diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,50143 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 8353,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 3.7323,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.7137,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.6367,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.5486,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.6136,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.5602,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.6057,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.6638,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.6412,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.6552,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.6624,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.4984,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.6175,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.6035,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.6981,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.5488,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.5959,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.6132,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.7469,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.6056,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.6174,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.5833,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.7458,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.51,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.6455,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 3.59,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 3.5142,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 3.6316,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.5492,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.5157,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.5312,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 3.6773,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 3.516,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.5887,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.552,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 3.6936,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 3.6376,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 3.4676,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 3.5527,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.6337,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.6403,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.5982,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 3.5006,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 3.5175,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.5362,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.5184,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.5962,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.5796,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.6106,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.5675,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 3.4991,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 3.6283,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 3.4481,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 3.5471,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 3.5264,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 3.5195,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 3.623,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.4824,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 3.6108,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.5415,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 3.5379,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.6074,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 3.6348,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.5117,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.6086,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 3.5531,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 3.5341,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.4518,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 3.4945,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.5467,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 3.5064,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 3.4558,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 3.6236,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 3.4112,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.4346,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 3.5229,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 3.5693,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 3.5549,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 3.5007,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.4937,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 3.6853,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 3.4964,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 3.5038,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 3.5303,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.6223,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 3.5823,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 3.4536,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 3.4622,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 3.5351,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.5045,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 3.5873,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 3.5361,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 3.6108,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 3.4927,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.5481,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.5281,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 3.5092,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 3.5297,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 3.5722,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.4638,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 3.5265,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 3.5187,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 3.5453,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 3.5852,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.5127,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 3.5489,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 3.5276,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 3.444,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 3.4195,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.601,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 3.455,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 3.5275,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 3.4837,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 3.5999,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.5134,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 3.5527,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 3.5631,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 3.471,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 3.5592,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.4264,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 3.5303,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 3.4599,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 3.55,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 3.3518,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 3.5255,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 3.5349,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 3.5229,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.56e-06,
+      "loss": 3.4806,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 3.4338,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 3.5073,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 3.5286,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64e-06,
+      "loss": 3.4175,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 3.4689,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.68e-06,
+      "loss": 3.5717,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.6018,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 3.5552,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 3.5714,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 3.533,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 3.5498,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.4394,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 3.525,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.84e-06,
+      "loss": 3.5113,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 3.5528,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 3.5723,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 3.538,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 3.4708,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 3.5312,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 3.546,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 3.5787,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 3.5079,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 3.5157,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 3.487,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 3.5814,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 3.5155,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.56,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 3.468,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 3.5285,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 3.5289,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 3.5315,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.4919,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 3.4659,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 3.4861,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 3.5563,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 3.4905,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.5122,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 3.4332,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 3.4551,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 3.5413,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 3.4896,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.495,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 3.4511,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 3.499,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 3.5349,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 3.5299,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 3.3971,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 3.5318,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 3.4478,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 3.4938,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 3.6172,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.3897,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 3.5722,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 3.5292,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 3.4581,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 3.3718,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 3.5642,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 3.451,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 3.4989,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 3.4142,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 3.3315,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.5248,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 3.5681,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 3.4855,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 3.5585,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 3.5826,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.5308,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 3.5823,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 3.4154,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 3.383,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 3.5263,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.4644,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 3.4609,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 3.4262,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 3.397,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 3.4506,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 3.4367,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 3.3999,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 3.4773,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 3.518,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.18e-06,
+      "loss": 3.5913,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.425,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.22e-06,
+      "loss": 3.4567,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.24e-06,
+      "loss": 3.4119,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 3.5247,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 3.5013,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3e-06,
+      "loss": 3.5418,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 3.4256,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.34e-06,
+      "loss": 3.4211,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 3.4789,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38e-06,
+      "loss": 3.4759,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 3.4964,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42e-06,
+      "loss": 3.4907,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 3.4452,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 3.4275,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48e-06,
+      "loss": 3.4328,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 3.38,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 3.5428,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 3.4562,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 3.5182,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.58e-06,
+      "loss": 3.4828,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.4033,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 3.539,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.5518,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.66e-06,
+      "loss": 3.4647,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 3.358,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7e-06,
+      "loss": 3.5417,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 3.4533,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 3.471,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 3.3832,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 3.4809,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.4727,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 3.4913,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 3.4193,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 3.4723,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 3.4168,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.4535,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 3.5212,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 3.4314,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 3.4973,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 3.4406,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 3.4547,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 3.5058,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 3.5257,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 3.5063,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 3.4964,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 3.4224,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 3.5307,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 3.5405,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 3.5068,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 3.5086,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 3.4756,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 3.5602,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 3.4704,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 3.4812,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 3.4746,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 3.4694,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 3.4963,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 3.5148,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 3.4962,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 3.4728,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 3.4845,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 3.4124,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 3.4928,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 3.451,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 3.4944,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 3.4845,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 3.4862,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 3.4852,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 3.4437,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 3.494,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.4525,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 3.4632,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 3.519,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 3.3807,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 3.4154,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 3.4645,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 3.4135,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 3.5097,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 3.406,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 3.4167,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 3.4954,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 3.5116,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.84e-06,
+      "loss": 3.4269,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.86e-06,
+      "loss": 3.4835,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 3.4997,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-06,
+      "loss": 3.4408,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-06,
+      "loss": 3.4042,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-06,
+      "loss": 3.4766,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 3.5762,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.98e-06,
+      "loss": 3.4793,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 3.5667,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-06,
+      "loss": 3.4949,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 3.3489,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 3.4937,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-06,
+      "loss": 3.4728,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-06,
+      "loss": 3.5092,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 3.4086,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 3.3792,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-06,
+      "loss": 3.4208,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-06,
+      "loss": 3.4187,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.4898,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 3.3771,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-06,
+      "loss": 3.495,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-06,
+      "loss": 3.4031,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 3.4497,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 3.4178,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 3.4735,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.34e-06,
+      "loss": 3.4925,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 3.433,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 3.4027,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.5192,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-06,
+      "loss": 3.5038,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 3.4815,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 3.405,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 3.4241,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.3476,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 3.4546,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 3.5562,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 3.4639,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 3.4512,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.5148,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 3.4237,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 3.4923,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 3.4351,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 3.5047,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.4776,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 3.4456,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 3.4186,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 3.5689,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 3.5343,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.4686,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 3.362,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 3.4372,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 3.506,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 3.4716,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 3.5198,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 3.3977,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 3.4112,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 3.4709,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 3.3945,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 3.4589,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 3.3037,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 3.4702,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 3.4397,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 3.4755,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 3.4415,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 3.4173,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 3.5845,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 3.4043,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 3.5315,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.3935,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 3.4366,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 3.4546,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 3.4882,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 3.4794,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 3.521,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 3.5121,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 3.4349,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 3.5197,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 3.4139,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 3.3723,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 3.3887,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 3.5284,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 3.4186,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 3.4267,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.4837,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 3.4861,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 3.4127,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 3.4252,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.58e-06,
+      "loss": 3.4129,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.3826,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 3.4218,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 3.4645,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 3.3702,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 3.4837,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7e-06,
+      "loss": 3.4199,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.72e-06,
+      "loss": 3.5145,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.74e-06,
+      "loss": 3.4834,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.76e-06,
+      "loss": 3.4762,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.78e-06,
+      "loss": 3.4538,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 3.497,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 3.4715,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 3.4676,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 3.5029,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.88e-06,
+      "loss": 3.43,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.9e-06,
+      "loss": 3.4504,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.92e-06,
+      "loss": 3.4617,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.94e-06,
+      "loss": 3.4499,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 3.4947,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 3.346,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.457,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 3.4013,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 3.5218,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.06e-06,
+      "loss": 3.3919,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.08e-06,
+      "loss": 3.4786,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.1e-06,
+      "loss": 3.4732,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 3.4729,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.14e-06,
+      "loss": 3.446,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.16e-06,
+      "loss": 3.478,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.18e-06,
+      "loss": 3.4943,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.2e-06,
+      "loss": 3.3914,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 3.5698,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.24e-06,
+      "loss": 3.5909,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26e-06,
+      "loss": 3.5257,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.28e-06,
+      "loss": 3.3866,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.3e-06,
+      "loss": 3.4127,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.32e-06,
+      "loss": 3.4817,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.34e-06,
+      "loss": 3.5687,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.36e-06,
+      "loss": 3.4527,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 3.4249,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.449,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.42e-06,
+      "loss": 3.3908,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.44e-06,
+      "loss": 3.4499,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.46e-06,
+      "loss": 3.5658,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.48e-06,
+      "loss": 3.4081,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.5e-06,
+      "loss": 3.5629,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.52e-06,
+      "loss": 3.4464,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 3.502,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 3.4991,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 3.4729,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 3.3907,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 3.4637,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 3.4289,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 3.5177,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 3.4505,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 3.4333,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 3.6156,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 3.476,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 3.532,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 3.4683,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 3.3907,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 3.3688,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 3.5018,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 3.3819,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 3.4631,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 3.4287,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 3.4252,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 3.4939,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 3.4412,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 3.4418,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 3.5411,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 3.5125,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 3.5112,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 3.5031,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 3.491,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.3861,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 3.5518,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 3.4181,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.16e-06,
+      "loss": 3.4794,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 3.4145,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.4515,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 3.51,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 3.4164,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 3.4122,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 3.488,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3e-06,
+      "loss": 3.4722,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.32e-06,
+      "loss": 3.5493,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 3.3499,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 3.5354,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38e-06,
+      "loss": 3.4701,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4e-06,
+      "loss": 3.382,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42e-06,
+      "loss": 3.3352,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 3.463,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 3.4408,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.48e-06,
+      "loss": 3.543,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.5e-06,
+      "loss": 3.4513,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.52e-06,
+      "loss": 3.5432,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.54e-06,
+      "loss": 3.4744,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.56e-06,
+      "loss": 3.4469,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58e-06,
+      "loss": 3.4394,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 3.4259,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 3.4721,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 3.4922,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66e-06,
+      "loss": 3.4895,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68e-06,
+      "loss": 3.4053,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.7e-06,
+      "loss": 3.4883,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.72e-06,
+      "loss": 3.4834,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.74e-06,
+      "loss": 3.4442,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 3.5049,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 3.5349,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.4556,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 3.4774,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.84e-06,
+      "loss": 3.4052,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.86e-06,
+      "loss": 3.4035,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.88e-06,
+      "loss": 3.4207,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9e-06,
+      "loss": 3.4196,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 3.5353,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 3.4284,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 3.4141,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 3.3429,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 3.5059,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999999599899999e-06,
+      "loss": 3.4364,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999998399600057e-06,
+      "loss": 3.5167,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99999639910037e-06,
+      "loss": 3.4654,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999993598401252e-06,
+      "loss": 3.4821,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999989997503157e-06,
+      "loss": 3.4842,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999985596406659e-06,
+      "loss": 3.4956,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999980395112462e-06,
+      "loss": 3.5023,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9999743936214e-06,
+      "loss": 3.4575,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99996759193443e-06,
+      "loss": 3.4686,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999959990052645e-06,
+      "loss": 3.5061,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99995158797726e-06,
+      "loss": 3.5496,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999942385709616e-06,
+      "loss": 3.4134,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999932383251192e-06,
+      "loss": 3.4391,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999921580603584e-06,
+      "loss": 3.4596,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999909977768524e-06,
+      "loss": 3.4648,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999897574747866e-06,
+      "loss": 3.3624,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999884371543598e-06,
+      "loss": 3.4494,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999870368157831e-06,
+      "loss": 3.3918,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999855564592807e-06,
+      "loss": 3.451,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999839960850894e-06,
+      "loss": 3.5059,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99982355693459e-06,
+      "loss": 3.4561,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99980635284652e-06,
+      "loss": 3.4667,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99978834858944e-06,
+      "loss": 3.5818,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999769544166226e-06,
+      "loss": 3.3243,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999749939579892e-06,
+      "loss": 3.4099,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999729534833574e-06,
+      "loss": 3.4348,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999708329930536e-06,
+      "loss": 3.5206,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999686324874175e-06,
+      "loss": 3.4191,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99966351966801e-06,
+      "loss": 3.5305,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999639914315693e-06,
+      "loss": 3.4442,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999615508820998e-06,
+      "loss": 3.5081,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999590303187836e-06,
+      "loss": 3.4355,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999564297420237e-06,
+      "loss": 3.4076,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999537491522365e-06,
+      "loss": 3.5814,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999509885498508e-06,
+      "loss": 3.5086,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999481479353087e-06,
+      "loss": 3.4397,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999452273090644e-06,
+      "loss": 3.4276,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999422266715857e-06,
+      "loss": 3.4327,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999391460233526e-06,
+      "loss": 3.4161,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999359853648582e-06,
+      "loss": 3.5419,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999327446966084e-06,
+      "loss": 3.5112,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999294240191218e-06,
+      "loss": 3.4719,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999260233329298e-06,
+      "loss": 3.4196,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999225426385766e-06,
+      "loss": 3.5287,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999189819366194e-06,
+      "loss": 3.4181,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99915341227628e-06,
+      "loss": 3.557,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999116205121848e-06,
+      "loss": 3.4411,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999078197908855e-06,
+      "loss": 3.4173,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999039390643386e-06,
+      "loss": 3.442,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998999783331649e-06,
+      "loss": 3.471,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998959375979982e-06,
+      "loss": 3.4069,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998918168594852e-06,
+      "loss": 3.392,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998876161182856e-06,
+      "loss": 3.4585,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998833353750714e-06,
+      "loss": 3.5137,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99878974630528e-06,
+      "loss": 3.4528,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998745338853531e-06,
+      "loss": 3.4334,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998700131402574e-06,
+      "loss": 3.4362,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998654123959645e-06,
+      "loss": 3.4802,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998607316532105e-06,
+      "loss": 3.3739,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998559709127448e-06,
+      "loss": 3.4597,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998511301753291e-06,
+      "loss": 3.6192,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99846209441738e-06,
+      "loss": 3.4244,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998412087127593e-06,
+      "loss": 3.3875,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998361279891933e-06,
+      "loss": 3.4249,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998309672718529e-06,
+      "loss": 3.4381,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99825726561564e-06,
+      "loss": 3.4812,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998204058591657e-06,
+      "loss": 3.489,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998150051655091e-06,
+      "loss": 3.4234,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99809524481459e-06,
+      "loss": 3.4915,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99803963807892e-06,
+      "loss": 3.3922,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997983231456982e-06,
+      "loss": 3.4557,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997926024957805e-06,
+      "loss": 3.5498,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997868018590542e-06,
+      "loss": 3.4711,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99780921236448e-06,
+      "loss": 3.4159,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997749606289026e-06,
+      "loss": 3.491,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997689200373723e-06,
+      "loss": 3.4376,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997627994628236e-06,
+      "loss": 3.4385,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997565989062361e-06,
+      "loss": 3.4295,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997503183686022e-06,
+      "loss": 3.3342,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997439578509268e-06,
+      "loss": 3.5081,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997375173542282e-06,
+      "loss": 3.5026,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99730996879537e-06,
+      "loss": 3.4188,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997243964278964e-06,
