loss": 1.466,
+      "step": 769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906153307778405e-05,
+      "loss": 1.4669,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895276946637138e-05,
+      "loss": 1.4334,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884389886406887e-05,
+      "loss": 1.4519,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587349214684611e-05,
+      "loss": 1.4772,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862583747732662e-05,
+      "loss": 1.4954,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851664708863738e-05,
+      "loss": 1.4561,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584073505005583e-05,
+      "loss": 1.4779,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 1.4495,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818843951985425e-05,
+      "loss": 1.4496,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807882552452156e-05,
+      "loss": 1.4842,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579691061243829e-05,
+      "loss": 1.4913,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785928151856345e-05,
+      "loss": 1.4653,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5774935190637924e-05,
+      "loss": 1.4686,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576393174873368e-05,
+      "loss": 1.4559,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57529178461133e-05,
+      "loss": 1.4531,
+      "step": 784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5741893502765452e-05,
+      "loss": 1.463,
+      "step": 785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573085873869775e-05,
+      "loss": 1.443,
+      "step": 786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571981357393671e-05,
+      "loss": 1.4838,
+      "step": 787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708758028527754e-05,
+      "loss": 1.4898,
+      "step": 788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569769212253511e-05,
+      "loss": 1.4804,
+      "step": 789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686615876041828e-05,
+      "loss": 1.4519,
+      "step": 790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567552930914972e-05,
+      "loss": 1.4605,
+      "step": 791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664432441979328e-05,
+      "loss": 1.4541,
+      "step": 792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653325294669884e-05,
+      "loss": 1.4347,
+      "step": 793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642207887379285e-05,
+      "loss": 1.4782,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563108024028404e-05,
+      "loss": 1.4529,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619942373579247e-05,
+      "loss": 1.4417,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608794307478546e-05,
+      "loss": 1.4251,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55976360622141e-05,
+      "loss": 1.4439,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 1.4629,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575289115214895e-05,
+      "loss": 1.4507,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556410045403667e-05,
+      "loss": 1.459,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552901694807683e-05,
+      "loss": 1.4471,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554169285785208e-05,
+      "loss": 1.4756,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553047396351232e-05,
+      "loss": 1.489,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519245032149083e-05,
+      "loss": 1.4525,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5508006084141293e-05,
+      "loss": 1.4207,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549675713988604e-05,
+      "loss": 1.4645,
+      "step": 807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485498219798555e-05,
+      "loss": 1.4323,
+      "step": 808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547422934431218e-05,
+      "loss": 1.4629,
+      "step": 809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462950533878318e-05,
+      "loss": 1.4372,
+      "step": 810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545166180896641e-05,
+      "loss": 1.4832,
+      "step": 811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440363190063874e-05,
+      "loss": 1.4516,
+      "step": 812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542905469767611e-05,
+      "loss": 1.4779,
+      "step": 813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417736352326412e-05,
+      "loss": 1.5046,
+      "step": 814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 1.4332,
+      "step": 815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539507018492383e-05,
+      "loss": 1.4691,
+      "step": 816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383722404006808e-05,
+      "loss": 1.4538,
+      "step": 817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372364852399518e-05,
+      "loss": 1.4502,
+      "step": 818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360997550714306e-05,
+      "loss": 1.4744,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349620519581205e-05,
+      "loss": 1.4713,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 1.436,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532683735157973e-05,
+      "loss": 1.4028,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315431256059558e-05,
+      "loss": 1.4586,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304015513787843e-05,
+      "loss": 1.4339,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529259014548253e-05,
+      "loss": 1.4808,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5281155171879033e-05,
+      "loss": 1.4945,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269710613730212e-05,
+      "loss": 1.4478,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52582564918063e-05,
+      "loss": 1.476,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246792826894906e-05,
+      "loss": 1.4955,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235319639800948e-05,
+      "loss": 1.4247,
+      "step": 830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223836951346634e-05,
+      "loss": 1.4823,
+      "step": 831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52123447823714e-05,
+      "loss": 1.4272,
+      "step": 832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200843153731905e-05,
+      "loss": 1.4587,
+      "step": 833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518933208630197e-05,
+      "loss": 1.4559,
+      "step": 834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517781160097254e-05,
+      "loss": 1.4662,
+      "step": 835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516628171865166e-05,
+      "loss": 1.4324,
+      "step": 836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154742460264426e-05,
+      "loss": 1.476,
+      "step": 837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143193846752952e-05,
+      "loss": 1.4496,
+      "step": 838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513163589907632e-05,
+      "loss": 1.4539,
+      "step": 839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120068638210573e-05,
+      "loss": 1.4293,
+      "step": 840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108492085148632e-05,
+      "loss": 1.472,
+      "step": 841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096906260900304e-05,
+      "loss": 1.4651,
+      "step": 842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 1.4603,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073706882967752e-05,
+      "loss": 1.4818,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062093371387097e-05,
+      "loss": 1.4606,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050470672827117e-05,
+      "loss": 1.4779,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038838808381355e-05,
+      "loss": 1.4874,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502719779915999e-05,
+      "loss": 1.4113,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015547666289798e-05,
+      "loss": 1.4374,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003888430914111e-05,
+      "loss": 1.4292,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992220114192786e-05,
+      "loss": 1.4292,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980542737302155e-05,
+      "loss": 1.4594,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968856321434997e-05,
+      "loss": 1.4484,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4957160887800494e-05,
+      "loss": 1.4373,
+      "step": 854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945456457624197e-05,
+      "loss": 1.4807,
+      "step": 855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933743052147975e-05,
+      "loss": 1.4404,
+      "step": 856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492202069263e-05,
+      "loss": 1.46,
+      "step": 857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491028940034468e-05,
+      "loss": 1.4469,
+      "step": 858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898549196582646e-05,
+      "loss": 1.4451,
+      "step": 859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886800102650696e-05,
+      "loss": 1.4524,
+      "step": 860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875042139871768e-05,
+      "loss": 1.4461,
+      "step": 861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863275329584886e-05,
+      "loss": 1.447,
+      "step": 862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851499693145137e-05,
+      "loss": 1.4469,
+      "step": 863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839715251923632e-05,
+      "loss": 1.4414,
+      "step": 864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 1.4285,
+      "step": 865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816120040699615e-05,
+      "loss": 1.4342,
+      "step": 866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480430931351906e-05,
+      "loss": 1.4509,
+      "step": 867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479248986720057e-05,
+      "loss": 1.4738,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780661723194757e-05,
+      "loss": 1.4563,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768824902968024e-05,
+      "loss": 1.521,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756979428002513e-05,
+      "loss": 1.4568,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745125319796079e-05,
+      "loss": 1.4406,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733262599862234e-05,
+      "loss": 1.4287,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721391289730135e-05,
+      "loss": 1.437,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709511410944523e-05,
+      "loss": 1.4509,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469762298506568e-05,
+      "loss": 1.4746,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685726033669412e-05,
+      "loss": 1.4444,
+      "step": 877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4673820578346993e-05,
+      "loss": 1.4404,
+      "step": 878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661906640705131e-05,
+      "loss": 1.4779,
+      "step": 879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649984242365926e-05,
+      "loss": 1.4599,
+      "step": 880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4638053404966836e-05,
+      "loss": 1.4744,
+      "step": 881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626114150160636e-05,
+      "loss": 1.4498,
+      "step": 882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461416649961537e-05,
+      "loss": 1.4404,
+      "step": 883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602210475014332e-05,
+      "loss": 1.4475,
+      "step": 884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590246098055995e-05,
+      "loss": 1.4309,
+      "step": 885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578273390454008e-05,
+      "loss": 1.4655,
+      "step": 886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 1.4709,
+      "step": 887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554303070249205e-05,
+      "loss": 1.4358,
+      "step": 888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454230550114911e-05,
+      "loss": 1.4988,
+      "step": 889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530299688410726e-05,
+      "loss": 1.4528,
+      "step": 890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45182856538229e-05,
+      "loss": 1.4629,
+      "step": 891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506263419189388e-05,
+      "loss": 1.4774,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4494233006328837e-05,
+      "loss": 1.4261,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482194437074744e-05,
+      "loss": 1.4343,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470147733275388e-05,
+      "loss": 1.4297,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458092916793826e-05,
+      "loss": 1.4154,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444603000950784e-05,
+      "loss": 1.4424,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433959033309888e-05,
+      "loss": 1.449,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442188001010707e-05,
+      "loss": 1.4517,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409792961821101e-05,
+      "loss": 1.4918,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397697910388248e-05,
+      "loss": 1.478,
+      "step": 901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4385594877759315e-05,
+      "loss": 1.4478,
+      "step": 902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373483885899582e-05,
+      "loss": 1.4515,
+      "step": 903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361364956788772e-05,
+      "loss": 1.4742,
+      "step": 904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349238112421025e-05,
+      "loss": 1.4386,
+      "step": 905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337103374804832e-05,
+      "loss": 1.4604,
+      "step": 906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324960765963019e-05,
+      "loss": 1.4778,
+      "step": 907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312810307932694e-05,
+      "loss": 1.4424,
+      "step": 908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 1.4604,
+      "step": 909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288485932526115e-05,
+      "loss": 1.441,
+      "step": 910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276312059295148e-05,
+      "loss": 1.4449,
+      "step": 911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264130425166145e-05,
+      "loss": 1.4723,
+      "step": 912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251941052247044e-05,
+      "loss": 1.4689,
+      "step": 913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239743962659822e-05,
+      "loss": 1.4505,
+      "step": 914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227539178540463e-05,
+      "loss": 1.4189,
+      "step": 915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215326722038911e-05,
+      "loss": 1.466,
+      "step": 916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420310661531904e-05,
+      "loss": 1.4542,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190878880558603e-05,
+      "loss": 1.4066,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178643539949197e-05,
+      "loss": 1.438,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166400615696232e-05,
+      "loss": 1.4295,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 1.4742,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141892105149993e-05,
+      "loss": 1.4895,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129626563336179e-05,
+      "loss": 1.4071,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117353526837643e-05,
+      "loss": 1.4448,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105073017928199e-05,
+      "loss": 1.4492,
+      "step": 925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.409278505889522e-05,
+      "loss": 1.4095,
+      "step": 926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080489672039608e-05,
+      "loss": 1.4226,
+      "step": 927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406818687967574e-05,
+      "loss": 1.4389,
+      "step": 928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405587670413143e-05,
+      "loss": 1.4541,
+      "step": 929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043559167747897e-05,
+      "loss": 1.4459,
+      "step": 930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 1.4642,
+      "step": 931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018902101894798e-05,
+      "loss": 1.4452,
+      "step": 932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006562617174292e-05,
+      "loss": 1.4317,
+      "step": 933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994215861112617e-05,
+      "loss": 1.4868,
+      "step": 934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398186185611738e-05,
+      "loss": 1.4224,
+      "step": 935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969500624609338e-05,
+      "loss": 1.4417,
+      "step": 936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957132189022373e-05,
+      "loss": 1.4421,
+      "step": 937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944756571803435e-05,
+      "loss": 1.4495,
+      "step": 938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932373795412503e-05,
+      "loss": 1.4646,
+      "step": 939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919983882322563e-05,
+      "loss": 1.4715,
+      "step": 940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907586855019538e-05,
+      "loss": 1.4487,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895182736002276e-05,
+      "loss": 1.4708,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882771547782476e-05,
+      "loss": 1.4655,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870353312884694e-05,
+      "loss": 1.4426,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385792805384625e-05,
+      "loss": 1.4322,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845495793217223e-05,
+      "loss": 1.4325,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38330565535604e-05,
+      "loss": 1.4002,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382061035745123e-05,
+      "loss": 1.4228,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808157227477788e-05,
+      "loss": 1.4595,
+      "step": 949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3795697186240733e-05,
+      "loss": 1.4879,
+      "step": 950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783230256353265e-05,
+      "loss": 1.4547,
+      "step": 951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770756460441098e-05,
+      "loss": 1.4835,
+      "step": 952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 1.4417,
+      "step": 953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745788361107715e-05,
+      "loss": 1.4346,
+      "step": 954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733294103000055e-05,
+      "loss": 1.4563,
+      "step": 955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37207930694947e-05,
+      "loss": 1.4432,
+      "step": 956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708285283279252e-05,
+      "loss": 1.4402,
+      "step": 957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695770767053565e-05,
+      "loss": 1.4544,
+      "step": 958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683249543529698e-05,
+      "loss": 1.448,
+      "step": 959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670721635431896e-05,
+      "loss": 1.4305,
+      "step": 960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658187065496533e-05,
+      "loss": 1.4435,
+      "step": 961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645645856472062e-05,
+      "loss": 1.4661,
+      "step": 962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633098031119e-05,
+      "loss": 1.4265,
+      "step": 963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620543612209861e-05,
+      "loss": 1.4269,
+      "step": 964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607982622529135e-05,
+      "loss": 1.4555,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595415084873224e-05,
+      "loss": 1.475,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582841022050424e-05,
+      "loss": 1.4067,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3570260456880873e-05,
+      "loss": 1.4449,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557673412196504e-05,
+      "loss": 1.4581,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545079910841007e-05,
+      "loss": 1.4327,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532479975669809e-05,
+      "loss": 1.4196,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519873629549991e-05,
+      "loss": 1.422,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507260895360274e-05,
+      "loss": 1.4365,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494641795990986e-05,
+      "loss": 1.4493,
+      "step": 974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 1.4469,
+      "step": 975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469384593332668e-05,
+      "loss": 1.4563,
+      "step": 976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456746535881872e-05,
+      "loss": 1.4547,
+      "step": 977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444102204927876e-05,
+      "loss": 1.4243,
+      "step": 978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431451623418344e-05,
+      "loss": 1.4522,
+      "step": 979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418794814312274e-05,
+      "loss": 1.447,
+      "step": 980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406131800579985e-05,
+      "loss": 1.4648,
+      "step": 981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393462605203041e-05,
+      "loss": 1.4489,
+      "step": 982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380787251174226e-05,
+      "loss": 1.4524,
+      "step": 983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368105761497507e-05,
+      "loss": 1.4381,
+      "step": 984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355418159187988e-05,
+      "loss": 1.434,
+      "step": 985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342724467271852e-05,
+      "loss": 1.4361,
+      "step": 986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330024708786354e-05,
+      "loss": 1.4675,
+      "step": 987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317318906779744e-05,
+      "loss": 1.4151,