+      "loss": 3.4835,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997177160003633e-06,
+      "loss": 3.4757,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997109555980064e-06,
+      "loss": 3.3467,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99704115221908e-06,
+      "loss": 3.3822,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996971948731626e-06,
+      "loss": 3.4902,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996901945528776e-06,
+      "loss": 3.4689,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996831142621736e-06,
+      "loss": 3.516,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996759540021837e-06,
+      "loss": 3.4637,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996687137740537e-06,
+      "loss": 3.4993,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996613935789422e-06,
+      "loss": 3.3946,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996539934180211e-06,
+      "loss": 3.4841,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996465132924746e-06,
+      "loss": 3.4859,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996389532034995e-06,
+      "loss": 3.4509,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996313131523061e-06,
+      "loss": 3.4146,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996235931401171e-06,
+      "loss": 3.5077,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996157931681678e-06,
+      "loss": 3.5915,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996079132377065e-06,
+      "loss": 3.4893,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995999533499946e-06,
+      "loss": 3.433,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995919135063055e-06,
+      "loss": 3.5199,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995837937079264e-06,
+      "loss": 3.43,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995755939561566e-06,
+      "loss": 3.441,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995673142523083e-06,
+      "loss": 3.5547,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995589545977066e-06,
+      "loss": 3.3807,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995505149936897e-06,
+      "loss": 3.4262,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995419954416077e-06,
+      "loss": 3.4942,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995333959428245e-06,
+      "loss": 3.5442,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995247164987164e-06,
+      "loss": 3.4893,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995159571106722e-06,
+      "loss": 3.5513,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995071177800938e-06,
+      "loss": 3.4865,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99498198508396e-06,
+      "loss": 3.6033,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99489199297006e-06,
+      "loss": 3.5204,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994801201473643e-06,
+      "loss": 3.4403,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994709610609239e-06,
+      "loss": 3.2906,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994617220391502e-06,
+      "loss": 3.3889,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994524030835224e-06,
+      "loss": 3.4951,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994430041955316e-06,
+      "loss": 3.5272,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99433525376682e-06,
+      "loss": 3.5216,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994239666284908e-06,
+      "loss": 3.4712,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994143279524875e-06,
+      "loss": 3.4363,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994046093502148e-06,
+      "loss": 3.4691,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99394810823228e-06,
+      "loss": 3.4324,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993849323730952e-06,
+      "loss": 3.4555,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993749740013978e-06,
+      "loss": 3.482,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993649357097289e-06,
+      "loss": 3.535,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993548174996954e-06,
+      "loss": 3.4017,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993446193729167e-06,
+      "loss": 3.507,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993343413310246e-06,
+      "loss": 3.4624,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99323983375664e-06,
+      "loss": 3.4813,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99313545508493e-06,
+      "loss": 3.4949,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993030277311817e-06,
+      "loss": 3.4387,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992924300454134e-06,
+      "loss": 3.4701,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992817524528844e-06,
+      "loss": 3.4476,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992709949553033e-06,
+      "loss": 3.4164,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992601575543916e-06,
+      "loss": 3.4357,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992492402518843e-06,
+      "loss": 3.4761,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99238243049528e-06,
+      "loss": 3.4447,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992271659490828e-06,
+      "loss": 3.474,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992160089523218e-06,
+      "loss": 3.3904,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992047720610304e-06,
+      "loss": 3.3564,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991934552770069e-06,
+      "loss": 3.4432,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991820586020622e-06,
+      "loss": 3.4285,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991705820380207e-06,
+      "loss": 3.4953,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991590255867188e-06,
+      "loss": 3.3296,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991473892500062e-06,
+      "loss": 3.4679,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99135673029745e-06,
+      "loss": 3.4901,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991238769278102e-06,
+      "loss": 3.491,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9911200094609e-06,
+      "loss": 3.4347,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991000450864846e-06,
+      "loss": 3.5869,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990880093509076e-06,
+      "loss": 3.415,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990758937412855e-06,
+      "loss": 3.3519,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990636982595567e-06,
+      "loss": 3.4252,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990514229076734e-06,
+      "loss": 3.4442,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990390676876001e-06,
+      "loss": 3.4142,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990266326013139e-06,
+      "loss": 3.4827,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990141176508051e-06,
+      "loss": 3.461,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990015228380766e-06,
+      "loss": 3.4888,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989888481651438e-06,
+      "loss": 3.3034,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989760936340355e-06,
+      "loss": 3.5789,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989632592467928e-06,
+      "loss": 3.5585,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989503450054698e-06,
+      "loss": 3.427,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989373509121332e-06,
+      "loss": 3.422,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989242769688625e-06,
+      "loss": 3.484,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989111231777504e-06,
+      "loss": 3.4144,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988978895409015e-06,
+      "loss": 3.4756,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988845760604341e-06,
+      "loss": 3.42,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988711827384788e-06,
+      "loss": 3.5353,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98857709577179e-06,
+      "loss": 3.4882,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98844156578691e-06,
+      "loss": 3.443,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988305237451839e-06,
+      "loss": 3.427,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988168110788394e-06,
+      "loss": 3.3769,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98803018581852e-06,
+      "loss": 3.3367,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987891462564293e-06,
+      "loss": 3.4368,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987751941047911e-06,
+      "loss": 3.4922,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987611621291705e-06,
+      "loss": 3.4996,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987470503318131e-06,
+      "loss": 3.4255,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987328587149776e-06,
+      "loss": 3.404,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987185872809348e-06,
+      "loss": 3.4403,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98704236031969e-06,
+      "loss": 3.5333,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98689804970377e-06,
+      "loss": 3.5433,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986752940984682e-06,
+      "loss": 3.4658,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98660703418565e-06,
+      "loss": 3.5124,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986460329330024e-06,
+      "loss": 3.5686,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986312826441282e-06,
+      "loss": 3.3707,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986164525543035e-06,
+      "loss": 3.3715,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98601542665901e-06,
+      "loss": 3.3967,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985865529813074e-06,
+      "loss": 3.5003,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985714835029216e-06,
+      "loss": 3.4562,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98556334233155e-06,
+      "loss": 3.4592,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985411051744324e-06,
+      "loss": 3.4199,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98525796329191e-06,
+      "loss": 3.5115,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985104076998808e-06,
+      "loss": 3.4228,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984949392889647e-06,
+      "loss": 3.4145,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98479391098918e-06,
+      "loss": 3.4315,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984637631322291e-06,
+      "loss": 3.4704,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984480553913993e-06,
+      "loss": 3.4898,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984322678789424e-06,
+      "loss": 3.4852,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984164005973849e-06,
+      "loss": 3.4049,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984004535492665e-06,
+      "loss": 3.3228,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983844267371387e-06,
+      "loss": 3.4605,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983683201635672e-06,
+      "loss": 3.5113,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983521338311293e-06,
+      "loss": 3.4863,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983358677424156e-06,
+      "loss": 3.5074,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983195219000291e-06,
+      "loss": 3.4363,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983030963065862e-06,
+      "loss": 3.4471,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982865909647152e-06,
+      "loss": 3.4031,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98270005877058e-06,
+      "loss": 3.4407,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982533410462685e-06,
+      "loss": 3.3964,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982365964750141e-06,
+      "loss": 3.5565,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982197721659743e-06,
+      "loss": 3.4656,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982028681218418e-06,
+      "loss": 3.4357,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98185884345322e-06,
+      "loss": 3.442,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981688208391328e-06,
+      "loss": 3.4018,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981516776060053e-06,
+      "loss": 3.4139,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981344546486829e-06,
+      "loss": 3.4799,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98117151969922e-06,
+      "loss": 3.4209,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980997695724916e-06,
+      "loss": 3.3628,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98082307459174e-06,
+      "loss": 3.3765,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980647656327634e-06,
+      "loss": 3.459,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980471440960671e-06,
+      "loss": 3.399,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980294428519059e-06,
+      "loss": 3.5074,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98011661903112e-06,
+      "loss": 3.4064,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979938012525315e-06,
+      "loss": 3.5303,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979758609030227e-06,
+      "loss": 3.473,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979578408574567e-06,
+      "loss": 3.5364,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979397411187175e-06,
+      "loss": 3.3546,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979215616897018e-06,
+      "loss": 3.4238,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97903302573319e-06,
+      "loss": 3.4091,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978849637724912e-06,
+      "loss": 3.4749,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978665452901536e-06,
+      "loss": 3.5488,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978480471292537e-06,
+      "loss": 3.5631,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97829469292752e-06,
+      "loss": 3.3966,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978108117836214e-06,
+      "loss": 3.4665,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977920746048484e-06,
+      "loss": 3.3947,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977732577594313e-06,
+      "loss": 3.3799,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977543612503817e-06,
+      "loss": 3.5093,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977353850807236e-06,
+      "loss": 3.4693,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977163292534943e-06,
+      "loss": 3.4665,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976971937717432e-06,
+      "loss": 3.4758,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97677978638533e-06,
+      "loss": 3.4435,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976586838569386e-06,
+      "loss": 3.4935,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97639309430048e-06,
+      "loss": 3.544,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976198553609622e-06,
+      "loss": 3.4769,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976003216527943e-06,
+      "loss": 3.4413,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975807083086703e-06,
+      "loss": 3.43,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975610153317297e-06,
+      "loss": 3.4612,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975412427251237e-06,
+      "loss": 3.4125,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97521390492017e-06,
+      "loss": 3.548,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975014586355863e-06,
+      "loss": 3.4561,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97481447159022e-06,
+      "loss": 3.4965,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974613560655265e-06,
+      "loss": 3.4463,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974411853583152e-06,
+      "loss": 3.4097,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974209350406163e-06,
+      "loss": 3.4715,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974006051156705e-06,
+      "loss": 3.4638,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973801955867315e-06,
+      "loss": 3.4089,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973597064570657e-06,
+      "loss": 3.4712,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973391377299522e-06,
+      "loss": 3.4774,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973184894086824e-06,
+      "loss": 3.3332,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972977614965616e-06,
+      "loss": 3.4522,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972769539969064e-06,
+      "loss": 3.5528,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972560669130472e-06,
+      "loss": 3.4287,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972351002483267e-06,
+      "loss": 3.483,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972140540061004e-06,
+      "loss": 3.5173,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971929281897365e-06,
+      "loss": 3.4507,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971717228026161e-06,
+      "loss": 3.4027,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971504378481326e-06,
+      "loss": 3.4355,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97129073329693e-06,
+      "loss": 3.46,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97107629250716e-06,
+      "loss": 3.411,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970861056146334e-06,
+      "loss": 3.4111,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970645024248904e-06,
+      "loss": 3.3984,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970428196849439e-06,
+      "loss": 3.4433,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970210573982645e-06,
+      "loss": 3.3969,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969992155683344e-06,
+      "loss": 3.4711,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969772941986493e-06,
+      "loss": 3.4748,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96955293292718e-06,
+      "loss": 3.5071,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969332128540611e-06,
+      "loss": 3.4882,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969110528862125e-06,
+      "loss": 3.4139,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968888133927186e-06,
+      "loss": 3.4757,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968664943771386e-06,
+      "loss": 3.4277,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968440958430445e-06,
+      "loss": 3.4082,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968216177940209e-06,
+      "loss": 3.4626,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967990602336653e-06,
+      "loss": 3.5199,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967764231655877e-06,
+      "loss": 3.457,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96753706593411e-06,
+      "loss": 3.5078,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967309105207707e-06,
+      "loss": 3.5274,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967080349513152e-06,
+      "loss": 3.487,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966850798887053e-06,
+      "loss": 3.4411,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966620453366149e-06,
+      "loss": 3.488,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966389312987303e-06,
+      "loss": 3.4129,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966157377787508e-06,
+      "loss": 3.4806,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965924647803884e-06,
+      "loss": 3.5187,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965691123073676e-06,
+      "loss": 3.5157,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965456803634256e-06,
+      "loss": 3.3987,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965221689523125e-06,
+      "loss": 3.4461,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964985780777912e-06,
+      "loss": 3.4897,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96474907743637e-06,
+      "loss": 3.3755,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964511579536384e-06,
+      "loss": 3.4337,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96427328711596e-06,
+      "loss": 3.4284,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964034200213235e-06,
+      "loss": 3.4223,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963794318866474e-06,
+      "loss": 3.3943,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963553643114065e-06,
+      "loss": 3.5218,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96331217299453e-06,
+      "loss": 3.4159,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963069908546509e-06,
+      "loss": 3.5294,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962826849808778e-06,
+      "loss": 3.4027,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962582996820232e-06,
+      "loss": 3.4916,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962338349619901e-06,
+      "loss": 3.4151,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962092908246936e-06,
+      "loss": 3.4302,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96184667274062e-06,
+      "loss": 3.3748,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961599643140357e-06,
+      "loss": 3.4246,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961351819485684e-06,
+      "loss": 3.4376,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961103201816264e-06,
+      "loss": 3.4757,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960853790171881e-06,
+      "loss": 3.4087,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960603584592456e-06,
+      "loss": 3.4498,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96035258511803e-06,
+      "loss": 3.3849,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96010079178877e-06,
+      "loss": 3.4172,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959848204644978e-06,
+      "loss": 3.3709,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959594823727075e-06,
+      "loss": 3.3803,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959340649075612e-06,
+      "loss": 3.4456,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959085680731269e-06,
+      "loss": 3.4788,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95882991873485e-06,
+      "loss": 3.4455,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958573363127286e-06,
+      "loss": 3.5325,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958316013949638e-06,
+      "loss": 3.4798,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958057871243091e-06,
+      "loss": 3.4614,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95779893504896e-06,
+      "loss": 3.4467,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957539205408684e-06,
+      "loss": 3.438,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957278682363829e-06,
+      "loss": 3.3815,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95701736595609e-06,
+      "loss": 3.5374,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956755256227288e-06,
+      "loss": 3.429,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956492353219371e-06,
+      "loss": 3.4506,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956228656974415e-06,
+      "loss": 3.4486,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95596416753462e-06,
+      "loss": 3.5174,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955698884942318e-06,
+      "loss": 3.4195,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95543280923996e-06,
+      "loss": 3.5185,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955165940470133e-06,
+      "loss": 3.4406,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954898278675547e-06,
+      "loss": 3.5067,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954629823899033e-06,
+      "loss": 3.4502,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954360576183562e-06,
+      "loss": 3.4141,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954090535572218e-06,
+      "loss": 3.4661,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953819702108222e-06,
+      "loss": 3.4849,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95354807583492e-06,
+      "loss": 3.4226,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953275656795778e-06,
+      "loss": 3.4517,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953002445034395e-06,
+      "loss": 3.4427,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9527284405945e-06,
+      "loss": 3.408,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952453643519939e-06,
+      "loss": 3.4097,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952178053854694e-06,
+      "loss": 3.434,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95190167164287e-06,
+      "loss": 3.5055,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9516244969287e-06,
+      "loss": 3.4616,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95134652975654e-06,
+      "loss": 3.5104,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951067770170879e-06,
+      "loss": 3.4968,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950788218216328e-06,
+      "loss": 3.3801,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950507873937626e-06,
+      "loss": 3.4596,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950226737379641e-06,
+      "loss": 3.4662,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949944808587365e-06,
+      "loss": 3.4973,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949662087605919e-06,
+      "loss": 3.4749,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949378574480548e-06,
+      "loss": 3.4564,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949094269256627e-06,
+      "loss": 3.4263,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948809171979655e-06,
+      "loss": 3.4164,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94852328269526e-06,
+      "loss": 3.3771,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948236601449194e-06,
+      "loss": 3.5144,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947949128287342e-06,
+      "loss": 3.4646,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947660863255705e-06,
+      "loss": 3.5283,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94737180640042e-06,
+      "loss": 3.4378,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947081957767748e-06,
+      "loss": 3.4082,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946791317404077e-06,
+      "loss": 3.479,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946499885355919e-06,
+      "loss": 3.5879,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946207661669914e-06,
+      "loss": 3.4253,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945914646392833e-06,
+      "loss": 3.4285,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945620839571568e-06,
+      "loss": 3.4648,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945326241253141e-06,
+      "loss": 3.3655,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945030851484698e-06,
+      "loss": 3.4475,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944734670313514e-06,
+      "loss": 3.5574,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944437697786988e-06,
+      "loss": 3.4024,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944139933952653e-06,
+      "loss": 3.3932,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943841378858157e-06,
+      "loss": 3.4593,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943542032551284e-06,
+      "loss": 3.4374,
+      "step": 876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943241895079938e-06,
+      "loss": 3.5098,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942940966492159e-06,
+      "loss": 3.4328,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942639246836101e-06,
+      "loss": 3.3775,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942336736160056e-06,
+      "loss": 3.4238,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942033434512436e-06,
+      "loss": 3.4318,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.94172934194178e-06,
+      "loss": 3.4538,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941424458496761e-06,
+      "loss": 3.4501,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941118784226164e-06,
+      "loss": 3.277,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940812319178914e-06,
+      "loss": 3.3692,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940505063404056e-06,
+      "loss": 3.3772,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940197016950766e-06,
+      "loss": 3.4768,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93988817986834e-06,
+      "loss": 3.5333,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939578552206207e-06,
+      "loss": 3.3658,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93926813401392e-06,
+      "loss": 3.562,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938956925341154e-06,
+      "loss": 3.4844,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938644926237721e-06,
+      "loss": 3.5074,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938332136753551e-06,
+      "loss": 3.4622,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9380185569387e-06,
+      "loss": 3.4205,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937704186843356e-06,
+      "loss": 3.4511,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937389026517829e-06,
+      "loss": 3.5285,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937073076012562e-06,
+      "loss": 3.4555,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936756335378114e-06,
+      "loss": 3.4349,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936438804665178e-06,
+      "loss": 3.5152,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936120483924574e-06,
+      "loss": 3.454,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935801373207243e-06,
+      "loss": 3.4421,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935481472564257e-06,
+      "loss": 3.4753,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935160782046811e-06,
+      "loss": 3.4171,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934839301706231e-06,
+      "loss": 3.5205,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934517031593966e-06,
+      "loss": 3.5813,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934193971761593e-06,
+      "loss": 3.442,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93387012226081e-06,
+      "loss": 3.4544,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933545483143452e-06,
+      "loss": 3.4787,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933220054461469e-06,
+      "loss": 3.4773,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932893836266945e-06,
+      "loss": 3.476,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932566828612089e-06,
+      "loss": 3.4484,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932239031549234e-06,
+      "loss": 3.4735,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93191044513084e-06,
+      "loss": 3.6045,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931581069409494e-06,
+      "loss": 3.5409,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931250904437913e-06,
+      "loss": 3.4061,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930919950268932e-06,
+      "loss": 3.4147,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930588206955518e-06,
+      "loss": 3.5364,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930255674550763e-06,
+      "loss": 3.3194,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929922353107888e-06,
+      "loss": 3.3904,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929588242680234e-06,
+      "loss": 3.4281,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929253343321276e-06,
+      "loss": 3.4378,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92891765508461e-06,
+      "loss": 3.5069,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928581178023957e-06,
+      "loss": 3.3201,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92824391219317e-06,
+      "loss": 3.4504,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927905857646223e-06,
+      "loss": 3.4363,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927567014437219e-06,
+      "loss": 3.4252,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927227382620386e-06,
+      "loss": 3.42,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92688696225008e-06,
+      "loss": 3.5463,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92654575338078e-06,
+      "loss": 3.4582,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926203756067094e-06,
+      "loss": 3.4687,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925860970363755e-06,
+      "loss": 3.3861,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925517396325623e-06,
+      "loss": 3.4457,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925173034007684e-06,
+      "loss": 3.4436,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92482788346505e-06,
+      "loss": 3.4245,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924481944752956e-06,
+      "loss": 3.3796,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924135217926769e-06,
+      "loss": 3.5144,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923787703041977e-06,
+      "loss": 3.434,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923439400154198e-06,
+      "loss": 3.4023,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923090309319174e-06,
+      "loss": 3.4006,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922740430592775e-06,
+      "loss": 3.5406,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922389764030991e-06,
+      "loss": 3.5019,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922038309689946e-06,
+      "loss": 3.5793,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921686067625888e-06,
+      "loss": 3.4171,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921333037895186e-06,
+      "loss": 3.3438,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920979220554343e-06,
+      "loss": 3.4729,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92062461565998e-06,
+      "loss": 3.4514,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920269223268853e-06,
+      "loss": 3.407,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919913043437832e-06,
+      "loss": 3.3449,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919556076223926e-06,
+      "loss": 3.3368,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919198321684263e-06,
+      "loss": 3.5235,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918839779876095e-06,
+      "loss": 3.4599,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918480450856807e-06,
+      "loss": 3.4033,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918120334683902e-06,
+      "loss": 3.4156,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917759431415015e-06,
+      "loss": 3.4437,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917397741107906e-06,
+      "loss": 3.4791,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917035263820459e-06,
+      "loss": 3.3432,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916671999610685e-06,
+      "loss": 3.49,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916307948536721e-06,
+      "loss": 3.4118,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915943110656827e-06,
+      "loss": 3.4488,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915577486029397e-06,
+      "loss": 3.4815,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91521107471294e-06,
+      "loss": 3.427,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9148438767661e-06,
+      "loss": 3.4866,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914475892247642e-06,
+      "loss": 3.4915,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91410712121646e-06,
+      "loss": 3.4183,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91373756373157e-06,
+      "loss": 3.4312,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913367219852116e-06,
+      "loss": 3.559,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912996089637369e-06,
+      "loss": 3.3622,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912624173146724e-06,
+      "loss": 3.4634,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912251470439703e-06,
+      "loss": 3.3249,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911877981575953e-06,
+      "loss": 3.337,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911503706615247e-06,
+      "loss": 3.4099,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911128645617485e-06,
+      "loss": 3.509,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91075279864269e-06,
+      "loss": 3.4353,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910376165751014e-06,
+      "loss": 3.4846,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909998747002733e-06,
+      "loss": 3.5272,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90962054245825e-06,
+      "loss": 3.4054,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909241552178089e-06,
+      "loss": 3.5047,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908861776222908e-06,
+      "loss": 3.4924,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908481214653485e-06,
+      "loss": 3.499,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908099867530724e-06,
+      "loss": 3.4075,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907717734915658e-06,
+      "loss": 3.381,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90733481686944e-06,
+      "loss": 3.4361,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906951113453356e-06,
+      "loss": 3.4889,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906566624728813e-06,
+      "loss": 3.439,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906181350757343e-06,
+      "loss": 3.498,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905795291600603e-06,
+      "loss": 3.4548,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905408447320385e-06,
+      "loss": 3.5406,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905020817978593e-06,
+      "loss": 3.4937,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904632403637267e-06,
+      "loss": 3.3841,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904243204358567e-06,
+      "loss": 3.3282,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903853220204782e-06,
+      "loss": 3.5177,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903462451238324e-06,
+      "loss": 3.4643,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90307089752173e-06,
+      "loss": 3.5231,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902678559117667e-06,
+      "loss": 3.4382,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902285436088925e-06,
+      "loss": 3.5048,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901891528498416e-06,
+      "loss": 3.4726,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901496836409183e-06,
+      "loss": 3.4126,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901101359884396e-06,
+      "loss": 3.4908,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900705098987342e-06,
+      "loss": 3.5464,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900308053781438e-06,
+      "loss": 3.3729,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899910224330232e-06,
+      "loss": 3.4135,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89951161069739e-06,
+      "loss": 3.4387,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899112212946706e-06,
+      "loss": 3.4158,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8987120311421e-06,
+      "loss": 3.4746,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898311065347616e-06,
+      "loss": 3.4884,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897909315627428e-06,
+      "loss": 3.3794,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897506782045828e-06,
+      "loss": 3.4427,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897103464667237e-06,
+      "loss": 3.4142,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896699363556206e-06,
+      "loss": 3.4166,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896294478777407e-06,
+      "loss": 3.4296,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895888810395633e-06,
+      "loss": 3.4977,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895482358475812e-06,
+      "loss": 3.3701,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895075123082989e-06,
+      "loss": 3.4539,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894667104282341e-06,
+      "loss": 3.5801,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894258302139166e-06,
+      "loss": 3.4414,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893848716718888e-06,
+      "loss": 3.3728,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89343834808706e-06,
+      "loss": 3.4917,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89302719630935e-06,
+      "loss": 3.4151,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89261526145157e-06,
+      "loss": 3.375,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892202543579637e-06,
+      "loss": 3.3959,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891789042759604e-06,
+      "loss": 3.348,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891374759057653e-06,
+      "loss": 3.485,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890959692540079e-06,
+      "loss": 3.3232,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890543843273312e-06,
+      "loss": 3.3411,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890127211323908e-06,
+      "loss": 3.3334,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889709796758537e-06,
+      "loss": 3.4661,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889291599644011e-06,
+      "loss": 3.3742,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888872620047251e-06,
+      "loss": 3.3336,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888452858035313e-06,
+      "loss": 3.4934,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888032313675377e-06,
+      "loss": 3.4962,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887610987034746e-06,
+      "loss": 3.4498,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88718887818085e-06,
+      "loss": 3.4989,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88676598718124e-06,
+      "loss": 3.4374,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8863423141036e-06,
+      "loss": 3.4783,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885917859015732e-06,
+      "loss": 3.4811,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885492621985567e-06,
+      "loss": 3.4486,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885066603081156e-06,
+      "loss": 3.4852,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884639802370686e-06,
+      "loss": 3.4248,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884212219922456e-06,
+      "loss": 3.5183,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8837838558049e-06,
+      "loss": 3.4566,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883354710086573e-06,
+      "loss": 3.4332,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882924782836151e-06,
+      "loss": 3.4053,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882494074122448e-06,
+      "loss": 3.3915,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882062584014386e-06,
+      "loss": 3.4641,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881630312581027e-06,
+      "loss": 3.3927,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881197259891548e-06,
+      "loss": 3.4503,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880763426015255e-06,
+      "loss": 3.4828,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880328811021581e-06,
+      "loss": 3.3531,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87989341498008e-06,
+      "loss": 3.5039,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879457237960435e-06,
+      "loss": 3.4963,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879020280032448e-06,
+      "loss": 3.4796,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878582541266054e-06,
+      "loss": 3.3917,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878144021731304e-06,
+      "loss": 3.471,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877704721498383e-06,
+      "loss": 3.4503,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877264640637592e-06,
+      "loss": 3.3974,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876823779219367e-06,
+      "loss": 3.4711,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87638213731426e-06,
+      "loss": 3.3885,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875939714992952e-06,
+      "loss": 3.4398,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87549651232625e-06,
+      "loss": 3.4797,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87505252938508e-06,
+      "loss": 3.4345,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874607766240501e-06,
+      "loss": 3.3822,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874162222963692e-06,
+      "loss": 3.5349,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873715899625958e-06,
+      "loss": 3.4794,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873268796298725e-06,
+      "loss": 3.4708,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872820913053552e-06,
+      "loss": 3.3982,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872372249962117e-06,
+      "loss": 3.5024,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871922807096224e-06,
+      "loss": 3.42,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8714725845278e-06,
+      "loss": 3.4886,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871021582328898e-06,
+      "loss": 3.3889,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870569800571703e-06,
+      "loss": 3.4452,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87011723932851e-06,
+      "loss": 3.4709,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869663898671754e-06,
+      "loss": 3.4966,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86920977867398e-06,
+      "loss": 3.3646,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868754879407873e-06,
+      "loss": 3.3671,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86829920094623e-06,
+      "loss": 3.4836,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867842743361978e-06,
+      "loss": 3.4284,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86738550672817e-06,
+      "loss": 3.4586,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866927491117984e-06,
+      "loss": 3.4851,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866468696604719e-06,
+      "loss": 3.4371,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866009123261798e-06,
+      "loss": 3.4586,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865548771162773e-06,
+      "loss": 3.4824,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86508764038132e-06,
+      "loss": 3.4262,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86462573099124e-06,
+      "loss": 3.5027,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864163043066453e-06,
+      "loss": 3.4617,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863699576681007e-06,
+      "loss": 3.4346,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86323533190908e-06,
+      "loss": 3.4038,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862770308824964e-06,
+      "loss": 3.455,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862304507503087e-06,
+      "loss": 3.4099,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861837928017992e-06,
+      "loss": 3.417,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86137057044435e-06,
+      "loss": 3.411,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86090243485696e-06,
+      "loss": 3.3934,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86043352133074e-06,
+      "loss": 3.5157,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859963829940736e-06,
+      "loss": 3.5535,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859493360762119e-06,
+      "loss": 3.4523,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859022113870178e-06,
+      "loss": 3.4419,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858550089340335e-06,
+      "loss": 3.498,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858077287248134e-06,
+      "loss": 3.428,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857603707669238e-06,
+      "loss": 3.5312,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857129350679442e-06,
+      "loss": 3.3741,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856654216354662e-06,
+      "loss": 3.4554,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856178304770935e-06,
+      "loss": 3.5287,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855701616004432e-06,
+      "loss": 3.5844,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855224150131437e-06,
+      "loss": 3.4962,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854745907228366e-06,
+      "loss": 3.3812,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854266887371755e-06,
+      "loss": 3.4187,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85378709063827e-06,
+      "loss": 3.4578,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853306517104694e-06,
+      "loss": 3.5438,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85282516684794e-06,
+      "loss": 3.4504,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852343039945042e-06,
+      "loss": 3.4058,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851860136473161e-06,
+      "loss": 3.4865,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85137645650958e-06,
+      "loss": 3.4567,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85089200013171e-06,
+      "loss": 3.5273,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850406767417078e-06,
+      "loss": 3.4436,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849920758443343e-06,
+      "loss": 3.4824,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849433973288286e-06,
+      "loss": 3.4265,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848946412029815e-06,
+      "loss": 3.5138,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848458074745954e-06,
+      "loss": 3.4375,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84796896151486e-06,
+      "loss": 3.3763,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847479072414811e-06,
+      "loss": 3.5061,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846988407524207e-06,
+      "loss": 3.4248,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846496966921574e-06,
+      "loss": 3.4441,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846004750685563e-06,
+      "loss": 3.4428,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845511758894948e-06,
+      "loss": 3.4055,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84501799162863e-06,
+      "loss": 3.4201,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844523448965626e-06,
+      "loss": 3.3958,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844028130985086e-06,
+      "loss": 3.5022,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843532037766281e-06,
+      "loss": 3.5112,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843035169388606e-06,
+      "loss": 3.4726,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842537525931579e-06,
+      "loss": 3.3546,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.84203910747484e-06,
+      "loss": 3.435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841539914098162e-06,
+      "loss": 3.4961,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841039945881432e-06,
+      "loss": 3.4936,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840539202904664e-06,
+      "loss": 3.4724,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840037685248001e-06,
+      "loss": 3.3769,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839535392991702e-06,
+      "loss": 3.4111,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839032326216153e-06,
+      "loss": 3.397,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838528485001871e-06,
+      "loss": 3.4133,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838023869429484e-06,
+      "loss": 3.4506,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837518479579753e-06,
+      "loss": 3.4172,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837012315533561e-06,
+      "loss": 3.5324,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.836505377371916e-06,
+      "loss": 3.3667,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835997665175946e-06,
+      "loss": 3.3949,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835489179026908e-06,
+      "loss": 3.44,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834979919006176e-06,
+      "loss": 3.5089,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834469885195255e-06,
+      "loss": 3.5538,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83395907767577e-06,
+      "loss": 3.4733,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83344749652947e-06,
+      "loss": 3.4981,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83293514183823e-06,
+      "loss": 3.476,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832422013684047e-06,
+      "loss": 3.476,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83190811214904e-06,
+      "loss": 3.4814,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831393437315455e-06,
+      "loss": 3.4533,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83087798926566e-06,
+      "loss": 3.5272,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830361768082151e-06,
+      "loss": 3.3497,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82984477384754e-06,
+      "loss": 3.4944,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829327006644567e-06,
+      "loss": 3.4417,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828808466556096e-06,
+      "loss": 3.4958,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828289153665116e-06,
+      "loss": 3.4052,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827769068054736e-06,
+      "loss": 3.4835,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82724820980819e-06,
+      "loss": 3.5239,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826726579008837e-06,
+      "loss": 3.402,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82620417574016e-06,
+      "loss": 3.4595,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825681000085762e-06,
+      "loss": 3.6014,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825157052129373e-06,
+      "loss": 3.4453,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824632331954846e-06,
+      "loss": 3.401,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824106839646159e-06,
+      "loss": 3.5249,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823580575287407e-06,
+      "loss": 3.4981,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823053538962818e-06,
+      "loss": 3.4027,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822525730756736e-06,
+      "loss": 3.3974,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821997150753633e-06,
+      "loss": 3.4588,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821467799038104e-06,
+      "loss": 3.5418,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820937675694862e-06,
+      "loss": 3.5825,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820406780808753e-06,
+      "loss": 3.5065,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819875114464738e-06,
+      "loss": 3.4312,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819342676747908e-06,
+      "loss": 3.3577,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81880946774347e-06,
+      "loss": 3.5159,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818275487536764e-06,
+      "loss": 3.3959,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817740736213244e-06,
+      "loss": 3.4912,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817205213858492e-06,
+      "loss": 3.4739,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.816668920558215e-06,
+      "loss": 3.5148,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81613185639824e-06,
+      "loss": 3.4953,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815594021464521e-06,
+      "loss": 3.5575,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81505541584313e-06,
+      "loss": 3.547,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814516039620267e-06,
+      "loss": 3.4553,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813975892882252e-06,
+      "loss": 3.4238,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813434975715534e-06,
+      "loss": 3.3869,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812893288206677e-06,
+      "loss": 3.4514,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812350830442377e-06,
+      "loss": 3.4368,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811807602509444e-06,
+      "loss": 3.3842,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81126360449482e-06,
+      "loss": 3.3989,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810718836485564e-06,
+      "loss": 3.4474,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810173298568864e-06,
+      "loss": 3.4798,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809626990832024e-06,
+      "loss": 3.5064,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809079913362478e-06,
+      "loss": 3.3892,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808532066247779e-06,
+      "loss": 3.4213,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807983449575603e-06,
+      "loss": 3.4372,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807434063433754e-06,
+      "loss": 3.543,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806883907910155e-06,
+      "loss": 3.4459,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80633298309285e-06,
+      "loss": 3.3945,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80578128907001e-06,
+      "loss": 3.4252,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80522882592993e-06,
+      "loss": 3.4469,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804675593761025e-06,
+      "loss": 3.3879,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804121592651835e-06,
+      "loss": 3.3693,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803566822691022e-06,
+      "loss": 3.4368,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803011283967371e-06,
+      "loss": 3.5264,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802454976569791e-06,
+      "loss": 3.4843,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801897900587312e-06,
+      "loss": 3.407,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80134005610909e-06,
+      "loss": 3.4185,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8007814432244e-06,
+      "loss": 3.4347,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800222062022648e-06,
+      "loss": 3.4312,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799661912593352e-06,
+      "loss": 3.479,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799100995026158e-06,
+      "loss": 3.4404,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79853930941084e-06,
+      "loss": 3.4743,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797976855837285e-06,
+      "loss": 3.4562,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797413634395513e-06,
+      "loss": 3.4056,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79684964517566e-06,
+      "loss": 3.4108,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796284888267984e-06,
+      "loss": 3.4373,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795719363762872e-06,
+      "loss": 3.4819,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795153071750828e-06,
+      "loss": 3.4884,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794586012322484e-06,
+      "loss": 3.5065,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794018185568592e-06,
+      "loss": 3.3809,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793449591580024e-06,
+      "loss": 3.459,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792880230447781e-06,
+      "loss": 3.4667,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792310102262982e-06,
+      "loss": 3.4113,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791739207116872e-06,
+      "loss": 3.3776,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791167545100815e-06,
+      "loss": 3.4509,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790595116306301e-06,
+      "loss": 3.398,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790021920824942e-06,
+      "loss": 3.4288,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789447958748468e-06,
+      "loss": 3.4163,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788873230168742e-06,
+      "loss": 3.4475,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78829773517774e-06,
+      "loss": 3.5106,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787721473867565e-06,
+      "loss": 3.5021,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78714444633044e-06,
+      "loss": 3.4387,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.786566652658717e-06,
+      "loss": 3.4539,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785988092944862e-06,
+      "loss": 3.4017,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78540876728147e-06,
+      "loss": 3.5145,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784828675761255e-06,
+      "loss": 3.422,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784247818477055e-06,
+      "loss": 3.3631,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783666195521831e-06,
+      "loss": 3.4832,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783083806988665e-06,
+      "loss": 3.4826,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782500652970763e-06,
+      "loss": 3.4674,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781916733561453e-06,
+      "loss": 3.4814,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781332048854187e-06,
+      "loss": 3.3681,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780746598942535e-06,
+      "loss": 3.4219,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780160383920195e-06,
+      "loss": 3.4729,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779573403880981e-06,
+      "loss": 3.4894,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778985658918839e-06,
+      "loss": 3.4641,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778397149127826e-06,
+      "loss": 3.5063,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77780787460213e-06,
+      "loss": 3.423,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77721783543606e-06,
+      "loss": 3.5069,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776627031724045e-06,
+      "loss": 3.4093,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776035463560634e-06,
+      "loss": 3.3707,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775443131040504e-06,
+      "loss": 3.5712,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774850034258453e-06,
+      "loss": 3.4364,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774256173309397e-06,
+      "loss": 3.4633,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773661548288382e-06,
+      "loss": 3.5011,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773066159290568e-06,
+      "loss": 3.5419,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772470006411243e-06,
+      "loss": 3.4116,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771873089745816e-06,
+      "loss": 3.4218,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771275409389813e-06,
+      "loss": 3.4673,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770676965438892e-06,
+      "loss": 3.3998,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770077757988826e-06,
+      "loss": 3.4746,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769477787135511e-06,
+      "loss": 3.3867,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768877052974968e-06,
+      "loss": 3.4318,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76827555560334e-06,
+      "loss": 3.4043,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767673295116885e-06,
+      "loss": 3.4302,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767070271611995e-06,
+      "loss": 3.5324,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766466485185174e-06,
+      "loss": 3.4538,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765861935933054e-06,
+      "loss": 3.4469,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765256623952385e-06,
+      "loss": 3.4713,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764650549340043e-06,
+      "loss": 3.4924,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764043712193025e-06,
+      "loss": 3.5396,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763436112608447e-06,
+      "loss": 3.3874,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762827750683549e-06,
+      "loss": 3.5168,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762218626515696e-06,
+      "loss": 3.4562,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761608740202369e-06,
+      "loss": 3.4624,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760998091841177e-06,
+      "loss": 3.4055,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760386681529845e-06,
+      "loss": 3.4444,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759774509366227e-06,
+      "loss": 3.4108,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75916157544829e-06,
+      "loss": 3.4805,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758547879874134e-06,
+      "loss": 3.3971,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757933422741973e-06,
+      "loss": 3.4098,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757318204150141e-06,
+      "loss": 3.3888,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7567022241971e-06,
+      "loss": 3.4124,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756085482981433e-06,
+      "loss": 3.4859,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75546798060184e-06,
+      "loss": 3.3863,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75484971715715e-06,
+      "loss": 3.5092,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754230692746305e-06,
+      "loss": 3.4252,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753610907468379e-06,
+      "loss": 3.505,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752990361422559e-06,
+      "loss": 3.4501,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752369054708159e-06,
+      "loss": 3.3542,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75174698742461e-06,
+      "loss": 3.4462,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751124159671472e-06,
+      "loss": 3.4958,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75050057154842e-06,
+      "loss": 3.4726,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74987622315525e-06,
+      "loss": 3.3973,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.749251114591888e-06,
+      "loss": 3.4591,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.748625245958375e-06,
+      "loss": 3.4772,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747998617354873e-06,
+      "loss": 3.3969,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747371228881668e-06,
+      "loss": 3.4959,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74674308063917e-06,
+      "loss": 3.4723,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746114172727904e-06,
+      "loss": 3.422,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745484505248524e-06,
+      "loss": 3.3604,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744854078301799e-06,
+      "loss": 3.4577,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744222891988624e-06,
+      "loss": 3.4515,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743590946410014e-06,
+      "loss": 3.4406,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742958241667107e-06,
+      "loss": 3.3814,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742324777861156e-06,
+      "loss": 3.4431,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741690555093548e-06,
+      "loss": 3.539,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741055573465779e-06,
+      "loss": 3.4675,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740419833079471e-06,
+      "loss": 3.522,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73978333403637e-06,
+      "loss": 3.4273,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739146076438342e-06,
+      "loss": 3.444,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.738508060387371e-06,
+      "loss": 3.5294,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737869285985568e-06,
+      "loss": 3.4892,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737229753335159e-06,
+      "loss": 3.4962,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.736589462538498e-06,
+      "loss": 3.4799,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735948413698055e-06,
+      "loss": 3.5173,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735306606916426e-06,
+      "loss": 3.3864,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734664042296324e-06,
+      "loss": 3.3832,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734020719940585e-06,
+      "loss": 3.4055,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733376639952165e-06,
+      "loss": 3.5478,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732731802434146e-06,
+      "loss": 3.4571,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732086207489726e-06,
+      "loss": 3.4482,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731439855222224e-06,
+      "loss": 3.408,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730792745735084e-06,
+      "loss": 3.4433,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730144879131871e-06,
+      "loss": 3.4144,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729496255516269e-06,
+      "loss": 3.5032,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728846874992082e-06,
+      "loss": 3.4343,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728196737663237e-06,
+      "loss": 3.5621,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.727545843633783e-06,
+      "loss": 3.4534,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726894193007887e-06,
+      "loss": 3.4108,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726241785889844e-06,
+      "loss": 3.4596,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72558862238406e-06,
+      "loss": 3.548,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72493470259507e-06,
+      "loss": 3.4469,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724280026627527e-06,
+      "loss": 3.3582,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.723624594586206e-06,
+      "loss": 3.4568,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722968406575998e-06,
+      "loss": 3.2977,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722311462701927e-06,
+      "loss": 3.45,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721653763069123e-06,
+      "loss": 3.4011,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720995307782849e-06,
+      "loss": 3.4524,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720336096948482e-06,
+      "loss": 3.4182,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719676130671523e-06,
+      "loss": 3.4401,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719015409057591e-06,
+      "loss": 3.454,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718353932212432e-06,
+      "loss": 3.3309,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717691700241904e-06,
+      "loss": 3.4321,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717028713251993e-06,
+      "loss": 3.4964,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716364971348804e-06,
+      "loss": 3.4112,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715700474638562e-06,
+      "loss": 3.3768,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715035223227612e-06,
+      "loss": 3.5292,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71436921722242e-06,
+      "loss": 3.4992,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71370245672958e-06,
+      "loss": 3.4212,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71303494185579e-06,
+      "loss": 3.411,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712366672707887e-06,
+      "loss": 3.5213,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711697649392817e-06,
+      "loss": 3.5411,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711027872017652e-06,
+      "loss": 3.4248,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710357340689584e-06,
+      "loss": 3.4408,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709686055515923e-06,
+      "loss": 3.349,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709014016604102e-06,
+      "loss": 3.4897,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708341224061675e-06,
+      "loss": 3.3594,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707667677996315e-06,
+      "loss": 3.5759,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706993378515815e-06,
+      "loss": 3.4144,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706318325728092e-06,
+      "loss": 3.4472,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705642519741182e-06,
+      "loss": 3.5168,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704965960663239e-06,
+      "loss": 3.3999,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70428864860254e-06,
+      "loss": 3.4043,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703610583667482e-06,
+      "loss": 3.4451,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702931765966583e-06,
+      "loss": 3.4659,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702252195608479e-06,
+      "loss": 3.4341,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701571872701933e-06,
+      "loss": 3.5117,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70089079735582e-06,
+      "loss": 3.4087,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70020896967914e-06,
+      "loss": 3.4372,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699526389781013e-06,
+      "loss": 3.3833,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69884305777068e-06,
+      "loss": 3.5057,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698158973757499e-06,
+      "loss": 3.4334,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697474137850953e-06,
+      "loss": 3.4676,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696788550160644e-06,
+      "loss": 3.4606,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696102210796291e-06,
+      "loss": 3.4541,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695415119867737e-06,
+      "loss": 3.3958,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694727277484944e-06,
+      "loss": 3.4482,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.694038683757993e-06,
+      "loss": 3.4384,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693349338797089e-06,
+      "loss": 3.4137,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692659242712552e-06,
+      "loss": 3.4738,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691968395614826e-06,
+      "loss": 3.4842,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691276797614476e-06,
+      "loss": 3.3738,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690584448822184e-06,
+      "loss": 3.4396,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689891349348753e-06,
+      "loss": 3.4539,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689197499305104e-06,
+      "loss": 3.4431,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.688502898802287e-06,
+      "loss": 3.4487,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687807547951463e-06,
+      "loss": 3.5356,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687111446863915e-06,
+      "loss": 3.4102,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686414595651049e-06,
+      "loss": 3.4849,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685716994424386e-06,
+      "loss": 3.5349,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685018643295571e-06,
+      "loss": 3.5032,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68431954237637e-06,
+      "loss": 3.4711,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.683619691778666e-06,
+      "loss": 3.5252,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682919091614462e-06,
+      "loss": 3.3281,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682217741995885e-06,
+      "loss": 3.4831,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.681515643035175e-06,
+      "loss": 3.3791,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.6808127948447e-06,
+      "loss": 3.4366,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68010919753694e-06,
+      "loss": 3.5015,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679404851224503e-06,
+      "loss": 3.471,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678699756020109e-06,
+      "loss": 3.4715,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677993912036602e-06,
+      "loss": 3.4389,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677287319386947e-06,
+      "loss": 3.568,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676579978184224e-06,
+      "loss": 3.3938,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67587188854164e-06,
+      "loss": 3.3897,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675163050572516e-06,
+      "loss": 3.4228,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674453464390292e-06,
+      "loss": 3.4214,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.673743130108533e-06,
+      "loss": 3.4224,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67303204784092e-06,
+      "loss": 3.4993,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672320217701254e-06,
+      "loss": 3.6335,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.671607639803458e-06,
+      "loss": 3.3976,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670894314261573e-06,
+      "loss": 3.3776,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670180241189755e-06,
+      "loss": 3.4756,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66946542070229e-06,
+      "loss": 3.4847,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668749852913575e-06,
+      "loss": 3.5321,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668033537938131e-06,
+      "loss": 3.4626,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.667316475890596e-06,
+      "loss": 3.4051,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666598666885729e-06,
+      "loss": 3.4032,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.665880111038407e-06,
+      "loss": 3.4516,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66516080846363e-06,
+      "loss": 3.3528,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664440759276512e-06,