+      "step": 988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304607084311246e-05,
+      "loss": 1.4257,
+      "step": 989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291889264451005e-05,
+      "loss": 1.4524,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279165470280067e-05,
+      "loss": 1.4706,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266435724890294e-05,
+      "loss": 1.3989,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253700051384371e-05,
+      "loss": 1.4371,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240958472875734e-05,
+      "loss": 1.467,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228211012488532e-05,
+      "loss": 1.4522,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215457693357592e-05,
+      "loss": 1.4401,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 1.4379,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189933571456937e-05,
+      "loss": 1.4209,
+      "step": 998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317716281500987e-05,
+      "loss": 1.4775,
+      "step": 999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316438629246428e-05,
+      "loss": 1.4497,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151604027007744e-05,
+      "loss": 1.407,
+      "step": 1001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138816041838251e-05,
+      "loss": 1.381,
+      "step": 1002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126022360164172e-05,
+      "loss": 1.4328,
+      "step": 1003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311322300520423e-05,
+      "loss": 1.434,
+      "step": 1004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310041800018742e-05,
+      "loss": 1.4502,
+      "step": 1005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087607368353015e-05,
+      "loss": 1.4059,
+      "step": 1006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074791132950484e-05,
+      "loss": 1.4649,
+      "step": 1007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306196931723947e-05,
+      "loss": 1.4077,
+      "step": 1008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304914194448975e-05,
+      "loss": 1.4146,
+      "step": 1009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036309037981171e-05,
+      "loss": 1.4213,
+      "step": 1010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023470621003641e-05,
+      "loss": 1.4146,
+      "step": 1011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3010626716857065e-05,
+      "loss": 1.4362,
+      "step": 1012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997777348851288e-05,
+      "loss": 1.4423,
+      "step": 1013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984922540306095e-05,
+      "loss": 1.4001,
+      "step": 1014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297206231455113e-05,
+      "loss": 1.4067,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959196694925878e-05,
+      "loss": 1.4175,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946325704779602e-05,
+      "loss": 1.461,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933449367471322e-05,
+      "loss": 1.4306,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 1.4215,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907680744853295e-05,
+      "loss": 1.4298,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289478850630993e-05,
+      "loss": 1.4484,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881891014137242e-05,
+      "loss": 1.4059,
+      "step": 1022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868988291742349e-05,
+      "loss": 1.4266,
+      "step": 1023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856080362541854e-05,
+      "loss": 1.447,
+      "step": 1024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284316724996181e-05,
+      "loss": 1.3914,
+      "step": 1025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830248977437685e-05,
+      "loss": 1.4234,
+      "step": 1026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 1.4268,
+      "step": 1027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804397046345805e-05,
+      "loss": 1.4197,
+      "step": 1028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279146343469563e-05,
+      "loss": 1.4388,
+      "step": 1029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778524756936438e-05,
+      "loss": 1.3809,
+      "step": 1030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765581036550095e-05,
+      "loss": 1.4651,
+      "step": 1031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752632297027605e-05,
+      "loss": 1.4117,
+      "step": 1032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273967856186909e-05,
+      "loss": 1.4079,
+      "step": 1033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2726719854583736e-05,
+      "loss": 1.4042,
+      "step": 1034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713756198689757e-05,
+      "loss": 1.4108,
+      "step": 1035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700787617714338e-05,
+      "loss": 1.4023,
+      "step": 1036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687814135193613e-05,
+      "loss": 1.437,
+      "step": 1037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674835774672603e-05,
+      "loss": 1.4315,
+      "step": 1038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661852559705192e-05,
+      "loss": 1.451,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648864513854067e-05,
+      "loss": 1.3935,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 1.3912,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262287402379521e-05,
+      "loss": 1.4047,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609871626756522e-05,
+      "loss": 1.4245,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596864493172119e-05,
+      "loss": 1.4104,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583852646648097e-05,
+      "loss": 1.4245,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570836110799101e-05,
+      "loss": 1.4174,
+      "step": 1046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557814909248297e-05,
+      "loss": 1.4157,
+      "step": 1047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544789065627314e-05,
+      "loss": 1.451,
+      "step": 1048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25317586035762e-05,
+      "loss": 1.4096,
+      "step": 1049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518723546743395e-05,
+      "loss": 1.4027,
+      "step": 1050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505683918785669e-05,
+      "loss": 1.4069,
+      "step": 1051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492639743368096e-05,
+      "loss": 1.4343,
+      "step": 1052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479591044163997e-05,
+      "loss": 1.4367,
+      "step": 1053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24665378448549e-05,
+      "loss": 1.4055,
+      "step": 1054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245348016913051e-05,
+      "loss": 1.4018,
+      "step": 1055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440418040688654e-05,
+      "loss": 1.4459,
+      "step": 1056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427351483235224e-05,
+      "loss": 1.4325,
+      "step": 1057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414280520484178e-05,
+      "loss": 1.4306,
+      "step": 1058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401205176157447e-05,
+      "loss": 1.4192,
+      "step": 1059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388125473984918e-05,
+      "loss": 1.415,
+      "step": 1060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375041437704394e-05,
+      "loss": 1.4515,
+      "step": 1061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236195309106154e-05,
+      "loss": 1.4482,
+      "step": 1062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 1.4425,
+      "step": 1063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335763561710562e-05,
+      "loss": 1.4142,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232266242653271e-05,
+      "loss": 1.419,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309557076052978e-05,
+      "loss": 1.4313,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296447534055715e-05,
+      "loss": 1.4579,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283333824332874e-05,
+      "loss": 1.4193,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270215970683977e-05,
+      "loss": 1.4407,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257093996916057e-05,
+      "loss": 1.4122,
+      "step": 1070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243967926843627e-05,
+      "loss": 1.4135,
+      "step": 1071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230837784288646e-05,
+      "loss": 1.4015,
+      "step": 1072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217703593080445e-05,
+      "loss": 1.4359,
+      "step": 1073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220456537705572e-05,
+      "loss": 1.4253,
+      "step": 1074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191423160058462e-05,
+      "loss": 1.4329,
+      "step": 1075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178276965939923e-05,
+      "loss": 1.4283,
+      "step": 1076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165126818558572e-05,
+      "loss": 1.4164,
+      "step": 1077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151972741780067e-05,
+      "loss": 1.4383,
+      "step": 1078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138814759477176e-05,
+      "loss": 1.396,
+      "step": 1079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125652895529766e-05,
+      "loss": 1.4212,
+      "step": 1080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112487173824755e-05,
+      "loss": 1.437,
+      "step": 1081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099317618256046e-05,
+      "loss": 1.3963,
+      "step": 1082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086144252724514e-05,
+      "loss": 1.4221,
+      "step": 1083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072967101137942e-05,
+      "loss": 1.4141,
+      "step": 1084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 1.3936,
+      "step": 1085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046601535465127e-05,
+      "loss": 1.3989,
+      "step": 1086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033413169228636e-05,
+      "loss": 1.4204,
+      "step": 1087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020221112636518e-05,
+      "loss": 1.3955,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007025389630484e-05,
+      "loss": 1.432,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993826024158887e-05,
+      "loss": 1.4429,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980623040176704e-05,
+      "loss": 1.4459,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196741646164547e-05,
+      "loss": 1.3941,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954206312533246e-05,
+      "loss": 1.449,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940992616814572e-05,
+      "loss": 1.4134,
+      "step": 1094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927775398470432e-05,
+      "loss": 1.4234,
+      "step": 1095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914554681488188e-05,
+      "loss": 1.4373,
+      "step": 1096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901330489861564e-05,
+      "loss": 1.4292,
+      "step": 1097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888102847590592e-05,
+      "loss": 1.436,
+      "step": 1098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1874871778681555e-05,
+      "loss": 1.4215,
+      "step": 1099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861637307146962e-05,
+      "loss": 1.4392,
+      "step": 1100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848399457005496e-05,
+      "loss": 1.4464,
+      "step": 1101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835158252281973e-05,
+      "loss": 1.3893,
+      "step": 1102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821913717007298e-05,
+      "loss": 1.4257,
+      "step": 1103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808665875218408e-05,
+      "loss": 1.4521,
+      "step": 1104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795414750958265e-05,
+      "loss": 1.4263,
+      "step": 1105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782160368275768e-05,
+      "loss": 1.4178,
+      "step": 1106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 1.4048,
+      "step": 1107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175564192386885e-05,
+      "loss": 1.4294,
+      "step": 1108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742377910271638e-05,
+      "loss": 1.423,
+      "step": 1109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729110734506396e-05,
+      "loss": 1.4175,
+      "step": 1110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715840420651153e-05,
+      "loss": 1.4191,
+      "step": 1111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702566992789646e-05,
+      "loss": 1.4478,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689290475011258e-05,
+      "loss": 1.405,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167601089141098e-05,
+      "loss": 1.3946,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662728266089365e-05,
+      "loss": 1.3863,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649442623152492e-05,
+      "loss": 1.4207,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636153986711906e-05,
+      "loss": 1.4515,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16228623808846e-05,
+      "loss": 1.404,
+      "step": 1118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609567829792945e-05,
+      "loss": 1.4337,
+      "step": 1119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596270357564656e-05,
+      "loss": 1.3824,
+      "step": 1120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582969988332757e-05,
+      "loss": 1.4093,
+      "step": 1121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569666746235527e-05,
+      "loss": 1.4239,
+      "step": 1122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556360655416457e-05,
+      "loss": 1.4214,
+      "step": 1123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543051740024204e-05,
+      "loss": 1.4275,
+      "step": 1124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529740024212566e-05,
+      "loss": 1.4607,
+      "step": 1125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516425532140403e-05,
+      "loss": 1.4283,
+      "step": 1126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503108287971627e-05,
+      "loss": 1.4132,
+      "step": 1127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148978831587515e-05,
+      "loss": 1.4139,
+      "step": 1128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 1.4361,
+      "step": 1129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1463140284599387e-05,
+      "loss": 1.4022,
+      "step": 1130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449812273782492e-05,
+      "loss": 1.4324,
+      "step": 1131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436481631762574e-05,
+      "loss": 1.3998,
+      "step": 1132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 1.4282,
+      "step": 1133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1409812550891272e-05,
+      "loss": 1.443,
+      "step": 1134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396474160440479e-05,
+      "loss": 1.4497,
+      "step": 1135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383133235587748e-05,
+      "loss": 1.4003,
+      "step": 1136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136978980054496e-05,
+      "loss": 1.4176,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356443879528554e-05,
+      "loss": 1.4364,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343095496759478e-05,
+      "loss": 1.412,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329744676463144e-05,
+      "loss": 1.4106,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316391442869394e-05,
+      "loss": 1.4241,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130303582021245e-05,
+      "loss": 1.3977,
+      "step": 1142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289677832730864e-05,
+      "loss": 1.4223,
+      "step": 1143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127631750466748e-05,
+      "loss": 1.4031,
+      "step": 1144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12629548602694e-05,
+      "loss": 1.3993,
+      "step": 1145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249589923787918e-05,
+      "loss": 1.3981,
+      "step": 1146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123622271947849e-05,
+      "loss": 1.4336,
+      "step": 1147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222853271600698e-05,
+      "loss": 1.4212,
+      "step": 1148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209481604418182e-05,
+      "loss": 1.4412,
+      "step": 1149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196107742198617e-05,
+      "loss": 1.413,
+      "step": 1150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 1.4033,
+      "step": 1151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116935352973891e-05,
+      "loss": 1.408,
+      "step": 1152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155973228053854e-05,
+      "loss": 1.3923,
+      "step": 1153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142590828441841e-05,
+      "loss": 1.3946,
+      "step": 1154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129206355190025e-05,
+      "loss": 1.4623,
+      "step": 1155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1115819832589314e-05,
+      "loss": 1.4357,
+      "step": 1156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102431284934345e-05,
+      "loss": 1.411,
+      "step": 1157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089040736523431e-05,
+      "loss": 1.4353,
+      "step": 1158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075648211658505e-05,
+      "loss": 1.4342,
+      "step": 1159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062253734645103e-05,
+      "loss": 1.389,
+      "step": 1160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048857329792284e-05,
+      "loss": 1.4411,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035459021412631e-05,
+      "loss": 1.4033,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022058833822158e-05,
+      "loss": 1.4379,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008656791340297e-05,
+      "loss": 1.3831,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099525291828986e-05,
+      "loss": 1.4047,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098184723899696e-05,
+      "loss": 1.3926,
+      "step": 1166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968439777791e-05,
+      "loss": 1.4389,
+      "step": 1167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955030559004616e-05,
+      "loss": 1.4335,
+      "step": 1168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941619606973633e-05,
+      "loss": 1.4063,
+      "step": 1169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928206946037019e-05,
+      "loss": 1.4208,
+      "step": 1170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914792600536843e-05,
+      "loss": 1.4029,
+      "step": 1171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090137659481824e-05,
+      "loss": 1.4178,
+      "step": 1172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 1.4212,
+      "step": 1173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087453970012128e-05,
+      "loss": 1.4422,
+      "step": 1174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861118859848067e-05,
+      "loss": 1.4171,
+      "step": 1175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847696456766632e-05,
+      "loss": 1.3952,
+      "step": 1176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083427251523672e-05,
+      "loss": 1.4057,
+      "step": 1177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820847059620879e-05,
+      "loss": 1.4514,
+      "step": 1178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08074201142844e-05,
+      "loss": 1.4041,
+      "step": 1179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793991703595279e-05,
+      "loss": 1.4302,
+      "step": 1180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780561851924168e-05,
+      "loss": 1.43,
+      "step": 1181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767130583644343e-05,
+      "loss": 1.4388,
+      "step": 1182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753697923131641e-05,
+      "loss": 1.4058,
+      "step": 1183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074026389476443e-05,
+      "loss": 1.4527,
+      "step": 1184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726828522923563e-05,
+      "loss": 1.441,
+      "step": 1185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0713391831992324e-05,
+      "loss": 1.3875,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06999538463564e-05,
+      "loss": 1.4054,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686514590403823e-05,
+      "loss": 1.4053,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673074088524926e-05,
+      "loss": 1.4269,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659632365112317e-05,
+      "loss": 1.3971,
+      "step": 1190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646189444560799e-05,
+      "loss": 1.4082,
+      "step": 1191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063274535126737e-05,
+      "loss": 1.3997,
+      "step": 1192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619300109631146e-05,
+      "loss": 1.4311,
+      "step": 1193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605853744053322e-05,
+      "loss": 1.4183,
+      "step": 1194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 1.4258,
+      "step": 1195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0578957738687847e-05,
+      "loss": 1.4312,
+      "step": 1196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565508147712618e-05,
+      "loss": 1.4241,
+      "step": 1197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055205753042055e-05,
+      "loss": 1.4418,
+      "step": 1198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538605911222603e-05,
+      "loss": 1.3553,
+      "step": 1199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525153314531548e-05,
+      "loss": 1.3915,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511699764761935e-05,
+      "loss": 1.4223,
+      "step": 1201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498245286330049e-05,
+      "loss": 1.4184,
+      "step": 1202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484789903653847e-05,
+      "loss": 1.4126,
+      "step": 1203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471333641152937e-05,
+      "loss": 1.4273,
+      "step": 1204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457876523248518e-05,
+      "loss": 1.4198,
+      "step": 1205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444418574363354e-05,
+      "loss": 1.3986,
+      "step": 1206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430959818921696e-05,
+      "loss": 1.4379,
+      "step": 1207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417500281349278e-05,
+      "loss": 1.4078,
+      "step": 1208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404039986073244e-05,
+      "loss": 1.455,
+      "step": 1209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390578957522117e-05,
+      "loss": 1.3984,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037711722012574e-05,
+      "loss": 1.4023,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363654798315264e-05,
+      "loss": 1.4335,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035019171652306e-05,
+      "loss": 1.4212,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033672799918271e-05,
+      "loss": 1.421,
+      "step": 1214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323263670728947e-05,
+      "loss": 1.428,
+      "step": 1215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309798755597606e-05,
+      "loss": 1.3919,
+      "step": 1216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 1.4459,
+      "step": 1217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282867263050848e-05,
+      "loss": 1.4202,
+      "step": 1218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269400734512256e-05,
+      "loss": 1.4651,
+      "step": 1219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255933717049658e-05,
+      "loss": 1.4357,
+      "step": 1220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024246623510377e-05,
+      "loss": 1.3973,
+      "step": 1221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0228998313116159e-05,
+      "loss": 1.4307,
+      "step": 1222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021552997552919e-05,
+      "loss": 1.4062,
+      "step": 1223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202061246785975e-05,
+      "loss": 1.4183,
+      "step": 1224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188592151330343e-05,
+      "loss": 1.4574,
+      "step": 1225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175122713606788e-05,
+      "loss": 1.3936,
+      "step": 1226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161652958060417e-05,
+      "loss": 1.4288,
+      "step": 1227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148182909136928e-05,
+      "loss": 1.4241,
+      "step": 1228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134712591282539e-05,
+      "loss": 1.4184,
+      "step": 1229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0121242028943958e-05,
+      "loss": 1.4038,
+      "step": 1230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107771246568346e-05,
+      "loss": 1.3999,
+      "step": 1231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094300268603254e-05,
+      "loss": 1.3925,
+      "step": 1232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080829119496587e-05,
+      "loss": 1.4496,
+      "step": 1233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067357823696566e-05,
+      "loss": 1.4145,
+      "step": 1234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005388640565168e-05,
+      "loss": 1.4367,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040414889810635e-05,
+      "loss": 1.425,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026943300622313e-05,
+      "loss": 1.4083,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001347166253573e-05,
+      "loss": 1.375,
+      "step": 1238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.4318,
+      "step": 1239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98652833746427e-06,
+      "loss": 1.4145,
+      "step": 1240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973056699377692e-06,
+      "loss": 1.4362,
+      "step": 1241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959585110189369e-06,
+      "loss": 1.3904,
+      "step": 1242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946113594348321e-06,
+      "loss": 1.3959,
+      "step": 1243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932642176303435e-06,
+      "loss": 1.4176,
+      "step": 1244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919170880503416e-06,
+      "loss": 1.4346,
+      "step": 1245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90569973139675e-06,
+      "loss": 1.4001,
+      "step": 1246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892228753431658e-06,
+      "loss": 1.4142,
+      "step": 1247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.878757971056043e-06,
+      "loss": 1.4019,
+      "step": 1248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865287408717464e-06,
+      "loss": 1.4017,
+      "step": 1249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851817090863073e-06,
+      "loss": 1.4535,
+      "step": 1250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838347041939585e-06,
+      "loss": 1.454,
+      "step": 1251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824877286393215e-06,
+      "loss": 1.3985,
+      "step": 1252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811407848669657e-06,
+      "loss": 1.4006,
+      "step": 1253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797938753214028e-06,
+      "loss": 1.4492,
+      "step": 1254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784470024470813e-06,
+      "loss": 1.4241,
+      "step": 1255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.771001686883841e-06,
+      "loss": 1.3917,
+      "step": 1256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757533764896235e-06,
+      "loss": 1.4106,
+      "step": 1257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744066282950345e-06,
+      "loss": 1.4246,
+      "step": 1258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730599265487746e-06,
+      "loss": 1.4046,
+      "step": 1259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717132736949157e-06,
+      "loss": 1.424,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 1.4006,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690201244402394e-06,
+      "loss": 1.4411,
+      "step": 1262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67673632927106e-06,
+      "loss": 1.4392,
+      "step": 1263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663272000817295e-06,
+      "loss": 1.4477,
+      "step": 1264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64980828347694e-06,
+      "loss": 1.3991,
+      "step": 1265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636345201684741e-06,
+      "loss": 1.4403,
+      "step": 1266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622882779874263e-06,
+      "loss": 1.4469,
+      "step": 1267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609421042477886e-06,
+      "loss": 1.4193,
+      "step": 1268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595960013926761e-06,
+      "loss": 1.4343,
+      "step": 1269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582499718650726e-06,
+      "loss": 1.4733,
+      "step": 1270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569040181078306e-06,
+      "loss": 1.3946,
+      "step": 1271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555581425636648e-06,
+      "loss": 1.3979,
+      "step": 1272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542123476751484e-06,
+      "loss": 1.4137,
+      "step": 1273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528666358847068e-06,
+      "loss": 1.4157,
+      "step": 1274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515210096346155e-06,
+      "loss": 1.4126,
+      "step": 1275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501754713669956e-06,
+      "loss": 1.3915,
+      "step": 1276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488300235238067e-06,
+      "loss": 1.4013,
+      "step": 1277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474846685468455e-06,
+      "loss": 1.4317,
+      "step": 1278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461394088777402e-06,
+      "loss": 1.4509,
+      "step": 1279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447942469579453e-06,
+      "loss": 1.4121,
+      "step": 1280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434491852287385e-06,
+      "loss": 1.4249,
+      "step": 1281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.421042261312156e-06,
+      "loss": 1.4044,
+      "step": 1282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 1.4026,
+      "step": 1283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.39414625594668e-06,
+      "loss": 1.3995,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38069989036886e-06,
+      "loss": 1.4156,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367254648732633e-06,
+      "loss": 1.4052,
+      "step": 1286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.353810555439203e-06,
+      "loss": 1.4294,
+      "step": 1287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340367634887688e-06,
+      "loss": 1.4152,
+      "step": 1288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326925911475075e-06,
+      "loss": 1.4253,
+      "step": 1289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31348540959618e-06,
+      "loss": 1.4066,
+      "step": 1290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300046153643603e-06,
+      "loss": 1.4084,
+      "step": 1291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286608168007678e-06,
+      "loss": 1.4732,
+      "step": 1292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27317147707644e-06,
+      "loss": 1.4182,
+      "step": 1293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259736105235572e-06,
+      "loss": 1.419,
+      "step": 1294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246302076868362e-06,
+      "loss": 1.4648,
+      "step": 1295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23286941635566e-06,
+      "loss": 1.4098,
+      "step": 1296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219438148075834e-06,
+      "loss": 1.459,
+      "step": 1297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206008296404724e-06,
+      "loss": 1.3825,
+      "step": 1298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192579885715602e-06,
+      "loss": 1.4047,
+      "step": 1299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179152940379123e-06,
+      "loss": 1.4363,
+      "step": 1300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165727484763283e-06,
+      "loss": 1.4092,
+      "step": 1301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15230354323337e-06,
+      "loss": 1.4519,
+      "step": 1302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138881140151931e-06,
+      "loss": 1.4394,
+      "step": 1303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.125460299878722e-06,
+      "loss": 1.4161,
+      "step": 1304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 1.4128,
+      "step": 1305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098623405181762e-06,
+      "loss": 1.4007,
+      "step": 1306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085207399463162e-06,
+      "loss": 1.4247,
+      "step": 1307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071793053962984e-06,
+      "loss": 1.3969,
+      "step": 1308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058380393026369e-06,
+      "loss": 1.4032,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044969440995389e-06,
+      "loss": 1.4219,
+      "step": 1310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.031560222209001e-06,
+      "loss": 1.4192,
+      "step": 1311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018152761003041e-06,
+      "loss": 1.3991,
+      "step": 1312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00474708171014e-06,
+      "loss": 1.4199,
+      "step": 1313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991343208659705e-06,
+      "loss": 1.4035,
+      "step": 1314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977941166177845e-06,
+      "loss": 1.4203,
+      "step": 1315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96454097858737e-06,
+      "loss": 1.4311,
+      "step": 1316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951142670207718e-06,
+      "loss": 1.446,
+      "step": 1317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9377462653549e-06,
+      "loss": 1.4314,
+      "step": 1318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924351788341495e-06,
+      "loss": 1.4253,
+      "step": 1319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910959263476575e-06,
+      "loss": 1.4308,
+      "step": 1320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897568715065658e-06,
+      "loss": 1.4245,
+      "step": 1321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884180167410687e-06,
+      "loss": 1.423,
+      "step": 1322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870793644809981e-06,
+      "loss": 1.4273,
+      "step": 1323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85740917155816e-06,
+      "loss": 1.4162,
+      "step": 1324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844026771946148e-06,
+      "loss": 1.4745,
+      "step": 1325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830646470261097e-06,
+      "loss": 1.4202,
+      "step": 1326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 1.3936,
+      "step": 1327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803892257801386e-06,
+      "loss": 1.4317,
+      "step": 1328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790518395581823e-06,
+      "loss": 1.4122,
+      "step": 1329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777146728399307e-06,
+      "loss": 1.4267,
+      "step": 1330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763777280521511e-06,
+      "loss": 1.4059,
+      "step": 1331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750410076212088e-06,
+      "loss": 1.4436,
+      "step": 1332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737045139730605e-06,
+      "loss": 1.3983,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723682495332523e-06,
+      "loss": 1.4219,
+      "step": 1334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.710322167269141e-06,
+      "loss": 1.4249,
+      "step": 1335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696964179787554e-06,
+      "loss": 1.4323,
+      "step": 1336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683608557130608e-06,
+      "loss": 1.4286,
+      "step": 1337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670255323536858e-06,
+      "loss": 1.4084,
+      "step": 1338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656904503240527e-06,
+      "loss": 1.458,
+      "step": 1339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643556120471449e-06,
+      "loss": 1.4302,
+      "step": 1340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63021019945504e-06,
+      "loss": 1.4034,
+      "step": 1341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616866764412257e-06,
+      "loss": 1.448,
+      "step": 1342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603525839559523e-06,
+      "loss": 1.3934,
+      "step": 1343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590187449108726e-06,
+      "loss": 1.3836,
+      "step": 1344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 1.4482,
+      "step": 1345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563518368237428e-06,
+      "loss": 1.4005,
+      "step": 1346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550187726217508e-06,
+      "loss": 1.3962,
+      "step": 1347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536859715400616e-06,
+      "loss": 1.4253,
+      "step": 1348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 1.4178,
+      "step": 1349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510211684124854e-06,
+      "loss": 1.4154,
+      "step": 1350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496891712028374e-06,
+      "loss": 1.4138,
+      "step": 1351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4835744678596e-06,
+      "loss": 1.4069,
+      "step": 1352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470259975787438e-06,
+      "loss": 1.4475,
+      "step": 1353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456948259975796e-06,
+      "loss": 1.4544,
+      "step": 1354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443639344583547e-06,
+      "loss": 1.3887,
+      "step": 1355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430333253764474e-06,
+      "loss": 1.4336,
+      "step": 1356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417030011667241e-06,
+      "loss": 1.4158,
+      "step": 1357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403729642435346e-06,
+      "loss": 1.4223,
+      "step": 1358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.390432170207059e-06,
+      "loss": 1.4075,
+      "step": 1359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377137619115401e-06,
+      "loss": 1.4417,
+      "step": 1360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363846013288096e-06,
+      "loss": 1.4264,
+      "step": 1361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350557376847513e-06,
+      "loss": 1.427,
+      "step": 1362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337271733910636e-06,
+      "loss": 1.4334,
+      "step": 1363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323989108589026e-06,
+      "loss": 1.4321,
+      "step": 1364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310709524988743e-06,
+      "loss": 1.4175,
+      "step": 1365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297433007210354e-06,
+      "loss": 1.4535,
+      "step": 1366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284159579348852e-06,
+      "loss": 1.4143,
+      "step": 1367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270889265493608e-06,
+      "loss": 1.404,
+      "step": 1368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257622089728362e-06,
+      "loss": 1.4321,
+      "step": 1369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244358076131154e-06,
+      "loss": 1.4055,
+      "step": 1370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 1.4155,
+      "step": 1371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217839631724236e-06,
+      "loss": 1.394,
+      "step": 1372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20458524904174e-06,
+      "loss": 1.3852,
+      "step": 1373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191334124781594e-06,
+      "loss": 1.4133,
+      "step": 1374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178086282992705e-06,
+      "loss": 1.4234,
+      "step": 1375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164841747718027e-06,
+      "loss": 1.4057,
+      "step": 1376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151600542994506e-06,
+      "loss": 1.4258,
+      "step": 1377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138362692853041e-06,
+      "loss": 1.4014,
+      "step": 1378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125128221318446e-06,
+      "loss": 1.4386,
+      "step": 1379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111897152409411e-06,
+      "loss": 1.4279,
+      "step": 1380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098669510138438e-06,
+      "loss": 1.4169,
+      "step": 1381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085445318511814e-06,
+      "loss": 1.3922,
+      "step": 1382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.072224601529574e-06,
+      "loss": 1.4313,
+      "step": 1383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059007383185431e-06,
+      "loss": 1.3947,
+      "step": 1384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045793687466757e-06,
+      "loss": 1.4506,
+      "step": 1385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032583538354534e-06,
+      "loss": 1.4282,
+      "step": 1386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0193769598233e-06,
+      "loss": 1.4262,
+      "step": 1387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006173975841115e-06,
+      "loss": 1.4244,
+      "step": 1388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992974610369521e-06,
+      "loss": 1.4164,
+      "step": 1389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979778887363485e-06,
+      "loss": 1.4111,
+      "step": 1390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966586830771367e-06,
+      "loss": 1.4142,
+      "step": 1391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953398464534878e-06,
+      "loss": 1.4271,
+      "step": 1392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 1.4247,
+      "step": 1393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927032898862061e-06,
+      "loss": 1.4244,
+      "step": 1394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91385574727549e-06,
+      "loss": 1.3996,
+      "step": 1395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900682381743956e-06,
+      "loss": 1.4436,
+      "step": 1396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887512826175247e-06,
+      "loss": 1.4222,
+      "step": 1397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874347104470234e-06,
+      "loss": 1.4079,
+      "step": 1398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861185240522828e-06,
+      "loss": 1.4044,
+      "step": 1399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848027258219938e-06,
+      "loss": 1.4518,
+      "step": 1400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834873181441426e-06,
+      "loss": 1.4517,
+      "step": 1401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821723034060082e-06,
+      "loss": 1.4276,
+      "step": 1402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808576839941541e-06,
+      "loss": 1.3862,
+      "step": 1403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795434622944282e-06,
+      "loss": 1.4251,
+      "step": 1404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782296406919557e-06,