+      "loss": 3.3697,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.663719963592293e-06,
+      "loss": 3.4892,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.662998421526326e-06,
+      "loss": 3.4028,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66227613319409e-06,
+      "loss": 3.5056,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661553098711177e-06,
+      "loss": 3.4418,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660829318193304e-06,
+      "loss": 3.5077,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660104791756302e-06,
+      "loss": 3.5394,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659379519516125e-06,
+      "loss": 3.4947,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658653501588848e-06,
+      "loss": 3.4952,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657926738090662e-06,
+      "loss": 3.4654,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657199229137876e-06,
+      "loss": 3.5055,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656470974846922e-06,
+      "loss": 3.5246,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65574197533435e-06,
+      "loss": 3.4758,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65501223071683e-06,
+      "loss": 3.454,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654281741111147e-06,
+      "loss": 3.4386,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.653550506634211e-06,
+      "loss": 3.4301,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65281852740305e-06,
+      "loss": 3.4843,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.652085803534808e-06,
+      "loss": 3.403,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65135233514675e-06,
+      "loss": 3.3784,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650618122356262e-06,
+      "loss": 3.4996,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649883165280845e-06,
+      "loss": 3.4821,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649147464038124e-06,
+      "loss": 3.3953,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648411018745838e-06,
+      "loss": 3.4086,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647673829521851e-06,
+      "loss": 3.422,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64693589648414e-06,
+      "loss": 3.3787,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.646197219750803e-06,
+      "loss": 3.4592,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645457799440062e-06,
+      "loss": 3.4165,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64471763567025e-06,
+      "loss": 3.4887,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643976728559825e-06,
+      "loss": 3.4709,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64323507822736e-06,
+      "loss": 3.4649,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642492684791551e-06,
+      "loss": 3.503,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641749548371209e-06,
+      "loss": 3.3386,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641005669085266e-06,
+      "loss": 3.3897,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640261047052773e-06,
+      "loss": 3.5379,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639515682392898e-06,
+      "loss": 3.4149,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638769575224928e-06,
+      "loss": 3.4057,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638022725668273e-06,
+      "loss": 3.4281,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63727513384246e-06,
+      "loss": 3.5258,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636526799867128e-06,
+      "loss": 3.5075,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635777723862043e-06,
+      "loss": 3.4512,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63502790594709e-06,
+      "loss": 3.4674,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.634277346242267e-06,
+      "loss": 3.4037,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.633526044867694e-06,
+      "loss": 3.4241,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.632774001943608e-06,
+      "loss": 3.4179,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63202121759037e-06,
+      "loss": 3.467,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.631267691928452e-06,
+      "loss": 3.4911,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.630513425078447e-06,
+      "loss": 3.4147,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629758417161073e-06,
+      "loss": 3.4814,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629002668297157e-06,
+      "loss": 3.4511,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62824617860765e-06,
+      "loss": 3.4969,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.627488948213622e-06,
+      "loss": 3.45,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626730977236261e-06,
+      "loss": 3.4622,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.625972265796868e-06,
+      "loss": 3.5547,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.625212814016872e-06,
+      "loss": 3.4872,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.624452622017814e-06,
+      "loss": 3.479,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.623691689921356e-06,
+      "loss": 3.4273,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622930017849274e-06,
+      "loss": 3.3889,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62216760592347e-06,
+      "loss": 3.5028,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.621404454265958e-06,
+      "loss": 3.3969,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620640562998875e-06,
+      "loss": 3.3672,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619875932244471e-06,
+      "loss": 3.5079,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619110562125121e-06,
+      "loss": 3.4405,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.618344452763313e-06,
+      "loss": 3.367,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.617577604281656e-06,
+      "loss": 3.3849,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616810016802874e-06,
+      "loss": 3.4091,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616041690449816e-06,
+      "loss": 3.4619,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61527262534544e-06,
+      "loss": 3.4432,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614502821612833e-06,
+      "loss": 3.4367,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.613732279375188e-06,
+      "loss": 3.4243,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612960998755826e-06,
+      "loss": 3.4623,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612188979878183e-06,
+      "loss": 3.3748,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.611416222865813e-06,
+      "loss": 3.3892,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610642727842385e-06,
+      "loss": 3.4544,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609868494931693e-06,
+      "loss": 3.4602,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609093524257644e-06,
+      "loss": 3.4346,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.608317815944265e-06,
+      "loss": 3.4639,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607541370115698e-06,
+      "loss": 3.3786,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.606764186896206e-06,
+      "loss": 3.3997,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605986266410172e-06,
+      "loss": 3.4518,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605207608782091e-06,
+      "loss": 3.4677,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.604428214136582e-06,
+      "loss": 3.5482,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603648082598378e-06,
+      "loss": 3.3928,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602867214292331e-06,
+      "loss": 3.4455,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602085609343411e-06,
+      "loss": 3.404,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.601303267876708e-06,
+      "loss": 3.4401,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600520190017426e-06,
+      "loss": 3.3899,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59973637589089e-06,
+      "loss": 3.4275,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598951825622541e-06,
+      "loss": 3.4668,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598166539337939e-06,
+      "loss": 3.4571,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59738051716276e-06,
+      "loss": 3.4183,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596593759222798e-06,
+      "loss": 3.3425,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59580626564397e-06,
+      "loss": 3.5323,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.595018036552302e-06,
+      "loss": 3.5066,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594229072073945e-06,
+      "loss": 3.5006,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593439372335165e-06,
+      "loss": 3.4826,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592648937462344e-06,
+      "loss": 3.5072,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.591857767581981e-06,
+      "loss": 3.4418,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.5910658628207e-06,
+      "loss": 3.375,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590273223305235e-06,
+      "loss": 3.5464,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58947984916244e-06,
+      "loss": 3.4553,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.588685740519285e-06,
+      "loss": 3.4317,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587890897502863e-06,
+      "loss": 3.3942,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587095320240375e-06,
+      "loss": 3.3888,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586299008859152e-06,
+      "loss": 3.4263,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.585501963486633e-06,
+      "loss": 3.4178,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584704184250373e-06,
+      "loss": 3.5389,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583905671278053e-06,
+      "loss": 3.4746,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583106424697467e-06,
+      "loss": 3.442,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.582306444636525e-06,
+      "loss": 3.452,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581505731223255e-06,
+      "loss": 3.4175,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580704284585807e-06,
+      "loss": 3.4718,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579902104852442e-06,
+      "loss": 3.4836,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579099192151537e-06,
+      "loss": 3.4608,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578295546611597e-06,
+      "loss": 3.4161,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577491168361232e-06,
+      "loss": 3.326,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576686057529181e-06,
+      "loss": 3.5082,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.575880214244287e-06,
+      "loss": 3.5094,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57507363863552e-06,
+      "loss": 3.4573,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574266330831968e-06,
+      "loss": 3.3548,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573458290962825e-06,
+      "loss": 3.4416,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572649519157416e-06,
+      "loss": 3.4077,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571840015545174e-06,
+      "loss": 3.4209,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571029780255651e-06,
+      "loss": 3.4435,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57021881341852e-06,
+      "loss": 3.4153,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569407115163566e-06,
+      "loss": 3.4762,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.568594685620696e-06,
+      "loss": 3.4059,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.567781524919927e-06,
+      "loss": 3.3931,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.566967633191402e-06,
+      "loss": 3.439,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56615301056537e-06,
+      "loss": 3.5619,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56533765717221e-06,
+      "loss": 3.3994,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.564521573142406e-06,
+      "loss": 3.4214,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.563704758606569e-06,
+      "loss": 3.4342,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562887213695417e-06,
+      "loss": 3.5368,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562068938539792e-06,
+      "loss": 3.4806,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56124993327065e-06,
+      "loss": 3.4746,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.560430198019065e-06,
+      "loss": 3.5128,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55960973291623e-06,
+      "loss": 3.4296,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55878853809345e-06,
+      "loss": 3.3987,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557966613682148e-06,
+      "loss": 3.5154,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557143959813864e-06,
+      "loss": 3.4476,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.556320576620259e-06,
+      "loss": 3.4539,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.555496464233108e-06,
+      "loss": 3.4608,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.554671622784296e-06,
+      "loss": 3.3701,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553846052405836e-06,
+      "loss": 3.4383,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553019753229849e-06,
+      "loss": 3.4197,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.552192725388578e-06,
+      "loss": 3.5768,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55136496901438e-06,
+      "loss": 3.45,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55053648423973e-06,
+      "loss": 3.52,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.549707271197216e-06,
+      "loss": 3.4685,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.548877330019549e-06,
+      "loss": 3.4477,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.548046660839549e-06,
+      "loss": 3.3781,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.547215263790158e-06,
+      "loss": 3.3809,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546383139004433e-06,
+      "loss": 3.5361,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.545550286615548e-06,
+      "loss": 3.3637,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54471670675679e-06,
+      "loss": 3.5042,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54388239956157e-06,
+      "loss": 3.4738,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.543047365163407e-06,
+      "loss": 3.4982,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.542211603695938e-06,
+      "loss": 3.4514,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541375115292922e-06,
+      "loss": 3.4945,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.540537900088229e-06,
+      "loss": 3.5765,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.539699958215847e-06,
+      "loss": 3.4971,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.538861289809881e-06,
+      "loss": 3.3979,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53802189500455e-06,
+      "loss": 3.4681,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537181773934192e-06,
+      "loss": 3.4144,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53634092673326e-06,
+      "loss": 3.4642,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.535499353536322e-06,
+      "loss": 3.5072,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.534657054478065e-06,
+      "loss": 3.4698,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.533814029693289e-06,
+      "loss": 3.5874,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532970279316913e-06,
+      "loss": 3.4076,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53212580348397e-06,
+      "loss": 3.4145,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53128060232961e-06,
+      "loss": 3.4387,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.530434675989098e-06,
+      "loss": 3.389,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.529588024597818e-06,
+      "loss": 3.4981,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.528740648291268e-06,
+      "loss": 3.5114,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52789254720506e-06,
+      "loss": 3.3644,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527043721474927e-06,
+      "loss": 3.3932,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.526194171236711e-06,
+      "loss": 3.4383,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525343896626378e-06,
+      "loss": 3.4384,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524492897780006e-06,
+      "loss": 3.4788,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.523641174833786e-06,
+      "loss": 3.4747,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.522788727924028e-06,
+      "loss": 3.3895,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52193555718716e-06,
+      "loss": 3.4462,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52108166275972e-06,
+      "loss": 3.4555,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520227044778371e-06,
+      "loss": 3.5151,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519371703379881e-06,
+      "loss": 3.4492,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51851563870114e-06,
+      "loss": 3.4663,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.517658850879152e-06,
+      "loss": 3.4714,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516801340051039e-06,
+      "loss": 3.5025,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515943106354037e-06,
+      "loss": 3.4057,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515084149925495e-06,
+      "loss": 3.49,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514224470902885e-06,
+      "loss": 3.374,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.513364069423785e-06,
+      "loss": 3.4688,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512502945625896e-06,
+      "loss": 3.5057,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.511641099647034e-06,
+      "loss": 3.4038,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510778531625127e-06,
+      "loss": 3.5229,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50991524169822e-06,
+      "loss": 3.349,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509051230004473e-06,
+      "loss": 3.3924,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508186496682166e-06,
+      "loss": 3.5668,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507321041869688e-06,
+      "loss": 3.53,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506454865705547e-06,
+      "loss": 3.5512,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505587968328366e-06,
+      "loss": 3.4585,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.504720349876883e-06,
+      "loss": 3.4284,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503852010489952e-06,
+      "loss": 3.4213,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502982950306543e-06,
+      "loss": 3.3304,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502113169465738e-06,
+      "loss": 3.4963,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501242668106739e-06,
+      "loss": 3.4745,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500371446368859e-06,
+      "loss": 3.4467,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.49949950439153e-06,
+      "loss": 3.4021,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.498626842314296e-06,
+      "loss": 3.4413,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.497753460276821e-06,
+      "loss": 3.3838,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.496879358418878e-06,
+      "loss": 3.5439,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49600453688036e-06,
+      "loss": 3.372,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49512899580127e-06,
+      "loss": 3.2784,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494252735321735e-06,
+      "loss": 3.4251,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.493375755581988e-06,
+      "loss": 3.543,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.492498056722382e-06,
+      "loss": 3.4366,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.491619638883382e-06,
+      "loss": 3.4537,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.490740502205573e-06,
+      "loss": 3.4077,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.489860646829652e-06,
+      "loss": 3.5194,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.488980072896428e-06,
+      "loss": 3.4065,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48809878054683e-06,
+      "loss": 3.4233,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4872167699219e-06,
+      "loss": 3.3805,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.486334041162797e-06,
+      "loss": 3.396,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.485450594410787e-06,
+      "loss": 3.5485,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.484566429807263e-06,
+      "loss": 3.3498,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483681547493723e-06,
+      "loss": 3.4103,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.482795947611787e-06,
+      "loss": 3.5373,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481909630303184e-06,
+      "loss": 3.4236,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481022595709759e-06,
+      "loss": 3.4292,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.480134843973473e-06,
+      "loss": 3.4684,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479246375236407e-06,
+      "loss": 3.5541,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478357189640746e-06,
+      "loss": 3.4179,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.477467287328797e-06,
+      "loss": 3.4855,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47657666844298e-06,
+      "loss": 3.5753,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47568533312583e-06,
+      "loss": 3.4184,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.474793281519996e-06,
+      "loss": 3.5113,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.473900513768243e-06,
+      "loss": 3.4222,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.473007030013447e-06,
+      "loss": 3.4083,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.472112830398602e-06,
+      "loss": 3.4007,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47121791506682e-06,
+      "loss": 3.4593,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.470322284161315e-06,
+      "loss": 3.422,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469425937825431e-06,
+      "loss": 3.4337,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.468528876202616e-06,
+      "loss": 3.4681,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.467631099436435e-06,
+      "loss": 3.4991,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.466732607670572e-06,
+      "loss": 3.4328,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.465833401048818e-06,
+      "loss": 3.4956,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464933479715084e-06,
+      "loss": 3.3379,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464032843813394e-06,
+      "loss": 3.4235,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46313149348788e-06,
+      "loss": 3.4687,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462229428882802e-06,
+      "loss": 3.489,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.461326650142522e-06,
+      "loss": 3.4994,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.460423157411522e-06,
+      "loss": 3.4423,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.459518950834396e-06,
+      "loss": 3.3912,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.458614030555854e-06,
+      "loss": 3.4498,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45770839672072e-06,
+      "loss": 3.4755,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45680204947393e-06,
+      "loss": 3.5378,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.455894988960537e-06,
+      "loss": 3.5047,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.454987215325705e-06,
+      "loss": 3.3773,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45407872871472e-06,
+      "loss": 3.3495,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45316952927297e-06,
+      "loss": 3.5069,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.452259617145965e-06,
+      "loss": 3.263,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451348992479326e-06,
+      "loss": 3.4,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450437655418792e-06,
+      "loss": 3.4671,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449525606110214e-06,
+      "loss": 3.4337,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448612844699554e-06,
+      "loss": 3.4405,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44769937133289e-06,
+      "loss": 3.4643,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446785186156415e-06,
+      "loss": 3.4444,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.445870289316436e-06,
+      "loss": 3.4015,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444954680959376e-06,
+      "loss": 3.3983,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444038361231763e-06,
+      "loss": 3.4704,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.443121330280245e-06,
+      "loss": 3.4409,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.442203588251589e-06,
+      "loss": 3.4104,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.441285135292666e-06,
+      "loss": 3.375,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.440365971550468e-06,
+      "loss": 3.4832,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439446097172096e-06,
+      "loss": 3.4562,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.438525512304767e-06,
+      "loss": 3.4294,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437604217095813e-06,
+      "loss": 3.4053,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436682211692676e-06,
+      "loss": 3.4432,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435759496242915e-06,
+      "loss": 3.4249,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4348360708942e-06,
+      "loss": 3.4444,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433911935794317e-06,
+      "loss": 3.4205,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432987091091165e-06,
+      "loss": 3.5518,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432061536932758e-06,
+      "loss": 3.4548,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431135273467217e-06,
+      "loss": 3.3886,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.430208300842784e-06,
+      "loss": 3.4138,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42928061920781e-06,
+      "loss": 3.4839,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428352228710764e-06,
+      "loss": 3.3968,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427423129500225e-06,
+      "loss": 3.482,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426493321724883e-06,
+      "loss": 3.5258,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425562805533548e-06,
+      "loss": 3.4394,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.424631581075137e-06,
+      "loss": 3.4073,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.423699648498685e-06,
+      "loss": 3.5521,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.42276700795334e-06,
+      "loss": 3.4594,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.421833659588358e-06,
+      "loss": 3.4543,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420899603553114e-06,
+      "loss": 3.4025,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419964839997093e-06,
+      "loss": 3.4903,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419029369069897e-06,
+      "loss": 3.4365,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418093190921238e-06,
+      "loss": 3.3804,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.417156305700941e-06,
+      "loss": 3.4116,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.416218713558946e-06,
+      "loss": 3.4794,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.415280414645304e-06,
+      "loss": 3.4243,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41434140911018e-06,
+      "loss": 3.3996,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.413401697103856e-06,
+      "loss": 3.361,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41246127877672e-06,
+      "loss": 3.5103,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41152015427928e-06,
+      "loss": 3.5108,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41057832376215e-06,
+      "loss": 3.388,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40963578737606e-06,
+      "loss": 3.4303,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.408692545271858e-06,
+      "loss": 3.4657,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.407748597600497e-06,
+      "loss": 3.4598,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.406803944513046e-06,
+      "loss": 3.4879,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40585858616069e-06,
+      "loss": 3.4073,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.404912522694721e-06,
+      "loss": 3.506,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40396575426655e-06,
+      "loss": 3.4518,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.403018281027696e-06,
+      "loss": 3.4482,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402070103129792e-06,
+      "loss": 3.4896,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.401121220724587e-06,
+      "loss": 3.4018,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40017163396394e-06,
+      "loss": 3.4156,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.399221342999818e-06,
+      "loss": 3.4644,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39827034798431e-06,
+      "loss": 3.3491,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.397318649069614e-06,
+      "loss": 3.4553,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.396366246408038e-06,
+      "loss": 3.547,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.395413140152003e-06,
+      "loss": 3.3889,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.394459330454047e-06,
+      "loss": 3.4466,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.393504817466817e-06,
+      "loss": 3.4886,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.392549601343071e-06,
+      "loss": 3.4099,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391593682235685e-06,
+      "loss": 3.3917,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.390637060297642e-06,
+      "loss": 3.4266,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.389679735682042e-06,
+      "loss": 3.3341,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.388721708542093e-06,
+      "loss": 3.3559,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.387762979031119e-06,
+      "loss": 3.4511,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.386803547302554e-06,
+      "loss": 3.3536,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.385843413509947e-06,
+      "loss": 3.3773,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.384882577806957e-06,
+      "loss": 3.5101,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.383921040347355e-06,
+      "loss": 3.5355,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382958801285026e-06,
+      "loss": 3.3682,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.381995860773969e-06,
+      "loss": 3.4817,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38103221896829e-06,
+      "loss": 3.522,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.380067876022213e-06,
+      "loss": 3.4949,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.379102832090067e-06,
+      "loss": 3.4869,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378137087326302e-06,
+      "loss": 3.4965,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377170641885474e-06,
+      "loss": 3.3685,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.376203495922254e-06,
+      "loss": 3.43,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.375235649591423e-06,
+      "loss": 3.4855,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.374267103047876e-06,
+      "loss": 3.4518,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37329785644662e-06,
+      "loss": 3.4053,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37232790994277e-06,
+      "loss": 3.4544,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371357263691557e-06,
+      "loss": 3.4062,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370385917848325e-06,
+      "loss": 3.5059,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369413872568528e-06,
+      "loss": 3.4349,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368441128007733e-06,
+      "loss": 3.489,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367467684321614e-06,
+      "loss": 3.489,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.366493541665967e-06,
+      "loss": 3.4259,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365518700196687e-06,
+      "loss": 3.4479,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364543160069792e-06,
+      "loss": 3.5363,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.363566921441407e-06,
+      "loss": 3.4674,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362589984467769e-06,
+      "loss": 3.4105,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.361612349305227e-06,
+      "loss": 3.4638,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36063401611024e-06,
+      "loss": 3.3841,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.359654985039382e-06,
+      "loss": 3.473,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358675256249338e-06,
+      "loss": 3.4798,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357694829896902e-06,
+      "loss": 3.4406,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356713706138983e-06,
+      "loss": 3.437,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355731885132598e-06,
+      "loss": 3.393,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354749367034881e-06,
+      "loss": 3.3939,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35376615200307e-06,
+      "loss": 3.5137,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352782240194522e-06,
+      "loss": 3.4382,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.3517976317667e-06,
+      "loss": 3.4022,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.350812326877183e-06,
+      "loss": 3.3839,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349826325683658e-06,
+      "loss": 3.4302,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.348839628343924e-06,
+      "loss": 3.4641,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347852235015892e-06,
+      "loss": 3.4728,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346864145857588e-06,
+      "loss": 3.4139,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345875361027141e-06,
+      "loss": 3.373,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.344885880682798e-06,
+      "loss": 3.4178,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.343895704982917e-06,
+      "loss": 3.3386,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.342904834085961e-06,
+      "loss": 3.3733,
+      "step": 1796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.341913268150516e-06,
+      "loss": 3.4104,
+      "step": 1797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.340921007335264e-06,
+      "loss": 3.4607,
+      "step": 1798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339928051799014e-06,
+      "loss": 3.4646,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.338934401700675e-06,
+      "loss": 3.3956,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.337940057199271e-06,
+      "loss": 3.3255,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.336945018453936e-06,
+      "loss": 3.5551,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335949285623918e-06,
+      "loss": 3.3944,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.334952858868574e-06,
+      "loss": 3.3146,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33395573834737e-06,
+      "loss": 3.4626,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.332957924219888e-06,
+      "loss": 3.463,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.331959416645814e-06,
+      "loss": 3.3906,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.330960215784955e-06,
+      "loss": 3.4151,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.329960321797216e-06,
+      "loss": 3.3609,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.328959734842626e-06,
+      "loss": 3.4024,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.327958455081319e-06,
+      "loss": 3.381,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.326956482673534e-06,
+      "loss": 3.3776,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.325953817779633e-06,
+      "loss": 3.4419,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.324950460560078e-06,
+      "loss": 3.4332,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323946411175449e-06,
+      "loss": 3.4388,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.322941669786433e-06,
+      "loss": 3.5133,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32193623655383e-06,
+      "loss": 3.5004,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.320930111638547e-06,
+      "loss": 3.4252,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319923295201607e-06,
+      "loss": 3.4357,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31891578740414e-06,
+      "loss": 3.4924,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.317907588407385e-06,
+      "loss": 3.3848,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.316898698372699e-06,
+      "loss": 3.3988,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31588911746154e-06,
+      "loss": 3.2586,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.314878845835484e-06,
+      "loss": 3.4339,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.313867883656216e-06,
+      "loss": 3.4339,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.312856231085528e-06,
+      "loss": 3.4757,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311843888285324e-06,
+      "loss": 3.4607,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.310830855417622e-06,
+      "loss": 3.5463,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309817132644548e-06,
+      "loss": 3.3454,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.308802720128337e-06,
+      "loss": 3.5152,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.307787618031332e-06,
+      "loss": 3.4292,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306771826515996e-06,
+      "loss": 3.47,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.305755345744894e-06,
+      "loss": 3.4421,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.304738175880703e-06,
+      "loss": 3.3485,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.303720317086211e-06,
+      "loss": 3.365,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.302701769524316e-06,
+      "loss": 3.447,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.301682533358028e-06,
+      "loss": 3.4869,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.300662608750463e-06,