+      "loss": 1.4127,
+      "step": 1405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769162215711356e-06,
+      "loss": 1.4599,
+      "step": 1406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.756032073156373e-06,
+      "loss": 1.3929,
+      "step": 1407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742906003083947e-06,
+      "loss": 1.4044,
+      "step": 1408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729784029316025e-06,
+      "loss": 1.4099,
+      "step": 1409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716666175667126e-06,
+      "loss": 1.4117,
+      "step": 1410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703552465944287e-06,
+      "loss": 1.4115,
+      "step": 1411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690442923947025e-06,
+      "loss": 1.4088,
+      "step": 1412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677337573467294e-06,
+      "loss": 1.4188,
+      "step": 1413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664236438289445e-06,
+      "loss": 1.3934,
+      "step": 1414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 1.4192,
+      "step": 1415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638046908938463e-06,
+      "loss": 1.4299,
+      "step": 1416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624958562295607e-06,
+      "loss": 1.4307,
+      "step": 1417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611874526015084e-06,
+      "loss": 1.4375,
+      "step": 1418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598794823842557e-06,
+      "loss": 1.3661,
+      "step": 1419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585719479515823e-06,
+      "loss": 1.3949,
+      "step": 1420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572648516764778e-06,
+      "loss": 1.437,
+      "step": 1421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5595819593113506e-06,
+      "loss": 1.3941,
+      "step": 1422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54651983086949e-06,
+      "loss": 1.4274,
+      "step": 1423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533462155145105e-06,
+      "loss": 1.4466,
+      "step": 1424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5204089558360076e-06,
+      "loss": 1.4003,
+      "step": 1425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5073602566319045e-06,
+      "loss": 1.4143,
+      "step": 1426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494316081214335e-06,
+      "loss": 1.443,
+      "step": 1427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481276453256609e-06,
+      "loss": 1.4122,
+      "step": 1428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468241396423801e-06,
+      "loss": 1.4183,
+      "step": 1429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455210934372692e-06,
+      "loss": 1.4174,
+      "step": 1430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.442185090751706e-06,
+      "loss": 1.4086,
+      "step": 1431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429163889200902e-06,
+      "loss": 1.4049,
+      "step": 1432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416147353351909e-06,
+      "loss": 1.3985,
+      "step": 1433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403135506827883e-06,
+      "loss": 1.4196,
+      "step": 1434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390128373243479e-06,
+      "loss": 1.4092,
+      "step": 1435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377125976204794e-06,
+      "loss": 1.4178,
+      "step": 1436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 1.434,
+      "step": 1437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351135486145937e-06,
+      "loss": 1.405,
+      "step": 1438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338147440294809e-06,
+      "loss": 1.4009,
+      "step": 1439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325164225327399e-06,
+      "loss": 1.4522,
+      "step": 1440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312185864806391e-06,
+      "loss": 1.4351,
+      "step": 1441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.299212382285665e-06,
+      "loss": 1.4419,
+      "step": 1442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2862438013102484e-06,
+      "loss": 1.4154,
+      "step": 1443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273280145416267e-06,
+      "loss": 1.4641,
+      "step": 1444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260321438130913e-06,
+      "loss": 1.4295,
+      "step": 1445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247367702972399e-06,
+      "loss": 1.4268,
+      "step": 1446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2344189634499075e-06,
+      "loss": 1.4351,
+      "step": 1447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2214752430635625e-06,
+      "loss": 1.4384,
+      "step": 1448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208536565304374e-06,
+      "loss": 1.4276,
+      "step": 1449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1956029536541985e-06,
+      "loss": 1.3848,
+      "step": 1450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 1.4191,
+      "step": 1451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1697510225623206e-06,
+      "loss": 1.4371,
+      "step": 1452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156832750038192e-06,
+      "loss": 1.4399,
+      "step": 1453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1439196374581486e-06,
+      "loss": 1.4363,
+      "step": 1454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1310117082576536e-06,
+      "loss": 1.4364,
+      "step": 1455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118108985862759e-06,
+      "loss": 1.4313,
+      "step": 1456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105211493690073e-06,
+      "loss": 1.4512,
+      "step": 1457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092319255146707e-06,
+      "loss": 1.4184,
+      "step": 1458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 1.4005,
+      "step": 1459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066550632528679e-06,
+      "loss": 1.429,
+      "step": 1460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053674295220399e-06,
+      "loss": 1.4177,
+      "step": 1461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040803305074126e-06,
+      "loss": 1.4053,
+      "step": 1462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027937685448871e-06,
+      "loss": 1.407,
+      "step": 1463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015077459693907e-06,
+      "loss": 1.3999,
+      "step": 1464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002222651148714e-06,
+      "loss": 1.4288,
+      "step": 1465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9893732831429395e-06,
+      "loss": 1.4121,
+      "step": 1466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976529378996357e-06,
+      "loss": 1.4025,
+      "step": 1467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963690962018832e-06,
+      "loss": 1.4332,
+      "step": 1468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950858055510254e-06,
+      "loss": 1.4276,
+      "step": 1469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938030682760532e-06,
+      "loss": 1.3853,
+      "step": 1470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925208867049522e-06,
+      "loss": 1.3989,
+      "step": 1471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912392631646987e-06,
+      "loss": 1.4663,
+      "step": 1472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89958199981258e-06,
+      "loss": 1.3965,
+      "step": 1473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8867769947957765e-06,
+      "loss": 1.4139,
+      "step": 1474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87397763983583e-06,
+      "loss": 1.4227,
+      "step": 1475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86118395816175e-06,
+      "loss": 1.4021,
+      "step": 1476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848395972992261e-06,
+      "loss": 1.4041,
+      "step": 1477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835613707535722e-06,
+      "loss": 1.3894,
+      "step": 1478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822837184990132e-06,
+      "loss": 1.4194,
+      "step": 1479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810066428543064e-06,
+      "loss": 1.4118,
+      "step": 1480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 1.4181,
+      "step": 1481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78454230664241e-06,
+      "loss": 1.4109,
+      "step": 1482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77178898751147e-06,
+      "loss": 1.427,
+      "step": 1483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75904152712427e-06,
+      "loss": 1.4102,
+      "step": 1484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7462999486156315e-06,
+      "loss": 1.38,
+      "step": 1485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733564275109707e-06,
+      "loss": 1.4179,
+      "step": 1486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720834529719939e-06,
+      "loss": 1.4111,
+      "step": 1487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708110735548996e-06,
+      "loss": 1.4182,
+      "step": 1488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695392915688759e-06,
+      "loss": 1.4095,
+      "step": 1489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682681093220263e-06,
+      "loss": 1.3986,
+      "step": 1490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6699752912136504e-06,
+      "loss": 1.4342,
+      "step": 1491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6572755327281506e-06,
+      "loss": 1.3955,
+      "step": 1492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644581840812019e-06,
+      "loss": 1.4023,
+      "step": 1493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631894238502495e-06,
+      "loss": 1.4388,
+      "step": 1494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619212748825776e-06,
+      "loss": 1.4117,
+      "step": 1495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606537394796963e-06,
+      "loss": 1.376,
+      "step": 1496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593868199420017e-06,
+      "loss": 1.4124,
+      "step": 1497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581205185687727e-06,
+      "loss": 1.4205,
+      "step": 1498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568548376581663e-06,
+      "loss": 1.3706,
+      "step": 1499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555897795072128e-06,
+      "loss": 1.4353,
+      "step": 1500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543253464118131e-06,
+      "loss": 1.4361,
+      "step": 1501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530615406667333e-06,
+      "loss": 1.414,
+      "step": 1502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 1.4163,
+      "step": 1503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505358204009018e-06,
+      "loss": 1.4093,
+      "step": 1504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492739104639727e-06,
+      "loss": 1.4346,
+      "step": 1505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480126370450015e-06,
+      "loss": 1.3917,
+      "step": 1506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467520024330194e-06,
+      "loss": 1.4255,
+      "step": 1507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454920089158992e-06,
+      "loss": 1.412,
+      "step": 1508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4423265878035015e-06,
+      "loss": 1.4055,
+      "step": 1509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42973954311913e-06,
+      "loss": 1.4048,
+      "step": 1510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4171589779495755e-06,
+      "loss": 1.4088,
+      "step": 1511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404584915126779e-06,
+      "loss": 1.4091,
+      "step": 1512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392017377470867e-06,
+      "loss": 1.4431,
+      "step": 1513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3794563877901385e-06,
+      "loss": 1.4198,
+      "step": 1514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366901968881002e-06,
+      "loss": 1.4297,
+      "step": 1515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3543541435279386e-06,
+      "loss": 1.4139,
+      "step": 1516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341812934503469e-06,
+      "loss": 1.3787,
+      "step": 1517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329278364568108e-06,
+      "loss": 1.402,
+      "step": 1518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.316750456470304e-06,
+      "loss": 1.4363,
+      "step": 1519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30422923294644e-06,
+      "loss": 1.435,
+      "step": 1520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2917147167207495e-06,
+      "loss": 1.413,
+      "step": 1521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279206930505303e-06,
+      "loss": 1.4146,
+      "step": 1522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266705896999949e-06,
+      "loss": 1.4396,
+      "step": 1523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254211638892287e-06,
+      "loss": 1.3909,
+      "step": 1524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 1.4303,
+      "step": 1525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229243539558907e-06,
+      "loss": 1.4078,
+      "step": 1526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216769743646734e-06,
+      "loss": 1.42,
+      "step": 1527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204302813759272e-06,
+      "loss": 1.3917,
+      "step": 1528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191842772522214e-06,
+      "loss": 1.4144,
+      "step": 1529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179389642548771e-06,
+      "loss": 1.4435,
+      "step": 1530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166943446439604e-06,
+      "loss": 1.4282,
+      "step": 1531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15450420678278e-06,
+      "loss": 1.4041,
+      "step": 1532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142071946153751e-06,
+      "loss": 1.4429,
+      "step": 1533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12964668711531e-06,
+      "loss": 1.3808,
+      "step": 1534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117228452217525e-06,
+      "loss": 1.4142,
+      "step": 1535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104817263997727e-06,
+      "loss": 1.3572,
+      "step": 1536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092413144980465e-06,
+      "loss": 1.4241,
+      "step": 1537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080016117677441e-06,
+      "loss": 1.3854,
+      "step": 1538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067626204587498e-06,
+      "loss": 1.4345,
+      "step": 1539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055243428196571e-06,
+      "loss": 1.4201,
+      "step": 1540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04286781097763e-06,
+      "loss": 1.3995,
+      "step": 1541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0304993753906635e-06,
+      "loss": 1.4507,
+      "step": 1542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018138143882621e-06,
+      "loss": 1.3998,
+      "step": 1543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0057841388873865e-06,
+      "loss": 1.4159,
+      "step": 1544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993437382825711e-06,
+      "loss": 1.4009,
+      "step": 1545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981097898105206e-06,
+      "loss": 1.4268,
+      "step": 1546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 1.4195,
+      "step": 1547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956440832252104e-06,
+      "loss": 1.4452,
+      "step": 1548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944123295868574e-06,
+      "loss": 1.4137,
+      "step": 1549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931813120324266e-06,
+      "loss": 1.4187,
+      "step": 1550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919510327960396e-06,
+      "loss": 1.4018,
+      "step": 1551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907214941104782e-06,
+      "loss": 1.3911,
+      "step": 1552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894926982071805e-06,
+      "loss": 1.4174,
+      "step": 1553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882646473162359e-06,
+      "loss": 1.4403,
+      "step": 1554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8703734366638235e-06,
+      "loss": 1.4147,
+      "step": 1555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858107894850011e-06,
+      "loss": 1.4166,
+      "step": 1556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 1.4094,
+      "step": 1557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8335993843037695e-06,
+      "loss": 1.4298,
+      "step": 1558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821356460050806e-06,
+      "loss": 1.3922,
+      "step": 1559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809121119441401e-06,
+      "loss": 1.4038,
+      "step": 1560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796893384680964e-06,
+      "loss": 1.3827,
+      "step": 1561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78467327796109e-06,
+      "loss": 1.4372,
+      "step": 1562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772460821459542e-06,
+      "loss": 1.4349,
+      "step": 1563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760256037340181e-06,
+      "loss": 1.4059,
+      "step": 1564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748058947752955e-06,
+      "loss": 1.464,
+      "step": 1565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735869574833859e-06,
+      "loss": 1.4369,
+      "step": 1566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723687940704857e-06,
+      "loss": 1.4179,
+      "step": 1567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711514067473886e-06,
+      "loss": 1.4146,
+      "step": 1568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 1.4211,
+      "step": 1569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687189692067311e-06,
+      "loss": 1.4439,
+      "step": 1570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675039234036983e-06,
+      "loss": 1.4272,
+      "step": 1571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66289662519517e-06,
+      "loss": 1.3974,
+      "step": 1572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650761887578977e-06,
+      "loss": 1.3889,
+      "step": 1573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638635043211228e-06,
+      "loss": 1.4319,
+      "step": 1574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626516114100424e-06,
+      "loss": 1.4532,
+      "step": 1575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614405122240689e-06,
+      "loss": 1.3909,
+      "step": 1576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602302089611755e-06,
+      "loss": 1.393,
+      "step": 1577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590207038178903e-06,
+      "loss": 1.4153,
+      "step": 1578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5781199898929315e-06,
+      "loss": 1.42,
+      "step": 1579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566040966690115e-06,
+      "loss": 1.4133,
+      "step": 1580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553969990492164e-06,
+      "loss": 1.409,
+      "step": 1581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541907083206176e-06,
+      "loss": 1.3809,
+      "step": 1582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529852266724616e-06,
+      "loss": 1.4213,
+      "step": 1583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51780556292526e-06,
+      "loss": 1.3801,
+      "step": 1584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5057669936711625e-06,
+      "loss": 1.3958,
+      "step": 1585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493736580810614e-06,
+      "loss": 1.4103,
+      "step": 1586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4817143461771024e-06,
+      "loss": 1.4095,
+      "step": 1587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469700311589277e-06,
+      "loss": 1.3936,
+      "step": 1588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457694498850892e-06,
+      "loss": 1.4261,
+      "step": 1589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445696929750797e-06,
+      "loss": 1.4052,
+      "step": 1590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 1.4148,
+      "step": 1591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421726609545992e-06,
+      "loss": 1.4189,
+      "step": 1592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409753901944006e-06,
+      "loss": 1.4501,
+      "step": 1593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397789524985675e-06,
+      "loss": 1.4259,
+      "step": 1594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385833500384632e-06,
+      "loss": 1.396,
+      "step": 1595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373885849839367e-06,
+      "loss": 1.4377,
+      "step": 1596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361946595033165e-06,
+      "loss": 1.3914,
+      "step": 1597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350015757634075e-06,
+      "loss": 1.4276,
+      "step": 1598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33809335929487e-06,
+      "loss": 1.4231,
+      "step": 1599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3261794216530104e-06,
+      "loss": 1.4165,
+      "step": 1600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314273966330591e-06,
+      "loss": 1.4263,
+      "step": 1601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302377014934322e-06,
+      "loss": 1.4188,
+      "step": 1602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290488589055484e-06,
+      "loss": 1.4147,
+      "step": 1603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278608710269863e-06,
+      "loss": 1.4193,
+      "step": 1604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266737400137765e-06,
+      "loss": 1.4034,
+      "step": 1605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254874680203923e-06,
+      "loss": 1.3796,
+      "step": 1606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24302057199749e-06,