+      "loss": 3.3654,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299641995864851e-06,
+      "loss": 3.4111,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.298620694864531e-06,
+      "loss": 3.4128,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.297598705912952e-06,
+      "loss": 3.3602,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.296576029173672e-06,
+      "loss": 3.3426,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295552664810363e-06,
+      "loss": 3.3388,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2945286129868e-06,
+      "loss": 3.4297,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.293503873866876e-06,
+      "loss": 3.3904,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.292478447614589e-06,
+      "loss": 3.4092,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291452334394048e-06,
+      "loss": 3.3305,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290425534369473e-06,
+      "loss": 3.5816,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289398047705191e-06,
+      "loss": 3.4465,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.288369874565644e-06,
+      "loss": 3.3478,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287341015115376e-06,
+      "loss": 3.4283,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.286311469519051e-06,
+      "loss": 3.4578,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285281237941435e-06,
+      "loss": 3.3536,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284250320547408e-06,
+      "loss": 3.4696,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283218717501955e-06,
+      "loss": 3.3647,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.282186428970176e-06,
+      "loss": 3.4651,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.281153455117277e-06,
+      "loss": 3.4686,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280119796108576e-06,
+      "loss": 3.4039,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2790854521095e-06,
+      "loss": 3.4466,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278050423285584e-06,
+      "loss": 3.4367,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.277014709802477e-06,
+      "loss": 3.4498,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275978311825932e-06,
+      "loss": 3.4468,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274941229521814e-06,
+      "loss": 3.4567,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2739034630561e-06,
+      "loss": 3.5132,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272865012594872e-06,
+      "loss": 3.3075,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271825878304323e-06,
+      "loss": 3.4585,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27078606035076e-06,
+      "loss": 3.5386,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26974555890059e-06,
+      "loss": 3.44,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26870437412034e-06,
+      "loss": 3.4767,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267662506176637e-06,
+      "loss": 3.5272,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266619955236225e-06,
+      "loss": 3.4838,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265576721465952e-06,
+      "loss": 3.4731,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.264532805032776e-06,
+      "loss": 3.4392,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263488206103768e-06,
+      "loss": 3.5319,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262442924846103e-06,
+      "loss": 3.314,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261396961427071e-06,
+      "loss": 3.3777,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260350316014064e-06,
+      "loss": 3.4001,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25930298877459e-06,
+      "loss": 3.4976,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.258254979876265e-06,
+      "loss": 3.4114,
+      "step": 1879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.257206289486806e-06,
+      "loss": 3.4747,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.256156917774053e-06,
+      "loss": 3.4383,
+      "step": 1881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.255106864905942e-06,
+      "loss": 3.3796,
+      "step": 1882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.254056131050525e-06,
+      "loss": 3.4304,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.253004716375962e-06,
+      "loss": 3.3864,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.251952621050521e-06,
+      "loss": 3.4629,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25089984524258e-06,
+      "loss": 3.4475,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.249846389120622e-06,
+      "loss": 3.5317,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24879225285325e-06,
+      "loss": 3.4265,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.247737436609158e-06,
+      "loss": 3.4188,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.246681940557166e-06,
+      "loss": 3.4428,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.245625764866191e-06,
+      "loss": 3.474,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.244568909705268e-06,
+      "loss": 3.3579,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243511375243533e-06,
+      "loss": 3.478,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.242453161650234e-06,
+      "loss": 3.3808,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.241394269094725e-06,
+      "loss": 3.4412,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.240334697746479e-06,
+      "loss": 3.4127,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239274447775062e-06,
+      "loss": 3.4123,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23821351935016e-06,
+      "loss": 3.3358,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.237151912641563e-06,
+      "loss": 3.4854,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.236089627819172e-06,
+      "loss": 3.376,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235026665052992e-06,
+      "loss": 3.3874,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.233963024513142e-06,
+      "loss": 3.5177,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.232898706369847e-06,
+      "loss": 3.4077,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23183371079344e-06,
+      "loss": 3.4352,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.230768037954363e-06,
+      "loss": 3.4884,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.229701688023166e-06,
+      "loss": 3.3862,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.228634661170506e-06,
+      "loss": 3.3058,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.227566957567154e-06,
+      "loss": 3.4257,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.226498577383983e-06,
+      "loss": 3.4242,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.225429520791974e-06,
+      "loss": 3.512,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.224359787962222e-06,
+      "loss": 3.3582,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22328937906593e-06,
+      "loss": 3.4845,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.222218294274398e-06,
+      "loss": 3.4133,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.221146533759051e-06,
+      "loss": 3.459,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.220074097691407e-06,
+      "loss": 3.4088,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.219000986243103e-06,
+      "loss": 3.4072,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.217927199585876e-06,
+      "loss": 3.3751,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21685273789158e-06,
+      "loss": 3.477,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.215777601332167e-06,
+      "loss": 3.3967,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.214701790079703e-06,
+      "loss": 3.3995,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.213625304306362e-06,
+      "loss": 3.4961,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.212548144184425e-06,
+      "loss": 3.4121,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21147030988628e-06,
+      "loss": 3.4712,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.210391801584422e-06,
+      "loss": 3.4452,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.209312619451459e-06,
+      "loss": 3.3748,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2082327636601e-06,
+      "loss": 3.3744,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.207152234383165e-06,
+      "loss": 3.3897,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.206071031793584e-06,
+      "loss": 3.4294,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.204989156064395e-06,
+      "loss": 3.4489,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203906607368735e-06,
+      "loss": 3.3879,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20282338587986e-06,
+      "loss": 3.5638,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.201739491771127e-06,
+      "loss": 3.4185,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.200654925216002e-06,
+      "loss": 3.3473,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199569686388061e-06,
+      "loss": 3.4936,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198483775460983e-06,
+      "loss": 3.4251,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19739719260856e-06,
+      "loss": 3.4143,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.196309938004686e-06,
+      "loss": 3.5226,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195222011823365e-06,
+      "loss": 3.3659,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.194133414238713e-06,
+      "loss": 3.4724,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193044145424946e-06,
+      "loss": 3.475,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19195420555639e-06,
+      "loss": 3.426,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19086359480748e-06,
+      "loss": 3.431,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189772313352757e-06,
+      "loss": 3.3937,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18868036136687e-06,
+      "loss": 3.5288,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.187587739024575e-06,
+      "loss": 3.4484,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.186494446500736e-06,
+      "loss": 3.531,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.185400483970321e-06,
+      "loss": 3.3458,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18430585160841e-06,
+      "loss": 3.3513,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183210549590188e-06,
+      "loss": 3.4238,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.182114578090947e-06,
+      "loss": 3.5227,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181017937286085e-06,
+      "loss": 3.4718,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17992062735111e-06,
+      "loss": 3.4554,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178822648461634e-06,
+      "loss": 3.5118,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17772400079338e-06,
+      "loss": 3.5415,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176624684522172e-06,
+      "loss": 3.4038,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175524699823947e-06,
+      "loss": 3.4934,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.174424046874746e-06,
+      "loss": 3.5123,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.173322725850718e-06,
+      "loss": 3.4139,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172220736928117e-06,
+      "loss": 3.4965,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171118080283306e-06,
+      "loss": 3.5287,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170014756092756e-06,
+      "loss": 3.3579,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16891076453304e-06,
+      "loss": 3.4905,
+      "step": 1962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.167806105780841e-06,
+      "loss": 3.3528,
+      "step": 1963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16670078001295e-06,
+      "loss": 3.3669,
+      "step": 1964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.165594787406267e-06,
+      "loss": 3.3776,
+      "step": 1965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.164488128137788e-06,
+      "loss": 3.4592,
+      "step": 1966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.163380802384627e-06,
+      "loss": 3.4525,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.162272810323998e-06,
+      "loss": 3.5357,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.161164152133226e-06,
+      "loss": 3.3807,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16005482798974e-06,
+      "loss": 3.4045,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.158944838071078e-06,
+      "loss": 3.4981,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.157834182554879e-06,
+      "loss": 3.4654,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.156722861618894e-06,
+      "loss": 3.5281,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15561087544098e-06,
+      "loss": 3.4191,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.154498224199099e-06,
+      "loss": 3.5252,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15338490807132e-06,
+      "loss": 3.3801,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.152270927235815e-06,
+      "loss": 3.497,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15115628187087e-06,
+      "loss": 3.429,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.150040972154869e-06,
+      "loss": 3.4537,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.148924998266308e-06,
+      "loss": 3.5076,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.147808360383787e-06,
+      "loss": 3.4715,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.146691058686014e-06,
+      "loss": 3.469,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.1455730933518e-06,
+      "loss": 3.4613,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.144454464560066e-06,
+      "loss": 3.5032,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.143335172489835e-06,
+      "loss": 3.5048,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142215217320241e-06,
+      "loss": 3.4591,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.141094599230523e-06,
+      "loss": 3.2999,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.139973318400019e-06,
+      "loss": 3.4062,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.138851375008182e-06,
+      "loss": 3.393,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.137728769234569e-06,
+      "loss": 3.3343,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.13660550125884e-06,
+      "loss": 3.4826,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.135481571260765e-06,
+      "loss": 3.4647,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.134356979420216e-06,
+      "loss": 3.4296,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.13323172591717e-06,
+      "loss": 3.4725,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.13210581093172e-06,
+      "loss": 3.4902,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.130979234644051e-06,
+      "loss": 3.4108,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.129851997234462e-06,
+      "loss": 3.4993,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.128724098883357e-06,
+      "loss": 3.4248,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.127595539771243e-06,
+      "loss": 3.4091,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.126466320078738e-06,
+      "loss": 3.4833,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.125336439986559e-06,
+      "loss": 3.5011,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.124205899675532e-06,
+      "loss": 3.4656,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.123074699326591e-06,
+      "loss": 3.4339,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.121942839120772e-06,
+      "loss": 3.3622,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120810319239218e-06,
+      "loss": 3.4238,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.119677139863177e-06,
+      "loss": 3.4281,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.118543301174003e-06,
+      "loss": 3.447,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.117408803353158e-06,
+      "loss": 3.4512,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.116273646582203e-06,
+      "loss": 3.3261,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.115137831042811e-06,
+      "loss": 3.4455,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.114001356916759e-06,
+      "loss": 3.4667,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112864224385925e-06,
+      "loss": 3.5228,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111726433632299e-06,
+      "loss": 3.4033,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.11058798483797e-06,
+      "loss": 3.5399,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10944887818514e-06,
+      "loss": 3.4542,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.108309113856104e-06,
+      "loss": 3.4083,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.107168692033278e-06,
+      "loss": 3.3697,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.106027612899169e-06,
+      "loss": 3.4164,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.1048858766364e-06,
+      "loss": 3.3832,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103743483427692e-06,
+      "loss": 3.4312,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.102600433455873e-06,
+      "loss": 3.3131,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.101456726903877e-06,
+      "loss": 3.4527,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.100312363954746e-06,
+      "loss": 3.4307,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09916734479162e-06,
+      "loss": 3.4337,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098021669597748e-06,
+      "loss": 3.3943,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096875338556486e-06,
+      "loss": 3.4369,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.095728351851291e-06,
+      "loss": 3.4742,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.094580709665728e-06,
+      "loss": 3.358,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093432412183467e-06,
+      "loss": 3.4086,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092283459588278e-06,
+      "loss": 3.321,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09113385206404e-06,
+      "loss": 3.4279,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08998358979474e-06,
+      "loss": 3.4617,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.088832672964462e-06,
+      "loss": 3.4417,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0876811017574e-06,
+      "loss": 3.3324,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.086528876357851e-06,
+      "loss": 3.4941,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08537599695022e-06,
+      "loss": 3.4323,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.084222463719007e-06,
+      "loss": 3.4117,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083068276848831e-06,
+      "loss": 3.5003,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.081913436524405e-06,
+      "loss": 3.3735,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08075794293055e-06,
+      "loss": 3.5008,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.079601796252188e-06,
+      "loss": 3.3901,
+      "step": 2041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078444996674353e-06,
+      "loss": 3.468,
+      "step": 2042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07728754438218e-06,
+      "loss": 3.4287,
+      "step": 2043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076129439560903e-06,
+      "loss": 3.4949,
+      "step": 2044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.074970682395868e-06,
+      "loss": 3.5339,
+      "step": 2045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.073811273072522e-06,
+      "loss": 3.4638,
+      "step": 2046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.072651211776417e-06,
+      "loss": 3.4053,
+      "step": 2047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.07149049869321e-06,
+      "loss": 3.4015,
+      "step": 2048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.07032913400866e-06,
+      "loss": 3.4316,
+      "step": 2049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.069167117908632e-06,
+      "loss": 3.3369,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.068004450579096e-06,
+      "loss": 3.577,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.066841132206124e-06,
+      "loss": 3.3476,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.065677162975894e-06,
+      "loss": 3.4297,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.064512543074688e-06,
+      "loss": 3.3906,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.063347272688893e-06,
+      "loss": 3.4032,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.062181352004998e-06,
+      "loss": 3.5255,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.061014781209594e-06,
+      "loss": 3.4658,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.059847560489382e-06,
+      "loss": 3.4077,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058679690031165e-06,
+      "loss": 3.4675,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.057511170021846e-06,
+      "loss": 3.3845,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.056342000648438e-06,
+      "loss": 3.4815,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.055172182098051e-06,
+      "loss": 3.4507,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.054001714557905e-06,
+      "loss": 3.3784,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.052830598215323e-06,
+      "loss": 3.4613,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051658833257729e-06,
+      "loss": 3.5217,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.050486419872652e-06,
+      "loss": 3.4542,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.049313358247727e-06,
+      "loss": 3.4479,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.048139648570686e-06,
+      "loss": 3.4354,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.046965291029375e-06,
+      "loss": 3.4462,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.045790285811735e-06,
+      "loss": 3.3337,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.044614633105816e-06,
+      "loss": 3.4392,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.043438333099768e-06,
+      "loss": 3.4294,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.042261385981845e-06,
+      "loss": 3.468,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.041083791940406e-06,
+      "loss": 3.4274,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.039905551163917e-06,
+      "loss": 3.4588,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.038726663840937e-06,
+      "loss": 3.4466,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03754713016014e-06,
+      "loss": 3.5208,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0363669503103e-06,
+      "loss": 3.3996,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.035186124480287e-06,
+      "loss": 3.382,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.034004652859085e-06,
+      "loss": 3.5843,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.032822535635776e-06,
+      "loss": 3.4648,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.031639772999544e-06,
+      "loss": 3.3698,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.030456365139682e-06,
+      "loss": 3.4501,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.029272312245579e-06,
+      "loss": 3.4877,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.028087614506732e-06,
+      "loss": 3.4345,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.026902272112738e-06,
+      "loss": 3.5083,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.025716285253305e-06,
+      "loss": 3.4471,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.024529654118233e-06,
+      "loss": 3.4417,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.023342378897435e-06,
+      "loss": 3.4097,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.022154459780916e-06,
+      "loss": 3.3709,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.020965896958795e-06,
+      "loss": 3.4064,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01977669062129e-06,
+      "loss": 3.4199,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01858684095872e-06,
+      "loss": 3.3949,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01739634816151e-06,
+      "loss": 3.3819,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.016205212420187e-06,
+      "loss": 3.4607,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.015013433925375e-06,
+      "loss": 3.3768,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.013821012867812e-06,
+      "loss": 3.5312,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01262794943833e-06,
+      "loss": 3.5273,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01143424382787e-06,
+      "loss": 3.4574,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.010239896227468e-06,
+      "loss": 3.4664,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.009044906828273e-06,
+      "loss": 3.3885,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.007849275821527e-06,
+      "loss": 3.4821,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00665300339858e-06,
+      "loss": 3.4242,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.005456089750883e-06,
+      "loss": 3.452,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.004258535069989e-06,
+      "loss": 3.4388,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.003060339547559e-06,
+      "loss": 3.5157,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001861503375349e-06,
+      "loss": 3.5952,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.000662026745219e-06,
+      "loss": 3.4434,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.999461909849137e-06,
+      "loss": 3.5574,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.998261152879167e-06,
+      "loss": 3.4454,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99705975602748e-06,
+      "loss": 3.4159,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.995857719486347e-06,
+      "loss": 3.516,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994655043448142e-06,
+      "loss": 3.4126,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99345172810534e-06,
+      "loss": 3.3998,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99224777365052e-06,
+      "loss": 3.3988,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.991043180276363e-06,
+      "loss": 3.4777,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989837948175655e-06,
+      "loss": 3.3906,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.988632077541277e-06,
+      "loss": 3.569,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987425568566219e-06,
+      "loss": 3.4713,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.986218421443569e-06,
+      "loss": 3.495,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.985010636366521e-06,
+      "loss": 3.519,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983802213528367e-06,
+      "loss": 3.3406,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982593153122505e-06,
+      "loss": 3.512,
+      "step": 2123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981383455342432e-06,
+      "loss": 3.408,
+      "step": 2124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980173120381746e-06,
+      "loss": 3.3892,
+      "step": 2125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978962148434151e-06,
+      "loss": 3.3864,
+      "step": 2126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.977750539693453e-06,
+      "loss": 3.4832,
+      "step": 2127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.976538294353555e-06,
+      "loss": 3.4567,
+      "step": 2128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.975325412608465e-06,
+      "loss": 3.4861,
+      "step": 2129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.974111894652294e-06,
+      "loss": 3.4151,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.97289774067925e-06,
+      "loss": 3.4271,
+      "step": 2131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.97168295088365e-06,
+      "loss": 3.3285,
+      "step": 2132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.970467525459909e-06,
+      "loss": 3.4473,
+      "step": 2133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.969251464602542e-06,
+      "loss": 3.5435,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.968034768506167e-06,
+      "loss": 3.5085,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.966817437365503e-06,
+      "loss": 3.4745,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.965599471375375e-06,
+      "loss": 3.3813,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.964380870730705e-06,
+      "loss": 3.4798,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.963161635626518e-06,
+      "loss": 3.4364,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96194176625794e-06,
+      "loss": 3.4707,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.960721262820198e-06,
+      "loss": 3.4072,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.959500125508622e-06,
+      "loss": 3.4118,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.958278354518644e-06,
+      "loss": 3.4604,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.957055950045794e-06,
+      "loss": 3.3993,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.955832912285707e-06,
+      "loss": 3.4959,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.954609241434118e-06,
+      "loss": 3.5042,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.953384937686865e-06,
+      "loss": 3.386,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.952160001239881e-06,
+      "loss": 3.5236,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.950934432289208e-06,
+      "loss": 3.4203,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.949708231030986e-06,
+      "loss": 3.4091,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.948481397661457e-06,
+      "loss": 3.3883,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.947253932376961e-06,
+      "loss": 3.515,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.946025835373942e-06,
+      "loss": 3.3662,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.944797106848949e-06,
+      "loss": 3.4341,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94356774699862e-06,
+      "loss": 3.4891,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.942337756019708e-06,
+      "loss": 3.5199,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.941107134109057e-06,
+      "loss": 3.4633,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.939875881463619e-06,
+      "loss": 3.4252,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.93864399828044e-06,
+      "loss": 3.4018,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.937411484756675e-06,
+      "loss": 3.607,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.936178341089572e-06,
+      "loss": 3.3762,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.934944567476484e-06,
+      "loss": 3.4753,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.933710164114866e-06,
+      "loss": 3.3827,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.932475131202269e-06,
+      "loss": 3.4649,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.93123946893635e-06,
+      "loss": 3.4315,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.930003177514864e-06,
+      "loss": 3.3904,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.928766257135666e-06,
+      "loss": 3.4247,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.927528707996713e-06,
+      "loss": 3.3834,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.926290530296061e-06,
+      "loss": 3.4613,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.925051724231875e-06,
+      "loss": 3.5339,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.923812290002403e-06,
+      "loss": 3.4249,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.922572227806013e-06,
+      "loss": 3.4312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.921331537841158e-06,
+      "loss": 3.4492,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9200902203064e-06,
+      "loss": 3.3744,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.918848275400403e-06,
+      "loss": 3.4706,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.917605703321923e-06,
+      "loss": 3.4019,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.916362504269826e-06,
+      "loss": 3.3875,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.915118678443068e-06,
+      "loss": 3.4424,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.913874226040715e-06,
+      "loss": 3.5368,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.912629147261929e-06,
+      "loss": 3.5816,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911383442305971e-06,
+      "loss": 3.4139,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.910137111372204e-06,
+      "loss": 3.4064,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.908890154660091e-06,
+      "loss": 3.4259,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.907642572369193e-06,
+      "loss": 3.3302,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.906394364699175e-06,
+      "loss": 3.4774,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9051455318498e-06,
+      "loss": 3.4576,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.903896074020934e-06,
+      "loss": 3.4541,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902645991412537e-06,
+      "loss": 3.507,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.90139528422467e-06,
+      "loss": 3.5329,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9001439526575e-06,
+      "loss": 3.4749,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.898891996911291e-06,
+      "loss": 3.5004,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897639417186401e-06,
+      "loss": 3.4779,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.896386213683297e-06,
+      "loss": 3.3413,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.895132386602542e-06,
+      "loss": 3.3945,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.893877936144794e-06,
+      "loss": 3.4588,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892622862510821e-06,
+      "loss": 3.5149,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.891367165901481e-06,
+      "loss": 3.4157,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.890110846517737e-06,
+      "loss": 3.3799,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88885390456065e-06,
+      "loss": 3.3539,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887596340231383e-06,
+      "loss": 3.474,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.886338153731193e-06,
+      "loss": 3.353,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.885079345261442e-06,
+      "loss": 3.4775,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88381991502359e-06,
+      "loss": 3.4513,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882559863219196e-06,
+      "loss": 3.4803,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.881299190049919e-06,
+      "loss": 3.4469,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.880037895717517e-06,
+      "loss": 3.4432,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.878775980423847e-06,
+      "loss": 3.5574,
+      "step": 2207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.877513444370867e-06,
+      "loss": 3.5739,
+      "step": 2208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.876250287760631e-06,
+      "loss": 3.532,
+      "step": 2209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.874986510795296e-06,
+      "loss": 3.5181,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.873722113677118e-06,
+      "loss": 3.4943,
+      "step": 2211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87245709660845e-06,
+      "loss": 3.4208,
+      "step": 2212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.871191459791747e-06,
+      "loss": 3.5023,
+      "step": 2213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.869925203429556e-06,
+      "loss": 3.3763,
+      "step": 2214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868658327724537e-06,
+      "loss": 3.4841,
+      "step": 2215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.867390832879433e-06,
+      "loss": 3.3225,
+      "step": 2216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8661227190971e-06,
+      "loss": 3.4267,
+      "step": 2217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.864853986580482e-06,
+      "loss": 3.4871,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.863584635532629e-06,
+      "loss": 3.407,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.862314666156691e-06,
+      "loss": 3.3615,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.861044078655909e-06,
+      "loss": 3.47,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.859772873233629e-06,
+      "loss": 3.3649,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.858501050093297e-06,
+      "loss": 3.4641,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.857228609438453e-06,
+      "loss": 3.4122,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85595555147274e-06,
+      "loss": 3.2953,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.854681876399898e-06,
+      "loss": 3.4344,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.853407584423765e-06,
+      "loss": 3.5014,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85213267574828e-06,
+      "loss": 3.4482,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.850857150577478e-06,
+      "loss": 3.4117,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.849581009115495e-06,
+      "loss": 3.4511,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.848304251566567e-06,
+      "loss": 3.5017,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.847026878135021e-06,
+      "loss": 3.4873,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84574888902529e-06,
+      "loss": 3.4487,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.844470284441906e-06,
+      "loss": 3.5821,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.843191064589494e-06,
+      "loss": 3.3459,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.841911229672781e-06,
+      "loss": 3.4507,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.840630779896592e-06,
+      "loss": 3.4001,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.839349715465851e-06,
+      "loss": 3.4935,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.838068036585579e-06,
+      "loss": 3.3951,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.836785743460895e-06,
+      "loss": 3.4178,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.835502836297017e-06,
+      "loss": 3.4002,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.834219315299264e-06,
+      "loss": 3.4151,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.832935180673048e-06,
+      "loss": 3.4488,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.831650432623882e-06,
+      "loss": 3.5063,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.830365071357378e-06,
+      "loss": 3.4277,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.829079097079246e-06,