+      "loss": 1.4083,
+      "step": 1607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231175097031978e-06,
+      "loss": 1.4302,
+      "step": 1608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219338276805243e-06,
+      "loss": 1.4249,
+      "step": 1609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207510132799436e-06,
+      "loss": 1.418,
+      "step": 1610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19569068648094e-06,
+      "loss": 1.3994,
+      "step": 1611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183879959300382e-06,
+      "loss": 1.4157,
+      "step": 1612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 1.4191,
+      "step": 1613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16028474807637e-06,
+      "loss": 1.4071,
+      "step": 1614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148500306854862e-06,
+      "loss": 1.4002,
+      "step": 1615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13672467041512e-06,
+      "loss": 1.4154,
+      "step": 1616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124957860128237e-06,
+      "loss": 1.4305,
+      "step": 1617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113199897349302e-06,
+      "loss": 1.4301,
+      "step": 1618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101450803417357e-06,
+      "loss": 1.4122,
+      "step": 1619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089710599655323e-06,
+      "loss": 1.4173,
+      "step": 1620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077979307370004e-06,
+      "loss": 1.4296,
+      "step": 1621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.06625694785203e-06,
+      "loss": 1.4296,
+      "step": 1622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054543542375809e-06,
+      "loss": 1.4357,
+      "step": 1623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042839112199509e-06,
+      "loss": 1.4236,
+      "step": 1624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031143678565005e-06,
+      "loss": 1.4312,
+      "step": 1625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019457262697847e-06,
+      "loss": 1.4347,
+      "step": 1626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007779885807216e-06,
+      "loss": 1.4047,
+      "step": 1627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996111569085889e-06,
+      "loss": 1.3761,
+      "step": 1628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984452333710207e-06,
+      "loss": 1.4198,
+      "step": 1629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972802200840013e-06,
+      "loss": 1.4048,
+      "step": 1630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961161191618649e-06,
+      "loss": 1.3961,
+      "step": 1631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949529327172885e-06,
+      "loss": 1.3892,
+      "step": 1632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937906628612905e-06,
+      "loss": 1.4188,
+      "step": 1633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92629311703225e-06,
+      "loss": 1.4169,
+      "step": 1634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 1.3839,
+      "step": 1635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903093739099699e-06,
+      "loss": 1.4374,
+      "step": 1636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89150791485137e-06,
+      "loss": 1.3997,
+      "step": 1637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879931361789431e-06,
+      "loss": 1.409,
+      "step": 1638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868364100923681e-06,
+      "loss": 1.3908,
+      "step": 1639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856806153247051e-06,
+      "loss": 1.3781,
+      "step": 1640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845257539735577e-06,
+      "loss": 1.436,
+      "step": 1641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833718281348342e-06,
+      "loss": 1.4124,
+      "step": 1642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822188399027461e-06,
+      "loss": 1.4265,
+      "step": 1643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810667913698032e-06,
+      "loss": 1.4339,
+      "step": 1644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7991568462680945e-06,
+      "loss": 1.4257,
+      "step": 1645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7876552176286005e-06,
+      "loss": 1.4237,
+      "step": 1646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77616304865337e-06,
+      "loss": 1.4457,
+      "step": 1647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7646803601990555e-06,
+      "loss": 1.4308,
+      "step": 1648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7532071731050975e-06,
+      "loss": 1.4101,
+      "step": 1649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7417435081937034e-06,
+      "loss": 1.4156,
+      "step": 1650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7302893862697915e-06,
+      "loss": 1.4223,
+      "step": 1651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7188448281209665e-06,
+      "loss": 1.4172,
+      "step": 1652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707409854517471e-06,
+      "loss": 1.4143,
+      "step": 1653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695984486212161e-06,
+      "loss": 1.3847,
+      "step": 1654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684568743940444e-06,
+      "loss": 1.4257,
+      "step": 1655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673162648420275e-06,
+      "loss": 1.3753,
+      "step": 1656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 1.4458,
+      "step": 1657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650379480418795e-06,
+      "loss": 1.4203,
+      "step": 1658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639002449285693e-06,
+      "loss": 1.3958,
+      "step": 1659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627635147600485e-06,
+      "loss": 1.4224,
+      "step": 1660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616277595993196e-06,
+      "loss": 1.4202,
+      "step": 1661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604929815076174e-06,
+      "loss": 1.4531,
+      "step": 1662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 1.3975,
+      "step": 1663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58226364767359e-06,
+      "loss": 1.3853,
+      "step": 1664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57094530232389e-06,
+      "loss": 1.3869,
+      "step": 1665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559636809936124e-06,
+      "loss": 1.4049,
+      "step": 1666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548338191033595e-06,
+      "loss": 1.3965,
+      "step": 1667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5370494661216835e-06,
+      "loss": 1.4413,
+      "step": 1668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525770655687821e-06,
+      "loss": 1.3794,
+      "step": 1669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514501780201449e-06,
+      "loss": 1.4207,
+      "step": 1670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5032428601139644e-06,
+      "loss": 1.3638,
+      "step": 1671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491993915858706e-06,
+      "loss": 1.4455,
+      "step": 1672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48075496785092e-06,
+      "loss": 1.3867,
+      "step": 1673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469526036487686e-06,
+      "loss": 1.3682,
+      "step": 1674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45830714214792e-06,
+      "loss": 1.4149,
+      "step": 1675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447098305192323e-06,
+      "loss": 1.4303,
+      "step": 1676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435899545963333e-06,
+      "loss": 1.4197,
+      "step": 1677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424710884785108e-06,
+      "loss": 1.4327,
+      "step": 1678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 1.4165,
+      "step": 1679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402363937785904e-06,
+      "loss": 1.39,
+      "step": 1680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391205692521453e-06,
+      "loss": 1.3899,
+      "step": 1681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3800576264207586e-06,
+      "loss": 1.3864,
+      "step": 1682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368919759715964e-06,
+      "loss": 1.4273,
+      "step": 1683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357792112620718e-06,
+      "loss": 1.428,
+      "step": 1684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346674705330117e-06,
+      "loss": 1.439,
+      "step": 1685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335567558020673e-06,
+      "loss": 1.3997,
+      "step": 1686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32447069085028e-06,
+      "loss": 1.4539,
+      "step": 1687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313384123958172e-06,
+      "loss": 1.3817,
+      "step": 1688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302307877464893e-06,
+      "loss": 1.3857,
+      "step": 1689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29124197147225e-06,
+      "loss": 1.4312,
+      "step": 1690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280186426063291e-06,
+      "loss": 1.4221,
+      "step": 1691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269141261302256e-06,
+      "loss": 1.3712,
+      "step": 1692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258106497234551e-06,
+      "loss": 1.4187,
+      "step": 1693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2470821538867e-06,
+      "loss": 1.4059,
+      "step": 1694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236068251266324e-06,
+      "loss": 1.4373,
+      "step": 1695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2250648093620805e-06,
+      "loss": 1.3805,
+      "step": 1696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214071848143655e-06,
+      "loss": 1.4287,
+      "step": 1697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203089387561709e-06,
+      "loss": 1.4663,
+      "step": 1698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192117447547845e-06,
+      "loss": 1.4063,
+      "step": 1699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1811560480145754e-06,
+      "loss": 1.4036,
+      "step": 1700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 1.4419,
+      "step": 1701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159264949944173e-06,
+      "loss": 1.3677,
+      "step": 1702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148335291136267e-06,
+      "loss": 1.4245,
+      "step": 1703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1374162522673436e-06,
+      "loss": 1.4243,
+      "step": 1704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126507853153891e-06,
+      "loss": 1.4446,
+      "step": 1705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115610113593117e-06,
+      "loss": 1.4163,
+      "step": 1706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104723053362867e-06,
+      "loss": 1.3921,
+      "step": 1707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093846692221597e-06,
+      "loss": 1.4464,
+      "step": 1708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082981049908362e-06,
+      "loss": 1.4018,
+      "step": 1709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072126146142754e-06,
+      "loss": 1.4098,
+      "step": 1710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061282000624885e-06,
+      "loss": 1.3902,
+      "step": 1711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050448633035326e-06,
+      "loss": 1.4542,
+      "step": 1712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039626063035107e-06,
+      "loss": 1.4084,
+      "step": 1713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028814310265654e-06,
+      "loss": 1.4316,
+      "step": 1714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018013394348752e-06,
+      "loss": 1.3865,
+      "step": 1715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007223334886531e-06,
+      "loss": 1.4123,
+      "step": 1716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996444151461417e-06,
+      "loss": 1.4235,
+      "step": 1717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98567586363609e-06,
+      "loss": 1.4444,
+      "step": 1718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9749184909534565e-06,
+      "loss": 1.3824,
+      "step": 1719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96417205293663e-06,
+      "loss": 1.3901,
+      "step": 1720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953436569088856e-06,
+      "loss": 1.4404,
+      "step": 1721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9427120588935154e-06,
+      "loss": 1.3755,
+      "step": 1722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 1.3892,
+      "step": 1723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921296037294022e-06,
+      "loss": 1.4174,
+      "step": 1724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9106045647569005e-06,
+      "loss": 1.3963,
+      "step": 1725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899924143606204e-06,
+      "loss": 1.4493,
+      "step": 1726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88925479322538e-06,
+      "loss": 1.4303,
+      "step": 1727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878596532977779e-06,
+      "loss": 1.4071,
+      "step": 1728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867949382206632e-06,
+      "loss": 1.3808,
+      "step": 1729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857313360235001e-06,
+      "loss": 1.3726,
+      "step": 1730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846688486365749e-06,
+      "loss": 1.4217,
+      "step": 1731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836074779881511e-06,
+      "loss": 1.3595,
+      "step": 1732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825472260044658e-06,
+      "loss": 1.4114,
+      "step": 1733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814880946097253e-06,
+      "loss": 1.4263,
+      "step": 1734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8043008572610253e-06,
+      "loss": 1.4336,
+      "step": 1735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7937320127373336e-06,
+      "loss": 1.3967,
+      "step": 1736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7831744317071194e-06,
+      "loss": 1.4458,
+      "step": 1737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7726281333308955e-06,
+      "loss": 1.4031,
+      "step": 1738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7620931367486925e-06,
+      "loss": 1.4104,
+      "step": 1739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75156946108003e-06,
+      "loss": 1.4095,
+      "step": 1740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7410571254238835e-06,
+      "loss": 1.4356,
+      "step": 1741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7305561488586484e-06,
+      "loss": 1.4106,
+      "step": 1742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7200665504420984e-06,
+      "loss": 1.4208,
+      "step": 1743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.709588349211364e-06,
+      "loss": 1.4341,
+      "step": 1744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 1.43,
+      "step": 1745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6886662143524034e-06,
+      "loss": 1.428,
+      "step": 1746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678222318694875e-06,
+      "loss": 1.4297,
+      "step": 1747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667789896164495e-06,
+      "loss": 1.4295,
+      "step": 1748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6573689656946177e-06,
+      "loss": 1.3863,
+      "step": 1749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6469595461977546e-06,
+      "loss": 1.3929,
+      "step": 1750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6365616565655193e-06,
+      "loss": 1.3622,
+      "step": 1751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626175315668603e-06,
+      "loss": 1.4386,
+      "step": 1752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.615800542356738e-06,
+      "loss": 1.405,
+      "step": 1753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6054373554586598e-06,
+      "loss": 1.4096,
+      "step": 1754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5950857737820834e-06,
+      "loss": 1.4112,
+      "step": 1755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5847458161136485e-06,
+      "loss": 1.4143,
+      "step": 1756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574417501218913e-06,
+      "loss": 1.4201,
+      "step": 1757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564100847842297e-06,
+      "loss": 1.4094,
+      "step": 1758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55379587470706e-06,
+      "loss": 1.465,
+      "step": 1759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5435026005152605e-06,
+      "loss": 1.3839,
+      "step": 1760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5332210439477334e-06,
+      "loss": 1.4181,
+      "step": 1761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5229512236640317e-06,
+      "loss": 1.4552,
+      "step": 1762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5126931583024215e-06,
+      "loss": 1.3647,
+      "step": 1763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502446866479837e-06,
+      "loss": 1.3912,
+      "step": 1764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4922123667918305e-06,
+      "loss": 1.4037,
+      "step": 1765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4819896778125706e-06,
+      "loss": 1.3915,
+      "step": 1766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 1.4405,
+      "step": 1767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4615798061697258e-06,
+      "loss": 1.4282,
+      "step": 1768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 1.4749,
+      "step": 1769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4412173997152785e-06,
+      "loss": 1.4226,
+      "step": 1770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4310540421407667e-06,
+      "loss": 1.4188,
+      "step": 1771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42090260626865e-06,
+      "loss": 1.4103,
+      "step": 1772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4107631105223528e-06,
+      "loss": 1.4137,
+      "step": 1773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4006355733036167e-06,
+      "loss": 1.3905,
+      "step": 1774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390520012992474e-06,
+      "loss": 1.4118,
+      "step": 1775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3804164479472313e-06,
+      "loss": 1.4119,
+      "step": 1776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3703248965044253e-06,
+      "loss": 1.3977,
+      "step": 1777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.360245376978779e-06,
+      "loss": 1.4287,
+      "step": 1778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3501779076631867e-06,
+      "loss": 1.422,
+      "step": 1779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3401225068286825e-06,
+      "loss": 1.4099,
+      "step": 1780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330079192724379e-06,
+      "loss": 1.3977,
+      "step": 1781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3200479835774656e-06,
+      "loss": 1.4194,
+      "step": 1782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310028897593164e-06,
+      "loss": 1.4134,
+      "step": 1783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3000219529546817e-06,
+      "loss": 1.4195,
+      "step": 1784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2900271678232045e-06,
+      "loss": 1.3736,
+      "step": 1785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280044560337845e-06,
+      "loss": 1.4008,
+      "step": 1786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270074148615615e-06,
+      "loss": 1.4302,
+      "step": 1787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2601159507513936e-06,
+      "loss": 1.423,
+      "step": 1788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 1.4087,
+      "step": 1789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240236268865632e-06,
+      "loss": 1.4402,
+      "step": 1790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230314820922883e-06,
+      "loss": 1.4111,
+      "step": 1791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2204056589956657e-06,
+      "loss": 1.4578,
+      "step": 1792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2105088010677e-06,
+      "loss": 1.3901,
+      "step": 1793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.200624265100376e-06,
+      "loss": 1.4016,
+      "step": 1794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190752069032719e-06,
+      "loss": 1.4004,
+      "step": 1795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1808922307813628e-06,
+      "loss": 1.4015,
+      "step": 1796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171044768240508e-06,
+      "loss": 1.4542,
+      "step": 1797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1612096992818973e-06,
+      "loss": 1.4189,
+      "step": 1798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151387041754784e-06,
+      "loss": 1.412,
+      "step": 1799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1415768134858947e-06,
+      "loss": 1.4241,
+      "step": 1800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131779032279397e-06,
+      "loss": 1.4304,
+      "step": 1801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1219937159168743e-06,
+      "loss": 1.3855,
+      "step": 1802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112220882157275e-06,
+      "loss": 1.4008,
+      "step": 1803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1024605487369076e-06,
+      "loss": 1.4351,
+      "step": 1804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0927127333693872e-06,
+      "loss": 1.4019,
+      "step": 1805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0829774537456124e-06,
+      "loss": 1.4486,
+      "step": 1806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073254727533732e-06,
+      "loss": 1.4091,
+      "step": 1807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0635445723791123e-06,
+      "loss": 1.4161,
+      "step": 1808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053847005904298e-06,