+      "loss": 3.4658,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.827792509995294e-06,
+      "loss": 3.4672,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.826505310311424e-06,
+      "loss": 3.4904,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.825217498233641e-06,
+      "loss": 3.466,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.823929073968048e-06,
+      "loss": 3.4002,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.822640037720842e-06,
+      "loss": 3.3105,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.821350389698322e-06,
+      "loss": 3.5171,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.820060130106884e-06,
+      "loss": 3.4262,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81876925915302e-06,
+      "loss": 3.4068,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.817477777043318e-06,
+      "loss": 3.3921,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.816185683984473e-06,
+      "loss": 3.4113,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.814892980183268e-06,
+      "loss": 3.4662,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.813599665846586e-06,
+      "loss": 3.5469,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81230574118141e-06,
+      "loss": 3.4186,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.811011206394823e-06,
+      "loss": 3.3804,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.809716061693999e-06,
+      "loss": 3.4155,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.808420307286213e-06,
+      "loss": 3.4085,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.807123943378839e-06,
+      "loss": 3.4817,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.805826970179345e-06,
+      "loss": 3.4582,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.804529387895302e-06,
+      "loss": 3.4043,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.803231196734372e-06,
+      "loss": 3.5151,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.801932396904318e-06,
+      "loss": 3.4623,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.800632988613001e-06,
+      "loss": 3.432,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.799332972068376e-06,
+      "loss": 3.414,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7980323474785e-06,
+      "loss": 3.502,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.796731115051526e-06,
+      "loss": 3.4629,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.795429274995702e-06,
+      "loss": 3.3266,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.794126827519372e-06,
+      "loss": 3.4308,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.792823772830984e-06,
+      "loss": 3.5029,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.791520111139073e-06,
+      "loss": 3.4827,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.790215842652283e-06,
+      "loss": 3.4342,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.788910967579346e-06,
+      "loss": 3.427,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.787605486129096e-06,
+      "loss": 3.4863,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78629939851046e-06,
+      "loss": 3.5004,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.784992704932467e-06,
+      "loss": 3.3429,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.783685405604238e-06,
+      "loss": 3.418,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782377500734993e-06,
+      "loss": 3.4365,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.781068990534053e-06,
+      "loss": 3.369,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.779759875210826e-06,
+      "loss": 3.4865,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778450154974827e-06,
+      "loss": 3.4431,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.777139830035662e-06,
+      "loss": 3.4684,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775828900603036e-06,
+      "loss": 3.4323,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77451736688675e-06,
+      "loss": 3.4124,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.773205229096701e-06,
+      "loss": 3.3944,
+      "step": 2289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771892487442885e-06,
+      "loss": 3.4633,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.770579142135394e-06,
+      "loss": 3.4378,
+      "step": 2291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.769265193384412e-06,
+      "loss": 3.5087,
+      "step": 2292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767950641400227e-06,
+      "loss": 3.3176,
+      "step": 2293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.766635486393218e-06,
+      "loss": 3.4231,
+      "step": 2294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765319728573864e-06,
+      "loss": 3.31,
+      "step": 2295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.764003368152736e-06,
+      "loss": 3.4,
+      "step": 2296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.762686405340509e-06,
+      "loss": 3.4102,
+      "step": 2297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.761368840347944e-06,
+      "loss": 3.3808,
+      "step": 2298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.760050673385909e-06,
+      "loss": 3.4969,
+      "step": 2299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.758731904665362e-06,
+      "loss": 3.4254,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.757412534397355e-06,
+      "loss": 3.4241,
+      "step": 2301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.756092562793046e-06,
+      "loss": 3.3945,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75477199006368e-06,
+      "loss": 3.4653,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.753450816420602e-06,
+      "loss": 3.4899,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.752129042075252e-06,
+      "loss": 3.4929,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75080666723917e-06,
+      "loss": 3.4529,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.749483692123982e-06,
+      "loss": 3.3336,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.748160116941424e-06,
+      "loss": 3.3334,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.746835941903317e-06,
+      "loss": 3.3377,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.745511167221583e-06,
+      "loss": 3.3832,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74418579310824e-06,
+      "loss": 3.4898,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.7428598197754e-06,
+      "loss": 3.4564,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.741533247435269e-06,
+      "loss": 3.5192,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.740206076300156e-06,
+      "loss": 3.4964,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.738878306582462e-06,
+      "loss": 3.4156,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.737549938494679e-06,
+      "loss": 3.435,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.736220972249402e-06,
+      "loss": 3.4045,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.734891408059319e-06,
+      "loss": 3.3937,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.733561246137209e-06,
+      "loss": 3.4245,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.732230486695956e-06,
+      "loss": 3.4312,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.730899129948534e-06,
+      "loss": 3.4696,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.729567176108011e-06,
+      "loss": 3.5284,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.728234625387556e-06,
+      "loss": 3.4558,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.726901478000428e-06,
+      "loss": 3.3692,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.725567734159986e-06,
+      "loss": 3.4604,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.724233394079679e-06,
+      "loss": 3.4494,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.722898457973058e-06,
+      "loss": 3.4463,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.721562926053763e-06,
+      "loss": 3.3796,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.720226798535538e-06,
+      "loss": 3.4125,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.718890075632212e-06,
+      "loss": 3.4696,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.717552757557714e-06,
+      "loss": 3.5178,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.716214844526072e-06,
+      "loss": 3.4181,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.714876336751404e-06,
+      "loss": 3.4299,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.713537234447924e-06,
+      "loss": 3.4937,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.71219753782994e-06,
+      "loss": 3.3706,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.710857247111861e-06,
+      "loss": 3.5165,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.709516362508186e-06,
+      "loss": 3.3761,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.708174884233509e-06,
+      "loss": 3.4121,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.706832812502521e-06,
+      "loss": 3.4521,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.705490147530006e-06,
+      "loss": 3.4669,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.704146889530846e-06,
+      "loss": 3.5158,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.702803038720016e-06,
+      "loss": 3.4264,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.701458595312583e-06,
+      "loss": 3.4818,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.700113559523716e-06,
+      "loss": 3.464,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69876793156867e-06,
+      "loss": 3.3066,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.697421711662803e-06,
+      "loss": 3.511,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.696074900021562e-06,
+      "loss": 3.5361,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.694727496860491e-06,
+      "loss": 3.5149,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.693379502395229e-06,
+      "loss": 3.4756,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.692030916841508e-06,
+      "loss": 3.4175,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.690681740415158e-06,
+      "loss": 3.3995,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689331973332099e-06,
+      "loss": 3.4163,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.687981615808347e-06,
+      "loss": 3.4125,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.686630668060015e-06,
+      "loss": 3.4755,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.68527913030331e-06,
+      "loss": 3.4528,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.683927002754528e-06,
+      "loss": 3.5118,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.682574285630068e-06,
+      "loss": 3.4529,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.681220979146415e-06,
+      "loss": 3.6052,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.679867083520153e-06,
+      "loss": 3.3955,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67851259896796e-06,
+      "loss": 3.4225,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.677157525706612e-06,
+      "loss": 3.3022,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675801863952968e-06,
+      "loss": 3.4822,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67444561392399e-06,
+      "loss": 3.447,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.673088775836735e-06,
+      "loss": 3.2917,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67173134990835e-06,
+      "loss": 3.4397,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.670373336356076e-06,
+      "loss": 3.4618,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66901473539725e-06,
+      "loss": 3.4522,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.667655547249305e-06,
+      "loss": 3.5648,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.666295772129763e-06,
+      "loss": 3.4062,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66493541025624e-06,
+      "loss": 3.4848,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.663574461846456e-06,
+      "loss": 3.4686,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66221292711821e-06,
+      "loss": 3.4727,
+      "step": 2372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660850806289404e-06,
+      "loss": 3.4555,
+      "step": 2373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.659488099578033e-06,
+      "loss": 3.3528,
+      "step": 2374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.658124807202183e-06,
+      "loss": 3.4704,
+      "step": 2375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656760929380037e-06,
+      "loss": 3.3921,
+      "step": 2376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65539646632987e-06,
+      "loss": 3.5519,
+      "step": 2377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.654031418270048e-06,
+      "loss": 3.5269,
+      "step": 2378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.652665785419036e-06,
+      "loss": 3.4453,
+      "step": 2379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.651299567995389e-06,
+      "loss": 3.4348,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.649932766217757e-06,
+      "loss": 3.547,
+      "step": 2381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.648565380304881e-06,
+      "loss": 3.4593,
+      "step": 2382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.647197410475601e-06,
+      "loss": 3.471,
+      "step": 2383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.645828856948843e-06,
+      "loss": 3.5368,
+      "step": 2384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.644459719943634e-06,
+      "loss": 3.4761,
+      "step": 2385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.643089999679086e-06,
+      "loss": 3.4353,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.641719696374413e-06,
+      "loss": 3.3858,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.640348810248917e-06,
+      "loss": 3.3611,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.638977341521997e-06,
+      "loss": 3.466,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.637605290413137e-06,
+      "loss": 3.3908,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.636232657141926e-06,
+      "loss": 3.413,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.634859441928038e-06,
+      "loss": 3.4535,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63348564499124e-06,
+      "loss": 3.3957,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.632111266551399e-06,
+      "loss": 3.3752,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.630736306828468e-06,
+      "loss": 3.3547,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.629360766042496e-06,
+      "loss": 3.3953,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.627984644413622e-06,
+      "loss": 3.411,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.626607942162085e-06,
+      "loss": 3.4887,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.62523065950821e-06,
+      "loss": 3.5024,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.623852796672418e-06,
+      "loss": 3.4837,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.62247435387522e-06,
+      "loss": 3.4775,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.621095331337226e-06,
+      "loss": 3.5324,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61971572927913e-06,
+      "loss": 3.4679,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.618335547921728e-06,
+      "loss": 3.3615,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.616954787485902e-06,
+      "loss": 3.4603,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61557344819263e-06,
+      "loss": 3.3329,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61419153026298e-06,
+      "loss": 3.497,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.612809033918113e-06,
+      "loss": 3.4666,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.611425959379288e-06,
+      "loss": 3.4919,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.610042306867847e-06,
+      "loss": 3.4437,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.608658076605234e-06,
+      "loss": 3.4304,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.607273268812981e-06,
+      "loss": 3.4609,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.60588788371271e-06,
+      "loss": 3.4666,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.60450192152614e-06,
+      "loss": 3.5747,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.60311538247508e-06,
+      "loss": 3.5285,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.601728266781434e-06,
+      "loss": 3.4233,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.60034057466719e-06,
+      "loss": 3.4326,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.598952306354438e-06,
+      "loss": 3.3904,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.597563462065357e-06,
+      "loss": 3.495,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.596174042022216e-06,
+      "loss": 3.3919,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.594784046447381e-06,
+      "loss": 3.405,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.593393475563303e-06,
+      "loss": 3.4851,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.59200232959253e-06,
+      "loss": 3.434,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.590610608757703e-06,
+      "loss": 3.3634,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.589218313281551e-06,
+      "loss": 3.3964,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.587825443386897e-06,
+      "loss": 3.4421,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.586431999296657e-06,
+      "loss": 3.4398,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.585037981233837e-06,
+      "loss": 3.4866,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583643389421535e-06,
+      "loss": 3.4607,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.582248224082944e-06,
+      "loss": 3.3825,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.580852485441345e-06,
+      "loss": 3.4097,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57945617372011e-06,
+      "loss": 3.3831,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.578059289142706e-06,
+      "loss": 3.337,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.576661831932691e-06,
+      "loss": 3.4534,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.575263802313715e-06,
+      "loss": 3.3669,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.573865200509518e-06,
+      "loss": 3.4417,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.572466026743929e-06,
+      "loss": 3.4614,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.571066281240877e-06,
+      "loss": 3.4904,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569665964224371e-06,
+      "loss": 3.3851,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.568265075918526e-06,
+      "loss": 3.4127,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.566863616547533e-06,
+      "loss": 3.4618,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565461586335686e-06,
+      "loss": 3.4532,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.564058985507362e-06,
+      "loss": 3.4564,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562655814287037e-06,
+      "loss": 3.4456,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.561252072899271e-06,
+      "loss": 3.4768,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559847761568723e-06,
+      "loss": 3.4522,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558442880520136e-06,
+      "loss": 3.4257,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.557037429978348e-06,
+      "loss": 3.4005,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555631410168286e-06,
+      "loss": 3.3375,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.554224821314971e-06,
+      "loss": 3.4137,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.552817663643513e-06,
+      "loss": 3.377,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551409937379113e-06,
+      "loss": 3.4783,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.550001642747066e-06,
+      "loss": 3.3926,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54859277997275e-06,
+      "loss": 3.4164,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.547183349281647e-06,
+      "loss": 3.3432,
+      "step": 2455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.545773350899316e-06,
+      "loss": 3.4211,
+      "step": 2456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544362785051415e-06,
+      "loss": 3.369,
+      "step": 2457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54295165196369e-06,
+      "loss": 3.4549,
+      "step": 2458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.541539951861983e-06,
+      "loss": 3.4268,
+      "step": 2459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54012768497222e-06,
+      "loss": 3.4489,
+      "step": 2460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.538714851520418e-06,
+      "loss": 3.5027,
+      "step": 2461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.537301451732688e-06,
+      "loss": 3.3972,
+      "step": 2462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.53588748583523e-06,
+      "loss": 3.3971,
+      "step": 2463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.534472954054338e-06,
+      "loss": 3.4016,
+      "step": 2464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.533057856616392e-06,
+      "loss": 3.4099,
+      "step": 2465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.531642193747865e-06,
+      "loss": 3.3862,
+      "step": 2466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.530225965675316e-06,
+      "loss": 3.3649,
+      "step": 2467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.528809172625403e-06,
+      "loss": 3.4512,
+      "step": 2468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.527391814824866e-06,
+      "loss": 3.4191,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.525973892500541e-06,
+      "loss": 3.5298,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.524555405879354e-06,
+      "loss": 3.4905,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.523136355188315e-06,
+      "loss": 3.4487,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52171674065453e-06,
+      "loss": 3.3239,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.520296562505196e-06,
+      "loss": 3.4273,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.518875820967598e-06,
+      "loss": 3.3697,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.517454516269112e-06,
+      "loss": 3.4873,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.516032648637201e-06,
+      "loss": 3.5316,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.514610218299424e-06,
+      "loss": 3.4843,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.513187225483424e-06,
+      "loss": 3.4356,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.511763670416938e-06,
+      "loss": 3.3947,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.510339553327792e-06,
+      "loss": 3.4947,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.5089148744439e-06,
+      "loss": 3.4416,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.50748963399327e-06,
+      "loss": 3.4554,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.506063832203998e-06,
+      "loss": 3.4823,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.504637469304264e-06,
+      "loss": 3.4594,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.50321054552235e-06,
+      "loss": 3.4655,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.501783061086614e-06,
+      "loss": 3.3955,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.500355016225519e-06,
+      "loss": 3.4247,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.498926411167601e-06,
+      "loss": 3.4843,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4974972461415e-06,
+      "loss": 3.5323,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.496067521375936e-06,
+      "loss": 3.4294,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.494637237099726e-06,
+      "loss": 3.519,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.493206393541767e-06,
+      "loss": 3.3981,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.491774990931054e-06,
+      "loss": 3.4354,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.49034302949667e-06,
+      "loss": 3.5093,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.488910509467785e-06,
+      "loss": 3.325,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.48747743107366e-06,
+      "loss": 3.4688,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.486043794543643e-06,
+      "loss": 3.449,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.484609600107174e-06,
+      "loss": 3.4429,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.483174847993785e-06,
+      "loss": 3.3771,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.481739538433088e-06,
+      "loss": 3.5072,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.480303671654795e-06,
+      "loss": 3.504,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.478867247888697e-06,
+      "loss": 3.3545,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.477430267364684e-06,
+      "loss": 3.4426,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.475992730312728e-06,
+      "loss": 3.5132,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.474554636962894e-06,
+      "loss": 3.5122,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.473115987545333e-06,
+      "loss": 3.4863,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.471676782290285e-06,
+      "loss": 3.2704,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.470237021428086e-06,
+      "loss": 3.4634,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468796705189147e-06,
+      "loss": 3.457,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.467355833803983e-06,
+      "loss": 3.4015,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465914407503188e-06,
+      "loss": 3.4521,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.464472426517452e-06,
+      "loss": 3.332,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.463029891077542e-06,
+      "loss": 3.5039,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.461586801414328e-06,
+      "loss": 3.4102,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46014315775876e-06,
+      "loss": 3.4972,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.458698960341877e-06,
+      "loss": 3.47,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45725420939481e-06,
+      "loss": 3.4207,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.455808905148777e-06,
+      "loss": 3.4014,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.454363047835087e-06,
+      "loss": 3.4165,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45291663768513e-06,
+      "loss": 3.3933,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45146967493039e-06,
+      "loss": 3.435,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.450022159802443e-06,
+      "loss": 3.4158,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.448574092532947e-06,
+      "loss": 3.4281,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.447125473353648e-06,
+      "loss": 3.4031,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.445676302496388e-06,
+      "loss": 3.4279,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.444226580193092e-06,
+      "loss": 3.4738,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442776306675769e-06,
+      "loss": 3.4371,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.441325482176524e-06,
+      "loss": 3.4168,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.439874106927544e-06,
+      "loss": 3.3968,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.438422181161112e-06,
+      "loss": 3.4046,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436969705109593e-06,
+      "loss": 3.3263,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.435516679005437e-06,
+      "loss": 3.3665,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.434063103081189e-06,
+      "loss": 3.4316,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43260897756948e-06,
+      "loss": 3.4289,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43115430270303e-06,
+      "loss": 3.5127,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.429699078714642e-06,
+      "loss": 3.4663,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42824330583721e-06,
+      "loss": 3.4903,
+      "step": 2538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.426786984303717e-06,
+      "loss": 3.3975,
+      "step": 2539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.425330114347233e-06,
+      "loss": 3.4041,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.423872696200914e-06,
+      "loss": 3.3546,
+      "step": 2541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.422414730098007e-06,
+      "loss": 3.3791,
+      "step": 2542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.420956216271842e-06,
+      "loss": 3.461,
+      "step": 2543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.419497154955845e-06,
+      "loss": 3.4097,
+      "step": 2544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.41803754638352e-06,
+      "loss": 3.4262,
+      "step": 2545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.416577390788461e-06,
+      "loss": 3.4608,
+      "step": 2546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.415116688404356e-06,
+      "loss": 3.4284,
+      "step": 2547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.413655439464973e-06,
+      "loss": 3.4209,
+      "step": 2548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.412193644204172e-06,
+      "loss": 3.4181,
+      "step": 2549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.410731302855897e-06,
+      "loss": 3.4134,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.409268415654183e-06,
+      "loss": 3.4589,
+      "step": 2551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.407804982833147e-06,
+      "loss": 3.4555,
+      "step": 2552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.406341004627e-06,
+      "loss": 3.5406,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.404876481270037e-06,
+      "loss": 3.4741,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.40341141299664e-06,
+      "loss": 3.4461,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.401945800041275e-06,
+      "loss": 3.4755,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.400479642638504e-06,
+      "loss": 3.4722,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.399012941022969e-06,
+      "loss": 3.297,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.397545695429398e-06,
+      "loss": 3.3603,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396077906092614e-06,
+      "loss": 3.4172,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.394609573247518e-06,
+      "loss": 3.4828,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.393140697129104e-06,
+      "loss": 3.3931,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.391671277972451e-06,
+      "loss": 3.4452,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.390201316012723e-06,
+      "loss": 3.4101,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.388730811485174e-06,
+      "loss": 3.4024,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.387259764625144e-06,
+      "loss": 3.5652,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.385788175668058e-06,
+      "loss": 3.4493,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.38431604484943e-06,
+      "loss": 3.363,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.38284337240486e-06,
+      "loss": 3.386,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.381370158570033e-06,
+      "loss": 3.4869,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.379896403580725e-06,
+      "loss": 3.2959,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.378422107672795e-06,
+      "loss": 3.3993,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.376947271082186e-06,
+      "loss": 3.3794,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.375471894044935e-06,
+      "loss": 3.53,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.373995976797158e-06,
+      "loss": 3.4547,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.372519519575063e-06,
+      "loss": 3.4139,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.371042522614942e-06,
+      "loss": 3.4206,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.369564986153174e-06,
+      "loss": 3.3342,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.368086910426222e-06,
+      "loss": 3.3832,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.366608295670639e-06,
+      "loss": 3.3498,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.365129142123062e-06,
+      "loss": 3.4435,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.363649450020216e-06,
+      "loss": 3.5072,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.362169219598905e-06,
+      "loss": 3.3807,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.360688451096032e-06,
+      "loss": 3.4346,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.359207144748578e-06,
+      "loss": 3.3293,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35772530079361e-06,
+      "loss": 3.3629,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35624291946828e-06,
+      "loss": 3.4565,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354760001009833e-06,
+      "loss": 3.3354,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.353276545655593e-06,
+      "loss": 3.4341,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.351792553642972e-06,
+      "loss": 3.3975,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.350308025209468e-06,
+      "loss": 3.4339,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.348822960592665e-06,
+      "loss": 3.5572,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.347337360030233e-06,
+      "loss": 3.4212,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.345851223759928e-06,
+      "loss": 3.4273,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.344364552019589e-06,
+      "loss": 3.5332,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.342877345047146e-06,
+      "loss": 3.317,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34138960308061e-06,
+      "loss": 3.4307,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.33990132635808e-06,
+      "loss": 3.3612,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.338412515117738e-06,
+      "loss": 3.4417,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.336923169597857e-06,
+      "loss": 3.3826,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.335433290036787e-06,
+      "loss": 3.4767,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333942876672973e-06,
+      "loss": 3.4867,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.332451929744937e-06,
+      "loss": 3.4264,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.330960449491292e-06,
+      "loss": 3.4841,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.329468436150734e-06,
+      "loss": 3.4015,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.327975889962046e-06,
+      "loss": 3.3915,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.326482811164093e-06,
+      "loss": 3.4806,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32498919999583e-06,
+      "loss": 3.3861,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.323495056696291e-06,
+      "loss": 3.5215,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.322000381504603e-06,
+      "loss": 3.4804,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.320505174659969e-06,
+      "loss": 3.4255,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319009436401685e-06,
+      "loss": 3.4499,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.317513166969129e-06,
+      "loss": 3.4572,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.316016366601765e-06,
+      "loss": 3.4677,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.314519035539138e-06,
+      "loss": 3.3661,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.31302117402088e-06,
+      "loss": 3.4021,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311522782286713e-06,
+      "loss": 3.3969,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310023860576438e-06,
+      "loss": 3.4175,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30852440912994e-06,
+      "loss": 3.4809,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.307024428187194e-06,
+      "loss": 3.4567,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305523917988257e-06,
+      "loss": 3.3783,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30402287877327e-06,
+      "loss": 3.5218,
+      "step": 2622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302521310782456e-06,
+      "loss": 3.485,
+      "step": 2623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301019214256131e-06,
+      "loss": 3.43,
+      "step": 2624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29951658943469e-06,
+      "loss": 3.4157,
+      "step": 2625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.298013436558609e-06,
+      "loss": 3.4151,
+      "step": 2626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296509755868457e-06,
+      "loss": 3.4753,
+      "step": 2627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29500554760488e-06,
+      "loss": 3.4286,
+      "step": 2628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.293500812008615e-06,
+      "loss": 3.4584,
+      "step": 2629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.291995549320476e-06,
+      "loss": 3.5257,
+      "step": 2630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.290489759781365e-06,
+      "loss": 3.4881,
+      "step": 2631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.288983443632274e-06,
+      "loss": 3.5049,
+      "step": 2632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.287476601114269e-06,
+      "loss": 3.3847,
+      "step": 2633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.285969232468506e-06,
+      "loss": 3.3573,
+      "step": 2634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.284461337936225e-06,
+      "loss": 3.4686,
+      "step": 2635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.282952917758749e-06,
+      "loss": 3.4308,
+      "step": 2636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.281443972177485e-06,
+      "loss": 3.3719,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.279934501433926e-06,
+      "loss": 3.3761,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.278424505769647e-06,
+      "loss": 3.4916,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.276913985426307e-06,
+      "loss": 3.4601,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.27540294064565e-06,
+      "loss": 3.3578,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.273891371669505e-06,
+      "loss": 3.4077,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.272379278739783e-06,
+      "loss": 3.4093,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.270866662098476e-06,
+      "loss": 3.4466,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.269353521987668e-06,
+      "loss": 3.386,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.267839858649522e-06,
+      "loss": 3.5648,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.26632567232628e-06,
+      "loss": 3.4124,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.264810963260276e-06,