+      "loss": 1.398,
+      "step": 1809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.044162045708996e-06,
+      "loss": 1.4179,
+      "step": 1810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 1.439,
+      "step": 1811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0248300144413157e-06,
+      "loss": 1.4201,
+      "step": 1812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0151829784538257e-06,
+      "loss": 1.4011,
+      "step": 1813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005548618915557e-06,
+      "loss": 1.4419,
+      "step": 1814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9959269533115042e-06,
+      "loss": 1.4286,
+      "step": 1815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.986317999103616e-06,
+      "loss": 1.4032,
+      "step": 1816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9767217737307805e-06,
+      "loss": 1.4012,
+      "step": 1817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9671382946087847e-06,
+      "loss": 1.4288,
+      "step": 1818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9575675791302704e-06,
+      "loss": 1.3999,
+      "step": 1819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9480096446647335e-06,
+      "loss": 1.4104,
+      "step": 1820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938464508558466e-06,
+      "loss": 1.4066,
+      "step": 1821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 1.3839,
+      "step": 1822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9194127006927208e-06,
+      "loss": 1.3762,
+      "step": 1823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90990606350957e-06,
+      "loss": 1.4122,
+      "step": 1824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9004122938382617e-06,
+      "loss": 1.4055,
+      "step": 1825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890931408908637e-06,
+      "loss": 1.4284,
+      "step": 1826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.881463425927161e-06,
+      "loss": 1.4247,
+      "step": 1827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872008362076869e-06,
+      "loss": 1.4412,
+      "step": 1828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.86256623451736e-06,
+      "loss": 1.4093,
+      "step": 1829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853137060384752e-06,
+      "loss": 1.3913,
+      "step": 1830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843720856791652e-06,
+      "loss": 1.389,
+      "step": 1831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8343176408271312e-06,
+      "loss": 1.4106,
+      "step": 1832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 1.412,
+      "step": 1833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8155502400222246e-06,
+      "loss": 1.4311,
+      "step": 1834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8061860892420013e-06,
+      "loss": 1.4099,
+      "step": 1835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796834994210621e-06,
+      "loss": 1.4217,
+      "step": 1836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7874969718989943e-06,
+      "loss": 1.4206,
+      "step": 1837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7781720392542965e-06,
+      "loss": 1.4071,
+      "step": 1838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7688602131999564e-06,
+      "loss": 1.4531,
+      "step": 1839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7595615106356154e-06,
+      "loss": 1.4134,
+      "step": 1840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7502759484370946e-06,
+      "loss": 1.4132,
+      "step": 1841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7410035434563677e-06,
+      "loss": 1.4422,
+      "step": 1842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731744312521536e-06,
+      "loss": 1.4067,
+      "step": 1843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7224982724367775e-06,
+      "loss": 1.4153,
+      "step": 1844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7132654399823444e-06,
+      "loss": 1.4058,
+      "step": 1845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7040458319145137e-06,
+      "loss": 1.4095,
+      "step": 1846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.694839464965563e-06,
+      "loss": 1.4099,
+      "step": 1847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6856463558437374e-06,
+      "loss": 1.4334,
+      "step": 1848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.676466521233225e-06,
+      "loss": 1.4101,
+      "step": 1849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6672999777941168e-06,
+      "loss": 1.4233,
+      "step": 1850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6581467421623842e-06,
+      "loss": 1.4069,
+      "step": 1851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6490068309498483e-06,
+      "loss": 1.423,
+      "step": 1852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639880260744151e-06,
+      "loss": 1.4084,
+      "step": 1853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6307670481087143e-06,
+      "loss": 1.4074,
+      "step": 1854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 1.414,
+      "step": 1855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6125807616811006e-06,
+      "loss": 1.4007,
+      "step": 1856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6035077208944416e-06,
+      "loss": 1.417,
+      "step": 1857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594448103689031e-06,
+      "loss": 1.4621,
+      "step": 1858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585401926506785e-06,
+      "loss": 1.4145,
+      "step": 1859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576369205765229e-06,
+      "loss": 1.4161,
+      "step": 1860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5673499578574644e-06,
+      "loss": 1.4211,
+      "step": 1861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5583441991521475e-06,
+      "loss": 1.4242,
+      "step": 1862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5493519459934423e-06,
+      "loss": 1.4277,
+      "step": 1863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.540373214701013e-06,
+      "loss": 1.407,
+      "step": 1864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5314080215699822e-06,
+      "loss": 1.4027,
+      "step": 1865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5224563828708904e-06,
+      "loss": 1.4089,
+      "step": 1866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5135183148496976e-06,
+      "loss": 1.4088,
+      "step": 1867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5045938337277263e-06,
+      "loss": 1.4164,
+      "step": 1868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4956829557016336e-06,
+      "loss": 1.4288,
+      "step": 1869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486785696943399e-06,
+      "loss": 1.4193,
+      "step": 1870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4779020736002836e-06,
+      "loss": 1.4078,
+      "step": 1871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4690321017947938e-06,
+      "loss": 1.3826,
+      "step": 1872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4601757976246685e-06,
+      "loss": 1.4206,
+      "step": 1873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451333177162847e-06,
+      "loss": 1.4287,
+      "step": 1874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 1.391,
+      "step": 1875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4336890515316215e-06,
+      "loss": 1.4306,
+      "step": 1876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 1.4387,
+      "step": 1877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4160998529873756e-06,
+      "loss": 1.4282,
+      "step": 1878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.407325891290817e-06,
+      "loss": 1.4262,
+      "step": 1879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3985657092176153e-06,
+      "loss": 1.4176,
+      "step": 1880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389819322666264e-06,
+      "loss": 1.3951,
+      "step": 1881
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3810867475101995e-06,
+      "loss": 1.4089,
+      "step": 1882
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3723679995978088e-06,
+      "loss": 1.3937,
+      "step": 1883
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363663094752381e-06,
+      "loss": 1.4173,
+      "step": 1884
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354972048772074e-06,
+      "loss": 1.3918,
+      "step": 1885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3462948774299034e-06,
+      "loss": 1.4081,
+      "step": 1886
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3376315964737008e-06,
+      "loss": 1.3908,
+      "step": 1887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.328982221626087e-06,
+      "loss": 1.388,
+      "step": 1888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.320346768584449e-06,
+      "loss": 1.4015,
+      "step": 1889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3117252530209067e-06,
+      "loss": 1.398,
+      "step": 1890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3031176905822805e-06,
+      "loss": 1.4473,
+      "step": 1891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2945240968900728e-06,
+      "loss": 1.437,
+      "step": 1892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2859444875404347e-06,
+      "loss": 1.3805,
+      "step": 1893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277378878104136e-06,
+      "loss": 1.4136,
+      "step": 1894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268827284126539e-06,
+      "loss": 1.4124,
+      "step": 1895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260289721127572e-06,
+      "loss": 1.4299,
+      "step": 1896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2517662046016975e-06,
+      "loss": 1.4258,
+      "step": 1897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2432567500178828e-06,
+      "loss": 1.4049,
+      "step": 1898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 1.4164,
+      "step": 1899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2262800884246828e-06,
+      "loss": 1.4298,
+      "step": 1900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2178129122255255e-06,
+      "loss": 1.4165,
+      "step": 1901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209359859588823e-06,
+      "loss": 1.3913,
+      "step": 1902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.200920945855669e-06,
+      "loss": 1.3793,
+      "step": 1903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192496186341484e-06,
+      "loss": 1.393,
+      "step": 1904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184085596336011e-06,
+      "loss": 1.4041,
+      "step": 1905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175689191103275e-06,
+      "loss": 1.432,
+      "step": 1906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1673069858815553e-06,
+      "loss": 1.399,
+      "step": 1907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158938995883365e-06,
+      "loss": 1.4137,
+      "step": 1908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150585236295415e-06,
+      "loss": 1.4139,
+      "step": 1909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1422457222785875e-06,
+      "loss": 1.394,
+      "step": 1910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133920468967915e-06,
+      "loss": 1.4032,
+      "step": 1911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.125609491472549e-06,
+      "loss": 1.4119,
+      "step": 1912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1173128048757307e-06,
+      "loss": 1.4106,
+      "step": 1913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109030424234766e-06,
+      "loss": 1.4013,
+      "step": 1914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1007623645810004e-06,
+      "loss": 1.4119,
+      "step": 1915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0925086409197815e-06,
+      "loss": 1.4201,
+      "step": 1916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0842692682304442e-06,
+      "loss": 1.4364,
+      "step": 1917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0760442614662803e-06,
+      "loss": 1.4069,
+      "step": 1918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0678336355545047e-06,
+      "loss": 1.4052,
+      "step": 1919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0596374053962363e-06,
+      "loss": 1.4003,
+      "step": 1920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 1.4184,
+      "step": 1921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043288191814037e-06,
+      "loss": 1.4118,
+      "step": 1922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035135238061601e-06,
+      "loss": 1.4286,
+      "step": 1923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026996739405611e-06,
+      "loss": 1.4069,
+      "step": 1924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0188727106162874e-06,
+      "loss": 1.4112,
+      "step": 1925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0107631664375805e-06,
+      "loss": 1.4428,
+      "step": 1926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0026681215871656e-06,
+      "loss": 1.3597,
+      "step": 1927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994587590756397e-06,
+      "loss": 1.4071,
+      "step": 1928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986521588610285e-06,
+      "loss": 1.4209,
+      "step": 1929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9784701297874774e-06,
+      "loss": 1.4286,
+      "step": 1930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9704332289002292e-06,
+      "loss": 1.4093,
+      "step": 1931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962410900534367e-06,
+      "loss": 1.4356,
+      "step": 1932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9544031592492763e-06,
+      "loss": 1.419,
+      "step": 1933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9464100195778757e-06,
+      "loss": 1.4432,
+      "step": 1934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938431496026569e-06,
+      "loss": 1.4134,
+      "step": 1935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9304676030752435e-06,
+      "loss": 1.4218,
+      "step": 1936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922518355177232e-06,
+      "loss": 1.4189,
+      "step": 1937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9145837667592914e-06,
+      "loss": 1.4032,
+      "step": 1938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9066638522215652e-06,
+      "loss": 1.4014,
+      "step": 1939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8987586259375724e-06,
+      "loss": 1.3984,
+      "step": 1940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.890868102254182e-06,
+      "loss": 1.4159,
+      "step": 1941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8829922954915626e-06,
+      "loss": 1.3878,
+      "step": 1942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 1.4252,
+      "step": 1943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8672848898757912e-06,
+      "loss": 1.4123,
+      "step": 1944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.859453319529343e-06,
+      "loss": 1.4161,
+      "step": 1945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8516365231170318e-06,
+      "loss": 1.4032,
+      "step": 1946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843834514825229e-06,
+      "loss": 1.4137,
+      "step": 1947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8360473088134722e-06,
+      "loss": 1.4124,
+      "step": 1948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82827491921443e-06,
+      "loss": 1.4236,
+      "step": 1949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8205173601338889e-06,
+      "loss": 1.4021,
+      "step": 1950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8127746456507077e-06,
+      "loss": 1.4445,
+      "step": 1951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8050467898168156e-06,
+      "loss": 1.3565,
+      "step": 1952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797333806657171e-06,
+      "loss": 1.4004,
+      "step": 1953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896357101697405e-06,
+      "loss": 1.4075,
+      "step": 1954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7819525143254758e-06,
+      "loss": 1.4117,
+      "step": 1955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7742842330682852e-06,
+      "loss": 1.3885,
+      "step": 1956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7666308803150045e-06,
+      "loss": 1.4212,
+      "step": 1957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7589924699553817e-06,
+      "loss": 1.4249,
+      "step": 1958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7513690158520457e-06,
+      "loss": 1.4277,
+      "step": 1959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7437605318404805e-06,
+      "loss": 1.4198,
+      "step": 1960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7361670317290014e-06,
+      "loss": 1.4164,
+      "step": 1961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7285885292987338e-06,
+      "loss": 1.4016,
+      "step": 1962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.721025038303581e-06,
+      "loss": 1.3758,
+      "step": 1963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7134765724701996e-06,
+      "loss": 1.4233,
+      "step": 1964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 1.4029,
+      "step": 1965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6984247710590296e-06,
+      "loss": 1.4298,
+      "step": 1966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6909214627981197e-06,
+      "loss": 1.4042,
+      "step": 1967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683433234332691e-06,
+      "loss": 1.42,
+      "step": 1968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6759600992528147e-06,
+      "loss": 1.394,
+      "step": 1969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.668502071121162e-06,
+      "loss": 1.4102,
+      "step": 1970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6610591634729966e-06,
+      "loss": 1.4242,
+      "step": 1971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6536313898161393e-06,
+      "loss": 1.4229,
+      "step": 1972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6462187636309345e-06,
+      "loss": 1.3867,
+      "step": 1973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6388212983702534e-06,
+      "loss": 1.4219,
+      "step": 1974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6314390074594412e-06,
+      "loss": 1.4008,
+      "step": 1975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6240719042963004e-06,
+      "loss": 1.4343,
+      "step": 1976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6167200022510799e-06,
+      "loss": 1.3907,
+      "step": 1977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6093833146664373e-06,
+      "loss": 1.4042,
+      "step": 1978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6020618548574108e-06,
+      "loss": 1.419,
+      "step": 1979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5947556361114102e-06,
+      "loss": 1.4662,
+      "step": 1980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 1.4036,
+      "step": 1981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5801889748198052e-06,
+      "loss": 1.3729,
+      "step": 1982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5729285587106135e-06,
+      "loss": 1.407,
+      "step": 1983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5656834365372376e-06,
+      "loss": 1.4342,
+      "step": 1984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5584536214485457e-06,
+      "loss": 1.3939,
+      "step": 1985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.551239126565618e-06,
+      "loss": 1.4017,
+      "step": 1986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 1.4398,
+      "step": 1987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5368561497623668e-06,
+      "loss": 1.4063,
+      "step": 1988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5296876939450978e-06,
+      "loss": 1.4254,
+      "step": 1989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5225346105396631e-06,
+      "loss": 1.4144,
+      "step": 1990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5153969125278933e-06,
+      "loss": 1.4077,
+      "step": 1991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5082746128636905e-06,
+      "loss": 1.3996,
+      "step": 1992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.501167724473016e-06,
+      "loss": 1.404,
+      "step": 1993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940762602538628e-06,
+      "loss": 1.3734,
+      "step": 1994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487000233076229e-06,
+      "loss": 1.4289,
+      "step": 1995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4799396557820977e-06,
+      "loss": 1.4054,
+      "step": 1996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4728945411854135e-06,
+      "loss": 1.4278,
+      "step": 1997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4658649020720539e-06,
+      "loss": 1.4281,
+      "step": 1998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4588507511998162e-06,
+      "loss": 1.4095,
+      "step": 1999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4518521012983856e-06,
+      "loss": 1.423,
+      "step": 2000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444868965069315e-06,
+      "loss": 1.4048,
+      "step": 2001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4379013551860054e-06,
+      "loss": 1.4124,
+      "step": 2002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4309492842936767e-06,
+      "loss": 1.4061,
+      "step": 2003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4240127650093494e-06,
+      "loss": 1.4067,
+      "step": 2004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4170918099218166e-06,
+      "loss": 1.4055,
+      "step": 2005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410186431591629e-06,
+      "loss": 1.4237,
+      "step": 2006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032966425510664e-06,
+      "loss": 1.4153,
+      "step": 2007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396422455304114e-06,
+      "loss": 1.4004,
+      "step": 2008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 1.4259,
+      "step": 2009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3827209360653948e-06,
+      "loss": 1.428,
+      "step": 2010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3758936289399349e-06,
+      "loss": 1.3996,
+      "step": 2011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3690819733406568e-06,
+      "loss": 1.395,
+      "step": 2012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3622859816297473e-06,
+      "loss": 1.4348,
+      "step": 2013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355505666140965e-06,
+      "loss": 1.4181,
+      "step": 2014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3487410391796163e-06,
+      "loss": 1.4018,
+      "step": 2015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419921130225399e-06,
+      "loss": 1.4368,
+      "step": 2016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3352588999180726e-06,
+      "loss": 1.4177,
+      "step": 2017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3285414120860407e-06,
+      "loss": 1.4101,
+      "step": 2018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3218396617177287e-06,
+      "loss": 1.3926,
+      "step": 2019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3151536609758586e-06,
+      "loss": 1.4359,
+      "step": 2020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3084834219945731e-06,
+      "loss": 1.3742,
+      "step": 2021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301828956879405e-06,
+      "loss": 1.4141,
+      "step": 2022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2951902777072656e-06,
+      "loss": 1.4122,
+      "step": 2023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885673965264078e-06,
+      "loss": 1.4001,
+      "step": 2024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2819603253564206e-06,
+      "loss": 1.4033,
+      "step": 2025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275369076188202e-06,
+      "loss": 1.417,
+      "step": 2026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2687936609839235e-06,
+      "loss": 1.3915,
+      "step": 2027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2622340916770348e-06,
+      "loss": 1.4458,
+      "step": 2028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255690380172222e-06,
+      "loss": 1.4219,
+      "step": 2029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2491625383453854e-06,
+      "loss": 1.3505,
+      "step": 2030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 1.4324,
+      "step": 2031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2361545110852458e-06,
+      "loss": 1.4368,
+      "step": 2032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2296743492596587e-06,
+      "loss": 1.4087,
+      "step": 2033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2232101043274437e-06,
+      "loss": 1.4215,
+      "step": 2034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216761788020291e-06,
+      "loss": 1.3989,
+      "step": 2035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2103294120409715e-06,
+      "loss": 1.4164,
+      "step": 2036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203912988063335e-06,
+      "loss": 1.3984,
+      "step": 2037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1975125277322797e-06,
+      "loss": 1.3942,
+      "step": 2038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1911280426637274e-06,
+      "loss": 1.4286,
+      "step": 2039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1847595444446115e-06,
+      "loss": 1.4454,
+      "step": 2040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1784070446328477e-06,
+      "loss": 1.4233,
+      "step": 2041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1720705547573263e-06,
+      "loss": 1.4366,
+      "step": 2042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1657500863178694e-06,
+      "loss": 1.4085,
+      "step": 2043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1594456507852293e-06,
+      "loss": 1.4177,
+      "step": 2044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153157259601062e-06,
+      "loss": 1.4098,
+      "step": 2045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1468849241778978e-06,
+      "loss": 1.4046,
+      "step": 2046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1406286558991376e-06,
+      "loss": 1.3881,
+      "step": 2047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1343884661190153e-06,
+      "loss": 1.4385,
+      "step": 2048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1281643661625896e-06,
+      "loss": 1.4053,
+      "step": 2049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1219563673257162e-06,
+      "loss": 1.4181,
+      "step": 2050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1157644808750313e-06,
+      "loss": 1.4003,
+      "step": 2051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1095887180479248e-06,
+      "loss": 1.4243,
+      "step": 2052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 1.3865,
+      "step": 2053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0972856080676908e-06,
+      "loss": 1.3965,
+      "step": 2054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0911582832429591e-06,
+      "loss": 1.433,
+      "step": 2055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0850471266985551e-06,
+      "loss": 1.4188,
+      "step": 2056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078952149525362e-06,
+      "loss": 1.3991,
+      "step": 2057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0728733627848898e-06,
+      "loss": 1.4269,
+      "step": 2058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.066810777509275e-06,
+      "loss": 1.3952,
+      "step": 2059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0607644047012477e-06,
+      "loss": 1.396,
+      "step": 2060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0547342553341144e-06,
+      "loss": 1.4093,
+      "step": 2061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0487203403517377e-06,
+      "loss": 1.4006,
+      "step": 2062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0427226706685178e-06,
+      "loss": 1.4354,
+      "step": 2063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367412571693746e-06,
+      "loss": 1.411,
+      "step": 2064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.030776110709718e-06,
+      "loss": 1.4394,
+      "step": 2065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0248272421154426e-06,
+      "loss": 1.3896,
+      "step": 2066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0188946621828977e-06,
+      "loss": 1.4024,
+      "step": 2067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0129783816788742e-06,
+      "loss": 1.4617,
+      "step": 2068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0070784113405763e-06,
+      "loss": 1.4496,
+      "step": 2069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011947618756156e-06,
+      "loss": 1.419,
+      "step": 2070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953274439619742e-07,
+      "loss": 1.3822,
+      "step": 2071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.894764682480006e-07,
+      "loss": 1.4048,
+      "step": 2072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.836418453523833e-07,
+      "loss": 1.3942,
+      "step": 2073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.778235858641338e-07,
+      "loss": 1.4006,
+      "step": 2074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 1.4166,
+      "step": 2075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.662361993172743e-07,
+      "loss": 1.3911,
+      "step": 2076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.604670932881211e-07,
+      "loss": 1.3788,
+      "step": 2077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547143927252134e-07,
+      "loss": 1.4045,
+      "step": 2078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.489781080688865e-07,
+      "loss": 1.4136,
+      "step": 2079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.432582497296749e-07,
+      "loss": 1.3926,
+      "step": 2080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.375548280883129e-07,
+      "loss": 1.4031,
+      "step": 2081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.318678534957004e-07,
+      "loss": 1.4135,
+      "step": 2082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.261973362728827e-07,
+      "loss": 1.4161,
+      "step": 2083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.205432867110442e-07,
+      "loss": 1.4136,
+      "step": 2084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.149057150714802e-07,
+      "loss": 1.4282,
+      "step": 2085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.092846315855841e-07,
+      "loss": 1.4191,
+      "step": 2086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 1.3781,
+      "step": 2087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.980919698507062e-07,
+      "loss": 1.4387,
+      "step": 2088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.925204119148189e-07,
+      "loss": 1.4262,
+      "step": 2089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.869653827587366e-07,
+      "loss": 1.4155,
+      "step": 2090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814268924640468e-07,
+      "loss": 1.414,
+      "step": 2091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.759049510823247e-07,
+      "loss": 1.3774,
+      "step": 2092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703995686351041e-07,
+      "loss": 1.3474,
+      "step": 2093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.649107551138725e-07,
+      "loss": 1.3959,
+      "step": 2094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.594385204800481e-07,
+      "loss": 1.4201,
+      "step": 2095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.539828746649548e-07,
+      "loss": 1.3937,
+      "step": 2096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 1.4629,
+      "step": 2097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.431213890657275e-07,
+      "loss": 1.4051,
+      "step": 2098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.377155689936434e-07,
+      "loss": 1.3918,
+      "step": 2099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323263771643575e-07,
+      "loss": 1.4117,
+      "step": 2100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269538233584884e-07,
+      "loss": 1.4098,
+      "step": 2101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.215979173264543e-07,
+      "loss": 1.4233,
+      "step": 2102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.162586687884655e-07,
+      "loss": 1.4087,
+      "step": 2103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.109360874344962e-07,
+      "loss": 1.4081,
+      "step": 2104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.056301829242785e-07,
+      "loss": 1.3627,
+      "step": 2105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.003409648872684e-07,
+      "loss": 1.4238,
+      "step": 2106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.950684429226463e-07,
+      "loss": 1.3886,
+      "step": 2107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.898126265992912e-07,
+      "loss": 1.4466,
+      "step": 2108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.845735254557608e-07,
+      "loss": 1.4093,
+      "step": 2109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.793511490002781e-07,
+      "loss": 1.4122,
+      "step": 2110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.741455067107162e-07,
+      "loss": 1.3634,
+      "step": 2111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689566080345701e-07,
+      "loss": 1.4139,
+      "step": 2112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.637844623889557e-07,
+      "loss": 1.3774,
+      "step": 2113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.586290791605822e-07,
+      "loss": 1.4175,
+      "step": 2114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534904677057353e-07,
+      "loss": 1.424,
+      "step": 2115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.483686373502641e-07,
+      "loss": 1.3959,
+      "step": 2116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.432635973895652e-07,
+      "loss": 1.3983,
+      "step": 2117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.381753570885542e-07,
+      "loss": 1.4078,
+      "step": 2118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 1.4166,
+      "step": 2119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280493123728282e-07,
+      "loss": 1.4051,
+      "step": 2120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230115263354431e-07,
+      "loss": 1.3808,
+      "step": 2121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179905767123773e-07,
+      "loss": 1.4038,
+      "step": 2122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129864726159408e-07,
+      "loss": 1.3807,
+      "step": 2123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079992231278676e-07,
+      "loss": 1.4401,
+      "step": 2124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.030288372993066e-07,
+      "loss": 1.3695,
+      "step": 2125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.980753241508032e-07,
+      "loss": 1.3985,
+      "step": 2126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931386926722772e-07,
+      "loss": 1.42,
+      "step": 2127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882189518230142e-07,
+      "loss": 1.3868,
+      "step": 2128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.833161105316421e-07,
+      "loss": 1.3879,
+      "step": 2129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.784301776961222e-07,
+      "loss": 1.4242,
+      "step": 2130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.735611621837257e-07,
+      "loss": 1.41,
+      "step": 2131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68709072831023e-07,
+      "loss": 1.4088,
+      "step": 2132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.638739184438681e-07,
+      "loss": 1.3907,
+      "step": 2133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590557077973747e-07,
+      "loss": 1.3923,
+      "step": 2134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.542544496359138e-07,
+      "loss": 1.4049,
+      "step": 2135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.494701526730851e-07,
+      "loss": 1.4361,
+      "step": 2136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.447028255917054e-07,
+      "loss": 1.4188,
+      "step": 2137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399524770437948e-07,
+      "loss": 1.426,
+      "step": 2138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.352191156505627e-07,
+      "loss": 1.3673,
+      "step": 2139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.305027500023841e-07,
+      "loss": 1.4091,
+      "step": 2140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 1.4175,
+      "step": 2141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.211210401484602e-07,
+      "loss": 1.4273,
+      "step": 2142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.164557129691828e-07,
+      "loss": 1.3831,
+      "step": 2143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118074155878672e-07,
+      "loss": 1.4149,
+      "step": 2144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071761564405121e-07,
+      "loss": 1.4264,
+      "step": 2145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.025619439321928e-07,
+      "loss": 1.4063,
+      "step": 2146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.979647864370486e-07,
+      "loss": 1.4068,
+      "step": 2147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933846922982645e-07,
+      "loss": 1.4306,
+      "step": 2148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888216698280646e-07,
+      "loss": 1.443,
+      "step": 2149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842757273076815e-07,
+      "loss": 1.4163,
+      "step": 2150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.797468729873546e-07,
+      "loss": 1.4039,
+      "step": 2151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752351150863133e-07,
+      "loss": 1.4216,
+      "step": 2152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.707404617927526e-07,
+      "loss": 1.3901,
+      "step": 2153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.662629212638293e-07,
+      "loss": 1.3846,
+      "step": 2154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618025016256445e-07,
+      "loss": 1.364,
+      "step": 2155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.57359210973224e-07,
+      "loss": 1.4025,
+      "step": 2156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.529330573705083e-07,
+      "loss": 1.4012,
+      "step": 2157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485240488503385e-07,
+      "loss": 1.4304,
+      "step": 2158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44132193414434e-07,
+      "loss": 1.3935,
+      "step": 2159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.397574990333898e-07,
+      "loss": 1.4097,
+      "step": 2160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.353999736466531e-07,
+      "loss": 1.391,
+      "step": 2161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.310596251625133e-07,
+      "loss": 1.4022,
+      "step": 2162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 1.425,
+      "step": 2163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.224304903792987e-07,
+      "loss": 1.4344,
+      "step": 2164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181417197408733e-07,
+      "loss": 1.4198,
+      "step": 2165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138701573263216e-07,
+      "loss": 1.3961,
+      "step": 2166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09615810887919e-07,
+      "loss": 1.4151,
+      "step": 2167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053786881467004e-07,
+      "loss": 1.3787,
+      "step": 2168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.011587967924414e-07,
+      "loss": 1.425,
+      "step": 2169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969561444836446e-07,
+      "loss": 1.3846,
+      "step": 2170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927707388475256e-07,
+      "loss": 1.4032,
+      "step": 2171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886025874799993e-07,
+      "loss": 1.4331,
+      "step": 2172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844516979456671e-07,
+      "loss": 1.4095,
+      "step": 2173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80318077777805e-07,
+      "loss": 1.3697,
+      "step": 2174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7620173447834426e-07,
+      "loss": 1.4085,
+      "step": 2175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721026755178626e-07,
+      "loss": 1.3939,
+      "step": 2176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6802090833557136e-07,
+      "loss": 1.4437,
+      "step": 2177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6395644033929264e-07,
+      "loss": 1.3978,
+      "step": 2178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5990927890545934e-07,
+      "loss": 1.4338,
+      "step": 2179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55879431379096e-07,
+      "loss": 1.3943,
+      "step": 2180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5186690507379894e-07,
+      "loss": 1.4553,
+      "step": 2181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4787170727173537e-07,
+      "loss": 1.4275,
+      "step": 2182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.438938452236219e-07,
+      "loss": 1.4332,
+      "step": 2183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.399333261487082e-07,
+      "loss": 1.4106,
+      "step": 2184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 1.4114,
+      "step": 2185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.320643456381157e-07,
+      "loss": 1.3941,
+      "step": 2186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2815589848351434e-07,
+      "loss": 1.387,
+      "step": 2187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.242648228642499e-07,
+      "loss": 1.4102,
+      "step": 2188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203911258420712e-07,
+      "loss": 1.3807,
+      "step": 2189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165348144471881e-07,
+      "loss": 1.366,
+      "step": 2190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.126958956782545e-07,
+      "loss": 1.3936,
+      "step": 2191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.088743765023617e-07,
+      "loss": 1.4206,
+      "step": 2192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 1.3979,
+      "step": 2193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0128356464016915e-07,
+      "loss": 1.4309,
+      "step": 2194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9751428573011175e-07,
+      "loss": 1.41,
+      "step": 2195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.937624339655599e-07,
+      "loss": 1.4027,
+      "step": 2196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9002801615558805e-07,
+      "loss": 1.4115,
+      "step": 2197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.863110390776348e-07,
+      "loss": 1.4181,
+      "step": 2198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8261150947748627e-07,
+      "loss": 1.4212,
+      "step": 2199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7892943406926064e-07,
+      "loss": 1.3837,
+      "step": 2200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7526481953539915e-07,
+      "loss": 1.4248,
+      "step": 2201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.716176725266607e-07,
+      "loss": 1.4061,
+      "step": 2202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.67987999662095e-07,
+      "loss": 1.4154,
+      "step": 2203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6437580752904155e-07,
+      "loss": 1.3998,
+      "step": 2204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.607811026831176e-07,
+      "loss": 1.4482,
+      "step": 2205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.572038916481979e-07,
+      "loss": 1.4039,
+      "step": 2206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5364418091641374e-07,
+      "loss": 1.4309,
+      "step": 2207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5010197694813285e-07,
+      "loss": 1.4082,
+      "step": 2208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4657728617195295e-07,
+      "loss": 1.3906,
+      "step": 2209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.430701149846849e-07,
+      "loss": 1.413,
+      "step": 2210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3958046975134494e-07,