+      "loss": 3.3992,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.263295731693923e-06,
+      "loss": 3.4205,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.261779977869719e-06,
+      "loss": 3.4483,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.260263702030244e-06,
+      "loss": 3.3985,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.258746904418164e-06,
+      "loss": 3.3331,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.257229585276228e-06,
+      "loss": 3.4053,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.255711744847265e-06,
+      "loss": 3.4192,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.254193383374192e-06,
+      "loss": 3.4386,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.252674501100008e-06,
+      "loss": 3.268,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.251155098267796e-06,
+      "loss": 3.4518,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.249635175120718e-06,
+      "loss": 3.4802,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.248114731902027e-06,
+      "loss": 3.5496,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.246593768855048e-06,
+      "loss": 3.3528,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.2450722862232e-06,
+      "loss": 3.3613,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.243550284249983e-06,
+      "loss": 3.4039,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.242027763178976e-06,
+      "loss": 3.4005,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.240504723253843e-06,
+      "loss": 3.4744,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.238981164718333e-06,
+      "loss": 3.4456,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.237457087816274e-06,
+      "loss": 3.5134,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.235932492791581e-06,
+      "loss": 3.4454,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.234407379888247e-06,
+      "loss": 3.4481,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.232881749350355e-06,
+      "loss": 3.4088,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.231355601422067e-06,
+      "loss": 3.4961,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.229828936347625e-06,
+      "loss": 3.4201,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.228301754371358e-06,
+      "loss": 3.4296,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.226774055737674e-06,
+      "loss": 3.2846,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.22524584069107e-06,
+      "loss": 3.5564,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.223717109476117e-06,
+      "loss": 3.4448,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.222187862337477e-06,
+      "loss": 3.4601,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.220658099519889e-06,
+      "loss": 3.3984,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.219127821268176e-06,
+      "loss": 3.402,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.217597027827247e-06,
+      "loss": 3.4603,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.216065719442084e-06,
+      "loss": 3.4782,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.214533896357762e-06,
+      "loss": 3.4625,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.213001558819433e-06,
+      "loss": 3.338,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.211468707072332e-06,
+      "loss": 3.3934,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.209935341361774e-06,
+      "loss": 3.3726,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.208401461933165e-06,
+      "loss": 3.4378,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.206867069031983e-06,
+      "loss": 3.4054,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.205332162903794e-06,
+      "loss": 3.4704,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20379674379424e-06,
+      "loss": 3.3473,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.202260811949056e-06,
+      "loss": 3.4938,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.200724367614049e-06,
+      "loss": 3.5132,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199187411035112e-06,
+      "loss": 3.3869,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.197649942458217e-06,
+      "loss": 3.4029,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.196111962129425e-06,
+      "loss": 3.5126,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.194573470294872e-06,
+      "loss": 3.4396,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.193034467200777e-06,
+      "loss": 3.5005,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.191494953093446e-06,
+      "loss": 3.4132,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.189954928219258e-06,
+      "loss": 3.3377,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.188414392824681e-06,
+      "loss": 3.421,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.186873347156264e-06,
+      "loss": 3.4493,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.185331791460633e-06,
+      "loss": 3.4146,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.183789725984502e-06,
+      "loss": 3.3761,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.182247150974657e-06,
+      "loss": 3.3983,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.180704066677978e-06,
+      "loss": 3.4082,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.179160473341418e-06,
+      "loss": 3.4621,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177616371212015e-06,
+      "loss": 3.3962,
+      "step": 2705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.176071760536882e-06,
+      "loss": 3.4691,
+      "step": 2706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.174526641563224e-06,
+      "loss": 3.4159,
+      "step": 2707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17298101453832e-06,
+      "loss": 3.4676,
+      "step": 2708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.171434879709533e-06,
+      "loss": 3.3719,
+      "step": 2709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169888237324303e-06,
+      "loss": 3.4643,
+      "step": 2710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.168341087630158e-06,
+      "loss": 3.5045,
+      "step": 2711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.166793430874705e-06,
+      "loss": 3.4,
+      "step": 2712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.165245267305627e-06,
+      "loss": 3.3939,
+      "step": 2713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.163696597170695e-06,
+      "loss": 3.3914,
+      "step": 2714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.162147420717756e-06,
+      "loss": 3.4239,
+      "step": 2715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.160597738194744e-06,
+      "loss": 3.4913,
+      "step": 2716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.159047549849665e-06,
+      "loss": 3.3813,
+      "step": 2717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.157496855930615e-06,
+      "loss": 3.3436,
+      "step": 2718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.155945656685766e-06,
+      "loss": 3.4255,
+      "step": 2719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.154393952363372e-06,
+      "loss": 3.3924,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.152841743211766e-06,
+      "loss": 3.4829,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.151289029479368e-06,
+      "loss": 3.5066,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14973581141467e-06,
+      "loss": 3.5151,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14818208926625e-06,
+      "loss": 3.3941,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.146627863282766e-06,
+      "loss": 3.4991,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.145073133712957e-06,
+      "loss": 3.4408,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14351790080564e-06,
+      "loss": 3.4126,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.141962164809718e-06,
+      "loss": 3.5014,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.140405925974166e-06,
+      "loss": 3.4127,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13884918454805e-06,
+      "loss": 3.451,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.137291940780507e-06,
+      "loss": 3.4972,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.135734194920762e-06,
+      "loss": 3.5569,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.134175947218111e-06,
+      "loss": 3.5399,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.132617197921939e-06,
+      "loss": 3.3606,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13105794728171e-06,
+      "loss": 3.4979,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.129498195546965e-06,
+      "loss": 3.3861,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.127937942967326e-06,
+      "loss": 3.4826,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.126377189792497e-06,
+      "loss": 3.5137,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12481593627226e-06,
+      "loss": 3.4216,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.123254182656475e-06,
+      "loss": 3.4407,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.121691929195092e-06,
+      "loss": 3.3881,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12012917613813e-06,
+      "loss": 3.3425,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11856592373569e-06,
+      "loss": 3.3346,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11700217223796e-06,
+      "loss": 3.4457,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.115437921895198e-06,
+      "loss": 3.485,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11387317295775e-06,
+      "loss": 3.3761,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.112307925676037e-06,
+      "loss": 3.5002,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.110742180300561e-06,
+      "loss": 3.3176,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.109175937081904e-06,
+      "loss": 3.5018,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.107609196270729e-06,
+      "loss": 3.583,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.106041958117775e-06,
+      "loss": 3.3851,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.104474222873864e-06,
+      "loss": 3.4999,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.102905990789895e-06,
+      "loss": 3.4734,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10133726211685e-06,
+      "loss": 3.4348,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.099768037105788e-06,
+      "loss": 3.4404,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.098198316007848e-06,
+      "loss": 3.4186,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.096628099074243e-06,
+      "loss": 3.4293,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.095057386556278e-06,
+      "loss": 3.4254,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.093486178705327e-06,
+      "loss": 3.5392,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.091914475772844e-06,
+      "loss": 3.4429,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.090342278010367e-06,
+      "loss": 3.4326,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08876958566951e-06,
+      "loss": 3.3731,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.087196399001965e-06,
+      "loss": 3.3754,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.085622718259505e-06,
+      "loss": 3.4731,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.084048543693985e-06,
+      "loss": 3.4582,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.082473875557333e-06,
+      "loss": 3.3382,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.080898714101559e-06,
+      "loss": 3.4795,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.079323059578755e-06,
+      "loss": 3.3737,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.077746912241084e-06,
+      "loss": 3.2744,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.076170272340795e-06,
+      "loss": 3.4084,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.074593140130215e-06,
+      "loss": 3.4552,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.073015515861747e-06,
+      "loss": 3.397,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.071437399787872e-06,
+      "loss": 3.3677,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.069858792161152e-06,
+      "loss": 3.4791,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.06827969323423e-06,
+      "loss": 3.4163,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.066700103259825e-06,
+      "loss": 3.487,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.06512002249073e-06,
+      "loss": 3.3864,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.063539451179828e-06,
+      "loss": 3.4849,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061958389580068e-06,
+      "loss": 3.4202,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.060376837944486e-06,
+      "loss": 3.5036,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.058794796526192e-06,
+      "loss": 3.4478,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.057212265578376e-06,
+      "loss": 3.4214,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.055629245354308e-06,
+      "loss": 3.4755,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.054045736107333e-06,
+      "loss": 3.4845,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.052461738090876e-06,
+      "loss": 3.4198,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05087725155844e-06,
+      "loss": 3.3101,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.049292276763607e-06,
+      "loss": 3.4625,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.047706813960035e-06,
+      "loss": 3.3709,
+      "step": 2788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.046120863401461e-06,
+      "loss": 3.4067,
+      "step": 2789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.044534425341703e-06,
+      "loss": 3.3915,
+      "step": 2790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042947500034654e-06,
+      "loss": 3.4239,
+      "step": 2791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.041360087734286e-06,
+      "loss": 3.4658,
+      "step": 2792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039772188694645e-06,
+      "loss": 3.4744,
+      "step": 2793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.038183803169862e-06,
+      "loss": 3.3731,
+      "step": 2794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.036594931414141e-06,
+      "loss": 3.3404,
+      "step": 2795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.035005573681765e-06,
+      "loss": 3.4759,
+      "step": 2796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.033415730227095e-06,
+      "loss": 3.5032,
+      "step": 2797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.031825401304568e-06,
+      "loss": 3.504,
+      "step": 2798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.030234587168702e-06,
+      "loss": 3.4602,
+      "step": 2799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.02864328807409e-06,
+      "loss": 3.4534,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.027051504275404e-06,
+      "loss": 3.4167,
+      "step": 2801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.025459236027393e-06,
+      "loss": 3.3105,
+      "step": 2802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.023866483584886e-06,
+      "loss": 3.4217,
+      "step": 2803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.022273247202783e-06,
+      "loss": 3.4291,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.020679527136067e-06,
+      "loss": 3.4491,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.019085323639796e-06,
+      "loss": 3.3336,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01749063696911e-06,
+      "loss": 3.4034,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01589546737922e-06,
+      "loss": 3.4211,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.014299815125417e-06,
+      "loss": 3.4255,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01270368046307e-06,
+      "loss": 3.5165,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.011107063647622e-06,
+      "loss": 3.3843,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.009509964934599e-06,
+      "loss": 3.4523,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.007912384579597e-06,
+      "loss": 3.4155,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.006314322838295e-06,
+      "loss": 3.4694,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.004715779966448e-06,
+      "loss": 3.4158,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.003116756219887e-06,
+      "loss": 3.462,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.001517251854514e-06,
+      "loss": 3.5356,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.99991726712632e-06,
+      "loss": 3.5669,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.998316802291364e-06,
+      "loss": 3.441,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.996715857605786e-06,
+      "loss": 3.3479,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.995114433325799e-06,
+      "loss": 3.3704,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.993512529707695e-06,
+      "loss": 3.4375,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.991910147007845e-06,
+      "loss": 3.421,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.990307285482693e-06,
+      "loss": 3.5201,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98870394538876e-06,
+      "loss": 3.4008,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.987100126982645e-06,
+      "loss": 3.3828,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.985495830521024e-06,
+      "loss": 3.3642,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.983891056260647e-06,
+      "loss": 3.5334,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.982285804458344e-06,
+      "loss": 3.4611,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98068007537102e-06,
+      "loss": 3.4919,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.979073869255654e-06,
+      "loss": 3.4245,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.977467186369303e-06,
+      "loss": 3.538,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.975860026969102e-06,
+      "loss": 3.4488,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.974252391312259e-06,
+      "loss": 3.4145,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.972644279656062e-06,
+      "loss": 3.4405,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.97103569225787e-06,
+      "loss": 3.3874,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.969426629375127e-06,
+      "loss": 3.4569,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.967817091265344e-06,
+      "loss": 3.4902,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.96620707818611e-06,
+      "loss": 3.4005,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.964596590395092e-06,
+      "loss": 3.4409,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.962985628150037e-06,
+      "loss": 3.4853,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.961374191708757e-06,
+      "loss": 3.3631,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.959762281329151e-06,
+      "loss": 3.4731,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.958149897269186e-06,
+      "loss": 3.4892,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.95653703978691e-06,
+      "loss": 3.3405,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.954923709140442e-06,
+      "loss": 3.4658,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.953309905587984e-06,
+      "loss": 3.3954,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.951695629387806e-06,
+      "loss": 3.338,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.950080880798256e-06,
+      "loss": 3.4273,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.948465660077762e-06,
+      "loss": 3.4621,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94684996748482e-06,
+      "loss": 3.4042,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.945233803278007e-06,
+      "loss": 3.3057,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.943617167715974e-06,
+      "loss": 3.4463,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.942000061057448e-06,
+      "loss": 3.4727,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.940382483561228e-06,
+      "loss": 3.3984,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.938764435486195e-06,
+      "loss": 3.3789,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9371459170913e-06,
+      "loss": 3.3825,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.935526928635571e-06,
+      "loss": 3.4652,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.933907470378108e-06,
+      "loss": 3.4679,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.932287542578093e-06,
+      "loss": 3.5223,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.930667145494776e-06,
+      "loss": 3.4159,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.929046279387488e-06,
+      "loss": 3.5637,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92742494451563e-06,
+      "loss": 3.4644,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.925803141138684e-06,
+      "loss": 3.4849,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9241808695162e-06,
+      "loss": 3.5148,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.922558129907806e-06,
+      "loss": 3.4584,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.920934922573208e-06,
+      "loss": 3.3564,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.919311247772183e-06,
+      "loss": 3.426,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.917687105764584e-06,
+      "loss": 3.4761,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.916062496810338e-06,
+      "loss": 3.4918,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.914437421169448e-06,
+      "loss": 3.4616,
+      "step": 2871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91281187910199e-06,
+      "loss": 3.4741,
+      "step": 2872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91118587086812e-06,
+      "loss": 3.388,
+      "step": 2873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.909559396728056e-06,
+      "loss": 3.5234,
+      "step": 2874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.907932456942108e-06,
+      "loss": 3.4502,
+      "step": 2875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.906305051770646e-06,
+      "loss": 3.4483,
+      "step": 2876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.904677181474123e-06,
+      "loss": 3.3346,
+      "step": 2877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.903048846313059e-06,
+      "loss": 3.4214,
+      "step": 2878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.901420046548059e-06,
+      "loss": 3.5707,
+      "step": 2879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.89979078243979e-06,
+      "loss": 3.5093,
+      "step": 2880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.898161054249003e-06,
+      "loss": 3.3716,
+      "step": 2881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.896530862236518e-06,
+      "loss": 3.4912,
+      "step": 2882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.894900206663232e-06,
+      "loss": 3.4511,
+      "step": 2883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.893269087790115e-06,
+      "loss": 3.5483,
+      "step": 2884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.89163750587821e-06,
+      "loss": 3.4212,
+      "step": 2885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.89000546118864e-06,
+      "loss": 3.4227,
+      "step": 2886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.888372953982589e-06,
+      "loss": 3.4362,
+      "step": 2887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.886739984521331e-06,
+      "loss": 3.3503,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.885106553066204e-06,
+      "loss": 3.4685,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.883472659878622e-06,
+      "loss": 3.4981,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.881838305220073e-06,
+      "loss": 3.4383,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.880203489352119e-06,
+      "loss": 3.495,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.878568212536398e-06,
+      "loss": 3.4063,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.876932475034615e-06,
+      "loss": 3.5101,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.87529627710856e-06,
+      "loss": 3.5329,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.873659619020084e-06,
+      "loss": 3.4103,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.872022501031122e-06,
+      "loss": 3.3618,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.870384923403677e-06,
+      "loss": 3.4117,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.868746886399826e-06,
+      "loss": 3.3996,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.867108390281723e-06,
+      "loss": 3.3569,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.86546943531159e-06,
+      "loss": 3.4787,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.863830021751726e-06,
+      "loss": 3.3157,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.862190149864504e-06,
+      "loss": 3.5046,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.860549819912366e-06,
+      "loss": 3.3636,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.858909032157836e-06,
+      "loss": 3.437,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.857267786863501e-06,
+      "loss": 3.4436,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.855626084292027e-06,
+      "loss": 3.4637,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.853983924706153e-06,
+      "loss": 3.3973,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85234130836869e-06,
+      "loss": 3.4266,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85069823554252e-06,
+      "loss": 3.5159,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.849054706490603e-06,
+      "loss": 3.3726,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.84741072147597e-06,
+      "loss": 3.3886,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.845766280761722e-06,
+      "loss": 3.4748,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.844121384611037e-06,
+      "loss": 3.4201,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.842476033287162e-06,
+      "loss": 3.4173,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.840830227053422e-06,
+      "loss": 3.4143,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83918396617321e-06,
+      "loss": 3.4709,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.837537250909994e-06,
+      "loss": 3.3472,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.835890081527314e-06,
+      "loss": 3.5215,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.834242458288784e-06,
+      "loss": 3.3857,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.832594381458088e-06,
+      "loss": 3.4023,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.830945851298986e-06,
+      "loss": 3.44,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.829296868075306e-06,
+      "loss": 3.4484,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827647432050955e-06,
+      "loss": 3.4691,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.825997543489908e-06,
+      "loss": 3.3739,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.824347202656211e-06,
+      "loss": 3.434,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.822696409813984e-06,
+      "loss": 3.46,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.821045165227424e-06,
+      "loss": 3.4462,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.819393469160791e-06,
+      "loss": 3.379,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.817741321878427e-06,
+      "loss": 3.3795,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81608872364474e-06,
+      "loss": 3.374,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81443567472421e-06,
+      "loss": 3.4128,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.812782175381394e-06,
+      "loss": 3.3344,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.811128225880916e-06,
+      "loss": 3.4944,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.809473826487475e-06,
+      "loss": 3.4087,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.807818977465839e-06,
+      "loss": 3.3047,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.806163679080854e-06,
+      "loss": 3.3491,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.804507931597432e-06,
+      "loss": 3.4556,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.802851735280557e-06,
+      "loss": 3.4929,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.801195090395289e-06,
+      "loss": 3.3547,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.799537997206756e-06,
+      "loss": 3.4794,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79788045598016e-06,
+      "loss": 3.3899,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.796222466980777e-06,
+      "loss": 3.4428,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.794564030473943e-06,
+      "loss": 3.3445,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792905146725083e-06,
+      "loss": 3.3758,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79124581599968e-06,
+      "loss": 3.3267,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789586038563297e-06,
+      "loss": 3.4362,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787925814681558e-06,
+      "loss": 3.3516,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.786265144620174e-06,
+      "loss": 3.3347,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.78460402864491e-06,
+      "loss": 3.5641,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.782942467021618e-06,
+      "loss": 3.5732,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.78128046001621e-06,
+      "loss": 3.4034,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779618007894678e-06,
+      "loss": 3.4938,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.777955110923075e-06,
+      "loss": 3.4205,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.776291769367535e-06,
+      "loss": 3.4511,
+      "step": 2955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774627983494259e-06,
+      "loss": 3.4391,
+      "step": 2956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.772963753569516e-06,
+      "loss": 3.375,
+      "step": 2957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.771299079859654e-06,
+      "loss": 3.4011,
+      "step": 2958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769633962631084e-06,
+      "loss": 3.425,
+      "step": 2959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.767968402150294e-06,
+      "loss": 3.3886,
+      "step": 2960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.766302398683839e-06,
+      "loss": 3.4317,
+      "step": 2961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764635952498345e-06,
+      "loss": 3.437,
+      "step": 2962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.762969063860511e-06,
+      "loss": 3.4595,
+      "step": 2963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.761301733037106e-06,
+      "loss": 3.478,
+      "step": 2964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.75963396029497e-06,
+      "loss": 3.4498,
+      "step": 2965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.757965745901015e-06,
+      "loss": 3.4683,
+      "step": 2966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.756297090122218e-06,
+      "loss": 3.4266,
+      "step": 2967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.754627993225632e-06,
+      "loss": 3.4364,
+      "step": 2968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.752958455478381e-06,
+      "loss": 3.3349,
+      "step": 2969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.751288477147656e-06,
+      "loss": 3.5207,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.749618058500722e-06,
+      "loss": 3.3805,
+      "step": 2971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.747947199804912e-06,
+      "loss": 3.4797,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74627590132763e-06,
+      "loss": 3.3429,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74460416333635e-06,
+      "loss": 3.4175,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.742931986098618e-06,
+      "loss": 3.4886,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.741259369882049e-06,
+      "loss": 3.3873,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.739586314954329e-06,
+      "loss": 3.5561,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.737912821583211e-06,
+      "loss": 3.4286,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.736238890036525e-06,
+      "loss": 3.3969,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.734564520582165e-06,
+      "loss": 3.4655,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.732889713488096e-06,
+      "loss": 3.3905,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.731214469022356e-06,
+      "loss": 3.5287,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.729538787453051e-06,
+      "loss": 3.3337,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.727862669048356e-06,
+      "loss": 3.4413,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.726186114076518e-06,
+      "loss": 3.4025,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.72450912280585e-06,
+      "loss": 3.5163,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.722831695504743e-06,
+      "loss": 3.4305,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.721153832441647e-06,
+      "loss": 3.4036,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.719475533885091e-06,
+      "loss": 3.4471,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.71779680010367e-06,
+      "loss": 3.4409,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.716117631366045e-06,
+      "loss": 3.4416,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.714438027940953e-06,
+      "loss": 3.4912,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.712757990097196e-06,
+      "loss": 3.5001,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.711077518103649e-06,
+      "loss": 3.3776,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.709396612229255e-06,
+      "loss": 3.4623,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.707715272743023e-06,
+      "loss": 3.4862,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.706033499914038e-06,
+      "loss": 3.4038,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.704351294011451e-06,
+      "loss": 3.3793,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.702668655304479e-06,
+      "loss": 3.3701,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.700985584062414e-06,
+      "loss": 3.4827,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.699302080554614e-06,
+      "loss": 3.3715,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.697618145050508e-06,
+      "loss": 3.4823,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.69593377781959e-06,
+      "loss": 3.411,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.694248979131431e-06,
+      "loss": 3.4323,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.692563749255662e-06,
+      "loss": 3.4199,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.690878088461987e-06,
+      "loss": 3.4539,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.689191997020184e-06,
+      "loss": 3.4565,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.68750547520009e-06,
+      "loss": 3.4565,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.685818523271618e-06,
+      "loss": 3.4229,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.684131141504748e-06,
+      "loss": 3.3842,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.682443330169527e-06,
+      "loss": 3.4164,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680755089536073e-06,
+      "loss": 3.3742,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.679066419874575e-06,
+      "loss": 3.4972,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.677377321455284e-06,
+      "loss": 3.3539,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.675687794548524e-06,
+      "loss": 3.4556,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.673997839424689e-06,
+      "loss": 3.4792,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.672307456354235e-06,
+      "loss": 3.4583,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.670616645607695e-06,
+      "loss": 3.5271,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.668925407455667e-06,
+      "loss": 3.389,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.667233742168814e-06,
+      "loss": 3.3724,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.66554165001787e-06,
+      "loss": 3.4928,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.663849131273637e-06,
+      "loss": 3.4636,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.66215618620699e-06,
+      "loss": 3.3171,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.660462815088863e-06,
+      "loss": 3.4543,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.658769018190267e-06,
+      "loss": 3.4223,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.657074795782274e-06,
+      "loss": 3.4462,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.655380148136028e-06,
+      "loss": 3.5136,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.653685075522743e-06,
+      "loss": 3.5615,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.651989578213696e-06,
+      "loss": 3.5469,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.650293656480234e-06,
+      "loss": 3.3539,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.648597310593775e-06,
+      "loss": 3.5554,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6469005408258e-06,
+      "loss": 3.3195,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.645203347447859e-06,
+      "loss": 3.4832,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.643505730731572e-06,
+      "loss": 3.3928,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.641807690948629e-06,
+      "loss": 3.4174,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.640109228370778e-06,
+      "loss": 3.3716,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.638410343269846e-06,
+      "loss": 3.3729,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636711035917722e-06,
+      "loss": 3.402,
+      "step": 3038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.63501130658636e-06,
+      "loss": 3.3955,
+      "step": 3039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.633311155547786e-06,
+      "loss": 3.4885,
+      "step": 3040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.631610583074093e-06,
+      "loss": 3.3688,
+      "step": 3041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.62990958943744e-06,
+      "loss": 3.4607,
+      "step": 3042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.628208174910057e-06,
+      "loss": 3.5156,
+      "step": 3043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.626506339764234e-06,
+      "loss": 3.3952,
+      "step": 3044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.624804084272335e-06,
+      "loss": 3.5149,
+      "step": 3045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.623101408706787e-06,
+      "loss": 3.4692,
+      "step": 3046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6213983133400885e-06,
+      "loss": 3.511,
+      "step": 3047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6196947984448025e-06,
+      "loss": 3.4212,