+      "loss": 1.3905,
+      "step": 2211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.36108356805146e-07,
+      "loss": 1.3897,
+      "step": 2212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32653782447474e-07,
+      "loss": 1.4095,
+      "step": 2213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2921675294789403e-07,
+      "loss": 1.3958,
+      "step": 2214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2579727454412404e-07,
+      "loss": 1.3988,
+      "step": 2215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223953534420321e-07,
+      "loss": 1.4179,
+      "step": 2216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1901099581561846e-07,
+      "loss": 1.4273,
+      "step": 2217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1564420780701435e-07,
+      "loss": 1.4104,
+      "step": 2218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.122949955264587e-07,
+      "loss": 1.3756,
+      "step": 2219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08963365052295e-07,
+      "loss": 1.37,
+      "step": 2220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.056493224309587e-07,
+      "loss": 1.4075,
+      "step": 2221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0235287367696763e-07,
+      "loss": 1.4137,
+      "step": 2222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9907402477290515e-07,
+      "loss": 1.4585,
+      "step": 2223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9581278166941784e-07,
+      "loss": 1.3869,
+      "step": 2224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9256915028519575e-07,
+      "loss": 1.4324,
+      "step": 2225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8934313650696877e-07,
+      "loss": 1.413,
+      "step": 2226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8613474618949365e-07,
+      "loss": 1.4032,
+      "step": 2227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.829439851555427e-07,
+      "loss": 1.4349,
+      "step": 2228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7977085919589253e-07,
+      "loss": 1.3958,
+      "step": 2229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.766153740693145e-07,
+      "loss": 1.4222,
+      "step": 2230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.734775355025687e-07,
+      "loss": 1.4152,
+      "step": 2231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7035734919038104e-07,
+      "loss": 1.4263,
+      "step": 2232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.672548207954495e-07,
+      "loss": 1.3792,
+      "step": 2233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64169955948419e-07,
+      "loss": 1.4,
+      "step": 2234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6110276024788216e-07,
+      "loss": 1.4008,
+      "step": 2235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5805323926036295e-07,
+      "loss": 1.4542,
+      "step": 2236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550213985203076e-07,
+      "loss": 1.4227,
+      "step": 2237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5200724353007464e-07,
+      "loss": 1.4351,
+      "step": 2238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4901077975992836e-07,
+      "loss": 1.4141,
+      "step": 2239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4603201264802423e-07,
+      "loss": 1.4146,
+      "step": 2240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.430709476003978e-07,
+      "loss": 1.4162,
+      "step": 2241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4012758999096387e-07,
+      "loss": 1.441,
+      "step": 2242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3720194516149818e-07,
+      "loss": 1.4139,
+      "step": 2243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3429401842162802e-07,
+      "loss": 1.4115,
+      "step": 2244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3140381504882736e-07,
+      "loss": 1.3897,
+      "step": 2245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2853134028840594e-07,
+      "loss": 1.4295,
+      "step": 2246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2567659935349373e-07,
+      "loss": 1.4185,
+      "step": 2247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228395974250386e-07,
+      "loss": 1.4085,
+      "step": 2248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200203396517997e-07,
+      "loss": 1.3919,
+      "step": 2249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172188311503276e-07,
+      "loss": 1.4247,
+      "step": 2250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1443507700495968e-07,
+      "loss": 1.3747,
+      "step": 2251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.116690822678158e-07,
+      "loss": 1.4032,
+      "step": 2252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0892085195878154e-07,
+      "loss": 1.4089,
+      "step": 2253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0619039106550387e-07,
+      "loss": 1.3928,
+      "step": 2254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0347770454338111e-07,
+      "loss": 1.4199,
+      "step": 2255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0078279731555517e-07,
+      "loss": 1.4043,
+      "step": 2256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9810567427289596e-07,
+      "loss": 1.4072,
+      "step": 2257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9544634027400367e-07,
+      "loss": 1.4195,
+      "step": 2258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9280480014519098e-07,
+      "loss": 1.4257,
+      "step": 2259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9018105868047533e-07,
+      "loss": 1.4141,
+      "step": 2260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8757512064157658e-07,
+      "loss": 1.3931,
+      "step": 2261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8498699075789938e-07,
+      "loss": 1.4455,
+      "step": 2262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8241667372653316e-07,
+      "loss": 1.3767,
+      "step": 2263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7986417421223758e-07,
+      "loss": 1.401,
+      "step": 2264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7732949684743593e-07,
+      "loss": 1.408,
+      "step": 2265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7481264623220527e-07,
+      "loss": 1.4039,
+      "step": 2266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.723136269342729e-07,
+      "loss": 1.3756,
+      "step": 2267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.698324434890042e-07,
+      "loss": 1.4071,
+      "step": 2268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6736910039939159e-07,
+      "loss": 1.4027,
+      "step": 2269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6492360213605563e-07,
+      "loss": 1.4184,
+      "step": 2270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.62495953137225e-07,
+      "loss": 1.3709,
+      "step": 2271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.600861578087387e-07,
+      "loss": 1.4198,
+      "step": 2272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5769422052403172e-07,
+      "loss": 1.3929,
+      "step": 2273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5532014562413044e-07,
+      "loss": 1.4342,
+      "step": 2274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5296393741764392e-07,
+      "loss": 1.4118,
+      "step": 2275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5062560018075267e-07,
+      "loss": 1.4245,
+      "step": 2276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.483051381572076e-07,
+      "loss": 1.407,
+      "step": 2277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4600255555831667e-07,
+      "loss": 1.3888,
+      "step": 2278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4371785656294045e-07,
+      "loss": 1.4328,
+      "step": 2279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.414510453174811e-07,
+      "loss": 1.4152,
+      "step": 2280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3920212593588113e-07,
+      "loss": 1.4169,
+      "step": 2281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3697110249960564e-07,
+      "loss": 1.3852,
+      "step": 2282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3475797905764808e-07,
+      "loss": 1.4647,
+      "step": 2283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3256275962651223e-07,
+      "loss": 1.3974,
+      "step": 2284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.303854481902067e-07,
+      "loss": 1.412,
+      "step": 2285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2822604870024403e-07,
+      "loss": 1.3942,
+      "step": 2286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2608456507562706e-07,
+      "loss": 1.4105,
+      "step": 2287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2396100120284137e-07,
+      "loss": 1.4018,
+      "step": 2288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.218553609358575e-07,
+      "loss": 1.4197,
+      "step": 2289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1976764809611075e-07,
+      "loss": 1.3854,
+      "step": 2290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1769786647250148e-07,
+      "loss": 1.3888,
+      "step": 2291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1564601982139157e-07,
+      "loss": 1.4168,
+      "step": 2292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1361211186658893e-07,
+      "loss": 1.3717,
+      "step": 2293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1159614629934867e-07,
+      "loss": 1.4255,
+      "step": 2294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0959812677835968e-07,
+      "loss": 1.378,
+      "step": 2295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0761805692974469e-07,
+      "loss": 1.4519,
+      "step": 2296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0565594034704918e-07,
+      "loss": 1.4173,
+      "step": 2297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0371178059123354e-07,
+      "loss": 1.3663,
+      "step": 2298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 1.384,
+      "step": 2299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.987734564114504e-08,
+      "loss": 1.415,
+      "step": 2300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.798707740582447e-08,
+      "loss": 1.3795,
+      "step": 2301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.61147799152795e-08,
+      "loss": 1.431,
+      "step": 2302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.42604565674654e-08,
+      "loss": 1.3884,
+      "step": 2303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.242411072771463e-08,
+      "loss": 1.4211,
+      "step": 2304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.060574572873238e-08,
+      "loss": 1.4233,
+      "step": 2305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.880536487059333e-08,
+      "loss": 1.4093,
+      "step": 2306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.70229714207338e-08,
+      "loss": 1.3561,
+      "step": 2307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.525856861394177e-08,
+      "loss": 1.3608,
+      "step": 2308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.351215965235915e-08,
+      "loss": 1.4111,
+      "step": 2309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.178374770546838e-08,
+      "loss": 1.3841,
+      "step": 2310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.007333591009359e-08,
+      "loss": 1.4259,
+      "step": 2311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.838092737038617e-08,
+      "loss": 1.41,
+      "step": 2312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.670652515782917e-08,
+      "loss": 1.3852,
+      "step": 2313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.505013231122294e-08,
+      "loss": 1.4258,
+      "step": 2314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.341175183668503e-08,
+      "loss": 1.4048,
+      "step": 2315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.179138670764363e-08,
+      "loss": 1.4076,
+      "step": 2316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.018903986483083e-08,
+      "loss": 1.4028,
+      "step": 2317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.860471421627712e-08,
+      "loss": 1.4391,
+      "step": 2318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.703841263730914e-08,
+      "loss": 1.3998,
+      "step": 2319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.549013797054083e-08,
+      "loss": 1.3834,
+      "step": 2320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.395989302587113e-08,
+      "loss": 1.4418,
+      "step": 2321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.244768058047635e-08,
+      "loss": 1.4167,
+      "step": 2322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.095350337880667e-08,
+      "loss": 1.4064,
+      "step": 2323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9477364132582984e-08,
+      "loss": 1.3917,
+      "step": 2324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.801926552078563e-08,
+      "loss": 1.3823,
+      "step": 2325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.657921018965784e-08,
+      "loss": 1.4183,
+      "step": 2326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.515720075269349e-08,
+      "loss": 1.392,
+      "step": 2327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.375323979063929e-08,
+      "loss": 1.397,
+      "step": 2328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.236732985148374e-08,
+      "loss": 1.4393,
+      "step": 2329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0999473450455974e-08,
+      "loss": 1.3858,
+      "step": 2330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.964967307002244e-08,
+      "loss": 1.393,
+      "step": 2331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.831793115987582e-08,
+      "loss": 1.3798,
+      "step": 2332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7004250136940547e-08,
+      "loss": 1.4265,
+      "step": 2333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.570863238536061e-08,
+      "loss": 1.3909,
+      "step": 2334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.443108025649623e-08,
+      "loss": 1.3877,
+      "step": 2335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.317159606892496e-08,
+      "loss": 1.4167,
+      "step": 2336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1930182108430584e-08,
+      "loss": 1.3908,
+      "step": 2337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.070684062800201e-08,
+      "loss": 1.4098,
+      "step": 2338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.950157384783104e-08,
+      "loss": 1.4247,
+      "step": 2339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8314383955304624e-08,
+      "loss": 1.4006,
+      "step": 2340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.714527310500371e-08,
+      "loss": 1.4254,
+      "step": 2341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.599424341869884e-08,
+      "loss": 1.3983,
+      "step": 2342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.486129698534457e-08,
+      "loss": 1.3959,
+      "step": 2343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3746435861078394e-08,
+      "loss": 1.3785,
+      "step": 2344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.264966206921294e-08,
+      "loss": 1.3994,
+      "step": 2345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.157097760023819e-08,
+      "loss": 1.3959,
+      "step": 2346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0510384411812645e-08,
+      "loss": 1.4115,
+      "step": 2347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.946788442876103e-08,
+      "loss": 1.4223,
+      "step": 2348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8443479543073248e-08,
+      "loss": 1.3927,
+      "step": 2349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7437171613898806e-08,
+      "loss": 1.4099,
+      "step": 2350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.644896246754236e-08,
+      "loss": 1.3855,
+      "step": 2351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.547885389746485e-08,
+      "loss": 1.4119,
+      "step": 2352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4526847664273488e-08,
+      "loss": 1.4095,
+      "step": 2353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3592945495725107e-08,
+      "loss": 1.404,
+      "step": 2354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2677149086718363e-08,
+      "loss": 1.3966,
+      "step": 2355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.177946009929377e-08,
+      "loss": 1.4018,
+      "step": 2356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0899880162630336e-08,
+      "loss": 1.3969,
+      "step": 2357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0038410873038926e-08,
+      "loss": 1.3972,
+      "step": 2358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9195053793964468e-08,
+      "loss": 1.4069,
+      "step": 2359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8369810455981517e-08,
+      "loss": 1.4139,
+      "step": 2360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7562682356786488e-08,
+      "loss": 1.3698,
+      "step": 2361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.677367096120319e-08,
+      "loss": 1.4116,
+      "step": 2362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6002777701175088e-08,
+      "loss": 1.4259,
+      "step": 2363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5250003975765258e-08,
+      "loss": 1.4139,
+      "step": 2364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.451535115114866e-08,
+      "loss": 1.418,
+      "step": 2365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3798820560616544e-08,
+      "loss": 1.4217,
+      "step": 2366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3100413504570918e-08,
+      "loss": 1.4104,
+      "step": 2367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2420131250520107e-08,
+      "loss": 1.3987,
+      "step": 2368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.175797503307874e-08,
+      "loss": 1.4022,
+      "step": 2369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1113946053968872e-08,
+      "loss": 1.4085,
+      "step": 2370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.048804548200888e-08,
+      "loss": 1.4098,
+      "step": 2371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.88027445312123e-09,
+      "loss": 1.4111,
+      "step": 2372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.290634070322491e-09,
+      "loss": 1.3987,
+      "step": 2373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.719125403726658e-09,
+      "loss": 1.4192,
+      "step": 2374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.16574949054072e-09,
+      "loss": 1.4065,
+      "step": 2375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.630507335062432e-09,
+      "loss": 1.4134,
+      "step": 2376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.113399908681429e-09,
+      "loss": 1.4095,
+      "step": 2377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.614428149873675e-09,
+      "loss": 1.4146,
+      "step": 2378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1335929642014626e-09,
+      "loss": 1.4057,
+      "step": 2379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6708952243123e-09,
+      "loss": 1.3838,
+      "step": 2380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.226335769936697e-09,
+      "loss": 1.3826,
+      "step": 2381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.799915407887046e-09,
+      "loss": 1.3651,
+      "step": 2382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.39163491205652e-09,
+      "loss": 1.3958,
+      "step": 2383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.001495023414626e-09,
+      "loss": 1.408,
+      "step": 2384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6294964500116492e-09,
+      "loss": 1.4037,
+      "step": 2385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2756398669697708e-09,
+      "loss": 1.416,
+      "step": 2386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9399259164897278e-09,
+      "loss": 1.3706,
+      "step": 2387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.622355207844152e-09,
+      "loss": 1.3874,
+      "step": 2388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.322928317378681e-09,
+      "loss": 1.3735,
+      "step": 2389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0416457885108486e-09,
+      "loss": 1.3909,
+      "step": 2390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7785081317289732e-09,
+      "loss": 1.3991,
+      "step": 2391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5335158245899372e-09,
+      "loss": 1.376,
+      "step": 2392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3066693117191886e-09,
+      "loss": 1.3894,
+      "step": 2393
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0979690048107395e-09,
+      "loss": 1.4344,
+      "step": 2394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.074152826271665e-10,
+      "loss": 1.4142,
+      "step": 2395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.350084909962807e-10,
+      "loss": 1.3851,
+      "step": 2396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.807489428111268e-10,
+      "loss": 1.407,
+      "step": 2397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.446369180322041e-10,
+      "loss": 1.4032,
+      "step": 2398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2667266368080486e-10,
+      "loss": 1.4018,
+      "step": 2399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2685639384789625e-10,
+      "loss": 1.4015,
+      "step": 2400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4518828968523857e-10,
+      "loss": 1.413,
+      "step": 2401
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.166849940760557e-11,
+      "loss": 1.4091,
+      "step": 2402
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.629713829500503e-11,
+      "loss": 1.4125,
+      "step": 2403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.074288690458232e-12,
+      "loss": 1.3919,
+      "step": 2404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.387,
+      "step": 2405
+    },
+    {
+      "epoch": 1.0,
+      "step": 2405,
+      "total_flos": 6236373044428800.0,
+      "train_loss": 0.8259951042781997,
+      "train_runtime": 431319.641,
+      "train_samples_per_second": 0.892,
+      "train_steps_per_second": 0.006
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2405,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 6236373044428800.0,
+  "trial_name": null,
+  "trial_params": null
+}