+      "step": 3048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.617990864293558e-06,
+      "loss": 3.4733,
+      "step": 3049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.616286511159054e-06,
+      "loss": 3.3742,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.614581739314055e-06,
+      "loss": 3.4391,
+      "step": 3051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.612876549031392e-06,
+      "loss": 3.3786,
+      "step": 3052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.611170940583965e-06,
+      "loss": 3.5368,
+      "step": 3053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.609464914244739e-06,
+      "loss": 3.4164,
+      "step": 3054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.607758470286746e-06,
+      "loss": 3.4533,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.606051608983085e-06,
+      "loss": 3.4067,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6043443306069235e-06,
+      "loss": 3.4585,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.602636635431492e-06,
+      "loss": 3.4362,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.600928523730092e-06,
+      "loss": 3.3763,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.59921999577609e-06,
+      "loss": 3.5347,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.597511051842917e-06,
+      "loss": 3.4386,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.595801692204075e-06,
+      "loss": 3.4547,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5940919171331265e-06,
+      "loss": 3.4121,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.592381726903706e-06,
+      "loss": 3.3915,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.590671121789512e-06,
+      "loss": 3.4279,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.58896010206431e-06,
+      "loss": 3.4037,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.587248668001932e-06,
+      "loss": 3.4661,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5855368198762745e-06,
+      "loss": 3.4218,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.583824557961304e-06,
+      "loss": 3.3719,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.582111882531047e-06,
+      "loss": 3.3714,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5803987938596025e-06,
+      "loss": 3.3475,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.578685292221135e-06,
+      "loss": 3.4603,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.57697137788987e-06,
+      "loss": 3.4681,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.575257051140104e-06,
+      "loss": 3.4449,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.573542312246198e-06,
+      "loss": 3.3932,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.571827161482579e-06,
+      "loss": 3.4466,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.57011159912374e-06,
+      "loss": 3.4111,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.568395625444236e-06,
+      "loss": 3.3533,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.566679240718695e-06,
+      "loss": 3.4073,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.564962445221809e-06,
+      "loss": 3.3879,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.56324523922833e-06,
+      "loss": 3.6099,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5615276230130785e-06,
+      "loss": 3.3478,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.559809596850947e-06,
+      "loss": 3.4677,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.558091161016885e-06,
+      "loss": 3.4595,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.556372315785912e-06,
+      "loss": 3.4123,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5546530614331105e-06,
+      "loss": 3.4128,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.552933398233632e-06,
+      "loss": 3.4864,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.551213326462692e-06,
+      "loss": 3.4285,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.549492846395569e-06,
+      "loss": 3.3628,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.547771958307607e-06,
+      "loss": 3.4567,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.54605066247422e-06,
+      "loss": 3.4834,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5443289591708834e-06,
+      "loss": 3.4601,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5426068486731395e-06,
+      "loss": 3.4229,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.54088433125659e-06,
+      "loss": 3.396,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.539161407196913e-06,
+      "loss": 3.49,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.537438076769843e-06,
+      "loss": 3.4498,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5357143402511805e-06,
+      "loss": 3.4517,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.533990197916793e-06,
+      "loss": 3.4424,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.532265650042612e-06,
+      "loss": 3.5042,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.530540696904635e-06,
+      "loss": 3.4328,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.528815338778923e-06,
+      "loss": 3.3898,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.527089575941602e-06,
+      "loss": 3.4027,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.525363408668864e-06,
+      "loss": 3.4447,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.523636837236964e-06,
+      "loss": 3.5117,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.521909861922223e-06,
+      "loss": 3.4117,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.520182483001026e-06,
+      "loss": 3.4297,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.518454700749821e-06,
+      "loss": 3.503,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.516726515445125e-06,
+      "loss": 3.3849,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.514997927363517e-06,
+      "loss": 3.4009,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.513268936781637e-06,
+      "loss": 3.5151,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.511539543976194e-06,
+      "loss": 3.4356,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.509809749223962e-06,
+      "loss": 3.3867,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.508079552801776e-06,
+      "loss": 3.5071,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.506348954986535e-06,
+      "loss": 3.4559,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.504617956055207e-06,
+      "loss": 3.4182,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.502886556284819e-06,
+      "loss": 3.3829,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5011547559524646e-06,
+      "loss": 3.4618,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.499422555335301e-06,
+      "loss": 3.4954,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.49768995471055e-06,
+      "loss": 3.387,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4959569543554966e-06,
+      "loss": 3.4645,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.494223554547491e-06,
+      "loss": 3.286,
+      "step": 3121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.492489755563945e-06,
+      "loss": 3.4432,
+      "step": 3122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4907555576823375e-06,
+      "loss": 3.465,
+      "step": 3123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.489020961180209e-06,
+      "loss": 3.3448,
+      "step": 3124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.487285966335163e-06,
+      "loss": 3.4239,
+      "step": 3125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.48555057342487e-06,
+      "loss": 3.4163,
+      "step": 3126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.48381478272706e-06,
+      "loss": 3.3715,
+      "step": 3127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.482078594519532e-06,
+      "loss": 3.4643,
+      "step": 3128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.480342009080144e-06,
+      "loss": 3.428,
+      "step": 3129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4786050266868184e-06,
+      "loss": 3.3119,
+      "step": 3130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.476867647617542e-06,
+      "loss": 3.3804,
+      "step": 3131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.475129872150366e-06,
+      "loss": 3.4964,
+      "step": 3132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.473391700563403e-06,
+      "loss": 3.4749,
+      "step": 3133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.471653133134832e-06,
+      "loss": 3.4197,
+      "step": 3134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.469914170142889e-06,
+      "loss": 3.402,
+      "step": 3135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.468174811865883e-06,
+      "loss": 3.3758,
+      "step": 3136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.466435058582178e-06,
+      "loss": 3.4238,
+      "step": 3137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4646949105702025e-06,
+      "loss": 3.3662,
+      "step": 3138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.462954368108452e-06,
+      "loss": 3.4611,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4612134314754844e-06,
+      "loss": 3.5038,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4594721009499164e-06,
+      "loss": 3.4102,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4577303768104325e-06,
+      "loss": 3.4593,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.455988259335776e-06,
+      "loss": 3.4057,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.454245748804757e-06,
+      "loss": 3.4991,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.452502845496246e-06,
+      "loss": 3.4909,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4507595496891795e-06,
+      "loss": 3.3797,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.449015861662552e-06,
+      "loss": 3.5184,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.447271781695424e-06,
+      "loss": 3.4756,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.445527310066919e-06,
+      "loss": 3.2832,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.443782447056221e-06,
+      "loss": 3.5145,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.442037192942577e-06,
+      "loss": 3.4411,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.440291548005301e-06,
+      "loss": 3.4298,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.438545512523764e-06,
+      "loss": 3.4947,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4367990867774e-06,
+      "loss": 3.4462,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.435052271045709e-06,
+      "loss": 3.4618,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.433305065608251e-06,
+      "loss": 3.5147,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.431557470744648e-06,
+      "loss": 3.4575,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.429809486734586e-06,
+      "loss": 3.48,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4280611138578105e-06,
+      "loss": 3.4821,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.426312352394134e-06,
+      "loss": 3.4512,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4245632026234274e-06,
+      "loss": 3.3727,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.422813664825623e-06,
+      "loss": 3.3185,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.421063739280718e-06,
+      "loss": 3.4701,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.419313426268771e-06,
+      "loss": 3.482,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.417562726069901e-06,
+      "loss": 3.4269,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4158116389642906e-06,
+      "loss": 3.3753,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4140601652321845e-06,
+      "loss": 3.4183,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.412308305153886e-06,
+      "loss": 3.4277,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410556059009766e-06,
+      "loss": 3.502,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.408803427080252e-06,
+      "loss": 3.536,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.407050409645837e-06,
+      "loss": 3.3385,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.405297006987071e-06,
+      "loss": 3.4144,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4035432193845704e-06,
+      "loss": 3.3936,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.401789047119012e-06,
+      "loss": 3.4075,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.400034490471133e-06,
+      "loss": 3.4836,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3982795497217316e-06,
+      "loss": 3.3922,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.396524225151669e-06,
+      "loss": 3.4174,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.39476851704187e-06,
+      "loss": 3.3394,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.393012425673311e-06,
+      "loss": 3.4078,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.391255951327045e-06,
+      "loss": 3.4358,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.389499094284176e-06,
+      "loss": 3.3895,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.387741854825867e-06,
+      "loss": 3.4302,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.385984233233352e-06,
+      "loss": 3.3865,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.384226229787918e-06,
+      "loss": 3.5098,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.382467844770917e-06,
+      "loss": 3.4515,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3807090784637594e-06,
+      "loss": 3.3859,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.37894993114792e-06,
+      "loss": 3.3801,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.37719040310493e-06,
+      "loss": 3.3302,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.375430494616389e-06,
+      "loss": 3.5257,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.373670205963948e-06,
+      "loss": 3.4421,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.371909537429327e-06,
+      "loss": 3.4246,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3701484892943e-06,
+      "loss": 3.4381,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.368387061840707e-06,
+      "loss": 3.4843,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3666252553504494e-06,
+      "loss": 3.3767,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.364863070105482e-06,
+      "loss": 3.3826,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.363100506387828e-06,
+      "loss": 3.4949,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.361337564479567e-06,
+      "loss": 3.5067,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3595742446628405e-06,
+      "loss": 3.3907,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.35781054721985e-06,
+      "loss": 3.4976,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.356046472432857e-06,
+      "loss": 3.4452,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.354282020584187e-06,
+      "loss": 3.4134,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3525171919562186e-06,
+      "loss": 3.3993,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.350751986831399e-06,
+      "loss": 3.4196,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.348986405492228e-06,
+      "loss": 3.3603,
+      "step": 3204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3472204482212715e-06,
+      "loss": 3.3814,
+      "step": 3205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.345454115301152e-06,
+      "loss": 3.4169,
+      "step": 3206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.343687407014556e-06,
+      "loss": 3.477,
+      "step": 3207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341920323644224e-06,
+      "loss": 3.4007,
+      "step": 3208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.340152865472963e-06,
+      "loss": 3.3636,
+      "step": 3209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.338385032783635e-06,
+      "loss": 3.3962,
+      "step": 3210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3366168258591656e-06,
+      "loss": 3.401,
+      "step": 3211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.334848244982535e-06,
+      "loss": 3.4206,
+      "step": 3212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.333079290436792e-06,
+      "loss": 3.4498,
+      "step": 3213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.331309962505037e-06,
+      "loss": 3.4787,
+      "step": 3214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.329540261470435e-06,
+      "loss": 3.4878,
+      "step": 3215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.327770187616205e-06,
+      "loss": 3.4329,
+      "step": 3216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.325999741225636e-06,
+      "loss": 3.3415,
+      "step": 3217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.324228922582064e-06,
+      "loss": 3.4987,
+      "step": 3218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.322457731968895e-06,
+      "loss": 3.4012,
+      "step": 3219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.320686169669587e-06,
+      "loss": 3.5068,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.318914235967664e-06,
+      "loss": 3.3608,
+      "step": 3221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3171419311467065e-06,
+      "loss": 3.474,
+      "step": 3222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.31536925549035e-06,
+      "loss": 3.3464,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.313596209282297e-06,
+      "loss": 3.4026,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.311822792806306e-06,
+      "loss": 3.3996,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.310049006346193e-06,
+      "loss": 3.4255,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.308274850185835e-06,
+      "loss": 3.3399,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3065003246091685e-06,
+      "loss": 3.545,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.304725429900188e-06,
+      "loss": 3.3468,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3029501663429485e-06,
+      "loss": 3.3627,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3011745342215625e-06,
+      "loss": 3.3565,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.299398533820202e-06,
+      "loss": 3.3605,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.297622165423098e-06,
+      "loss": 3.433,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.295845429314543e-06,
+      "loss": 3.4423,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.294068325778883e-06,
+      "loss": 3.4252,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.292290855100526e-06,
+      "loss": 3.3528,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2905130175639385e-06,
+      "loss": 3.4225,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2887348134536464e-06,
+      "loss": 3.4944,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.286956243054233e-06,
+      "loss": 3.4554,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2851773066503415e-06,
+      "loss": 3.4393,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.283398004526672e-06,
+      "loss": 3.4009,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.281618336967984e-06,
+      "loss": 3.4646,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.279838304259095e-06,
+      "loss": 3.4843,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.278057906684883e-06,
+      "loss": 3.4371,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.27627714453028e-06,
+      "loss": 3.5733,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.274496018080284e-06,
+      "loss": 3.3895,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2727145276199414e-06,
+      "loss": 3.5215,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.270932673434366e-06,
+      "loss": 3.4328,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.269150455808723e-06,
+      "loss": 3.4257,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.267367875028238e-06,
+      "loss": 3.4516,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.265584931378199e-06,
+      "loss": 3.4513,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.263801625143945e-06,
+      "loss": 3.4576,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.262017956610876e-06,
+      "loss": 3.3264,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2602339260644524e-06,
+      "loss": 3.5462,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.258449533790191e-06,
+      "loss": 3.4276,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.256664780073663e-06,
+      "loss": 3.44,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.254879665200503e-06,
+      "loss": 3.3855,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.253094189456399e-06,
+      "loss": 3.3194,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2513083531271e-06,
+      "loss": 3.3781,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2495221564984096e-06,
+      "loss": 3.4013,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.247735599856191e-06,
+      "loss": 3.4239,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.245948683486367e-06,
+      "loss": 3.4142,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.244161407674914e-06,
+      "loss": 3.5828,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.242373772707867e-06,
+      "loss": 3.37,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.240585778871319e-06,
+      "loss": 3.4095,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238797426451423e-06,
+      "loss": 3.4531,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.237008715734384e-06,
+      "loss": 3.4693,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.235219647006468e-06,
+      "loss": 3.3098,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2334302205539985e-06,
+      "loss": 3.3258,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.231640436663356e-06,
+      "loss": 3.4398,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.229850295620975e-06,
+      "loss": 3.423,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.228059797713352e-06,
+      "loss": 3.4765,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2262689432270364e-06,
+      "loss": 3.4756,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.224477732448638e-06,
+      "loss": 3.4544,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.222686165664823e-06,
+      "loss": 3.464,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.220894243162311e-06,
+      "loss": 3.4212,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.219101965227882e-06,
+      "loss": 3.5172,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.217309332148375e-06,
+      "loss": 3.4636,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.215516344210679e-06,
+      "loss": 3.442,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2137230017017464e-06,
+      "loss": 3.3365,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2119293049085825e-06,
+      "loss": 3.4288,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.210135254118251e-06,
+      "loss": 3.4164,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.208340849617873e-06,
+      "loss": 3.4206,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.206546091694624e-06,
+      "loss": 3.426,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.204750980635735e-06,
+      "loss": 3.3777,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.202955516728498e-06,
+      "loss": 3.5137,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.20115970026026e-06,
+      "loss": 3.3668,
+      "step": 3287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.199363531518421e-06,
+      "loss": 3.4782,
+      "step": 3288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.197567010790439e-06,
+      "loss": 3.4232,
+      "step": 3289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.195770138363833e-06,
+      "loss": 3.5008,
+      "step": 3290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1939729145261725e-06,
+      "loss": 3.354,
+      "step": 3291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.192175339565085e-06,
+      "loss": 3.472,
+      "step": 3292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1903774137682546e-06,
+      "loss": 3.4173,
+      "step": 3293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18857913742342e-06,
+      "loss": 3.5249,
+      "step": 3294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1867805108183806e-06,
+      "loss": 3.38,
+      "step": 3295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.184981534240986e-06,
+      "loss": 3.4525,
+      "step": 3296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.183182207979144e-06,
+      "loss": 3.3996,
+      "step": 3297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.181382532320821e-06,
+      "loss": 3.4574,
+      "step": 3298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.179582507554035e-06,
+      "loss": 3.3671,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.177782133966864e-06,
+      "loss": 3.3963,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.175981411847438e-06,
+      "loss": 3.4505,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.174180341483945e-06,
+      "loss": 3.4829,
+      "step": 3302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.172378923164629e-06,
+      "loss": 3.509,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.170577157177789e-06,
+      "loss": 3.4549,
+      "step": 3304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.168775043811779e-06,
+      "loss": 3.3731,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.166972583355007e-06,
+      "loss": 3.3633,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1651697760959435e-06,
+      "loss": 3.4201,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.163366622323108e-06,
+      "loss": 3.4041,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.161563122325076e-06,
+      "loss": 3.3463,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.15975927639048e-06,
+      "loss": 3.3413,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1579550848080085e-06,
+      "loss": 3.484,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.156150547866402e-06,
+      "loss": 3.4906,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.154345665854462e-06,
+      "loss": 3.4171,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1525404390610384e-06,
+      "loss": 3.4565,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.150734867775042e-06,
+      "loss": 3.3992,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.148928952285437e-06,
+      "loss": 3.4059,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.147122692881241e-06,
+      "loss": 3.4418,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.145316089851526e-06,
+      "loss": 3.4588,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.143509143485423e-06,
+      "loss": 3.4483,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.141701854072116e-06,
+      "loss": 3.4928,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.139894221900842e-06,
+      "loss": 3.4825,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.138086247260894e-06,
+      "loss": 3.4745,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.136277930441623e-06,
+      "loss": 3.4117,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.13446927173243e-06,
+      "loss": 3.5122,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1326602714227735e-06,
+      "loss": 3.5388,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.130850929802165e-06,
+      "loss": 3.4977,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.129041247160172e-06,
+      "loss": 3.4729,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.127231223786417e-06,
+      "loss": 3.3615,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.125420859970575e-06,
+      "loss": 3.4197,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1236101560023775e-06,
+      "loss": 3.436,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.121799112171609e-06,
+      "loss": 3.4249,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.119987728768109e-06,
+      "loss": 3.3965,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.118176006081772e-06,
+      "loss": 3.4841,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.116363944402542e-06,
+      "loss": 3.5233,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.114551544020428e-06,
+      "loss": 3.4144,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.112738805225483e-06,
+      "loss": 3.4335,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.110925728307818e-06,
+      "loss": 3.435,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.109112313557597e-06,
+      "loss": 3.4145,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1072985612650416e-06,
+      "loss": 3.4396,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.105484471720421e-06,
+      "loss": 3.4695,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.103670045214065e-06,
+      "loss": 3.4039,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.101855282036353e-06,
+      "loss": 3.4931,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.10004018247772e-06,
+      "loss": 3.4326,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.098224746828655e-06,
+      "loss": 3.3676,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.096408975379701e-06,
+      "loss": 3.561,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.09459286842145e-06,
+      "loss": 3.4538,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.092776426244558e-06,
+      "loss": 3.3749,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.090959649139724e-06,
+      "loss": 3.4276,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.089142537397707e-06,
+      "loss": 3.3144,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.087325091309315e-06,
+      "loss": 3.3531,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.085507311165415e-06,
+      "loss": 3.3655,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.083689197256923e-06,
+      "loss": 3.3594,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.081870749874811e-06,
+      "loss": 3.4102,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.080051969310102e-06,
+      "loss": 3.4495,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.078232855853875e-06,
+      "loss": 3.3798,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.076413409797258e-06,
+      "loss": 3.4188,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.07459363143144e-06,
+      "loss": 3.4449,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.072773521047655e-06,
+      "loss": 3.3753,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.070953078937193e-06,
+      "loss": 3.3321,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.069132305391401e-06,
+      "loss": 3.3525,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.067311200701672e-06,
+      "loss": 3.4123,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.065489765159457e-06,
+      "loss": 3.3214,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0636679990562585e-06,
+      "loss": 3.3874,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0618459026836316e-06,
+      "loss": 3.5224,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.060023476333186e-06,
+      "loss": 3.4661,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.058200720296581e-06,
+      "loss": 3.3768,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.05637763486553e-06,
+      "loss": 3.4102,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.054554220331803e-06,
+      "loss": 3.3845,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.052730476987216e-06,
+      "loss": 3.4552,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.050906405123643e-06,
+      "loss": 3.4074,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.049082005033005e-06,
+      "loss": 3.3954,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.047257277007283e-06,
+      "loss": 3.314,
+      "step": 3372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.045432221338505e-06,
+      "loss": 3.4029,
+      "step": 3373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.043606838318753e-06,
+      "loss": 3.5424,
+      "step": 3374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.04178112824016e-06,
+      "loss": 3.341,
+      "step": 3375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.039955091394914e-06,
+      "loss": 3.4589,
+      "step": 3376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.038128728075255e-06,
+      "loss": 3.4828,
+      "step": 3377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0363020385734725e-06,
+      "loss": 3.3031,
+      "step": 3378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0344750231819095e-06,
+      "loss": 3.5283,
+      "step": 3379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.032647682192963e-06,
+      "loss": 3.4237,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.030820015899079e-06,
+      "loss": 3.4379,
+      "step": 3381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0289920245927586e-06,
+      "loss": 3.4798,
+      "step": 3382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0271637085665535e-06,
+      "loss": 3.4771,
+      "step": 3383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.025335068113066e-06,
+      "loss": 3.4492,
+      "step": 3384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.023506103524954e-06,
+      "loss": 3.3557,
+      "step": 3385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.021676815094923e-06,
+      "loss": 3.4351,
+      "step": 3386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.019847203115733e-06,
+      "loss": 3.4747,
+      "step": 3387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.018017267880195e-06,
+      "loss": 3.4333,
+      "step": 3388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.016187009681173e-06,
+      "loss": 3.4586,
+      "step": 3389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.01435642881158e-06,
+      "loss": 3.4566,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.012525525564382e-06,
+      "loss": 3.4549,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0106943002325975e-06,
+      "loss": 3.4157,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.008862753109297e-06,
+      "loss": 3.4722,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.007030884487598e-06,
+      "loss": 3.419,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.005198694660676e-06,
+      "loss": 3.4332,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0033661839217535e-06,
+      "loss": 3.5075,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.001533352564105e-06,
+      "loss": 3.4173,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.999700200881057e-06,
+      "loss": 3.3737,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.997866729165988e-06,
+      "loss": 3.3901,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.996032937712326e-06,
+      "loss": 3.3541,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.994198826813552e-06,
+      "loss": 3.378,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.992364396763196e-06,
+      "loss": 3.401,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.990529647854839e-06,
+      "loss": 3.4206,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.988694580382117e-06,
+      "loss": 3.3258,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.986859194638713e-06,
+      "loss": 3.4498,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.985023490918362e-06,
+      "loss": 3.4614,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.983187469514849e-06,
+      "loss": 3.4287,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.981351130722014e-06,
+      "loss": 3.4946,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.979514474833741e-06,
+      "loss": 3.4596,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97767750214397e-06,
+      "loss": 3.3772,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97584021294669e-06,
+      "loss": 3.4447,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.974002607535941e-06,
+      "loss": 3.5639,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9721646862058145e-06,
+      "loss": 3.3569,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9703264492504495e-06,
+      "loss": 3.4005,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9684878969640366e-06,
+      "loss": 3.4776,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.966649029640822e-06,
+      "loss": 3.3432,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9648098475750935e-06,
+      "loss": 3.3756,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.962970351061195e-06,
+      "loss": 3.4359,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.961130540393522e-06,
+      "loss": 3.3277,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.959290415866514e-06,
+      "loss": 3.3564,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9574499777746675e-06,
+      "loss": 3.4318,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9556092264125255e-06,
+      "loss": 3.4736,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.953768162074681e-06,
+      "loss": 3.4444,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.951926785055777e-06,
+      "loss": 3.5207,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.950085095650511e-06,
+      "loss": 3.4279,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9482430941536235e-06,
+      "loss": 3.4428,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.946400780859911e-06,
+      "loss": 3.5103,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.944558156064214e-06,
+      "loss": 3.439,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.94271522006143e-06,
+      "loss": 3.411,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.940871973146501e-06,
+      "loss": 3.4665,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.939028415614419e-06,
+      "loss": 3.4099,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.937184547760228e-06,
+      "loss": 3.4244,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.935340369879021e-06,
+      "loss": 3.3754,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.93349588226594e-06,
+      "loss": 3.3637,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.931651085216177e-06,
+      "loss": 3.5317,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.929805979024971e-06,
+      "loss": 3.4088,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9279605639876165e-06,
+      "loss": 3.3877,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.926114840399451e-06,
+      "loss": 3.3683,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.924268808555865e-06,
+      "loss": 3.3956,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.922422468752298e-06,
+      "loss": 3.5099,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.920575821284237e-06,
+      "loss": 3.488,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9187288664472205e-06,
+      "loss": 3.4102,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9168816045368355e-06,
+      "loss": 3.386,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.915034035848716e-06,
+      "loss": 3.3702,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.913186160678549e-06,
+      "loss": 3.3417,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.911337979322067e-06,
+      "loss": 3.3184,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.909489492075054e-06,
+      "loss": 3.4198,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.90764069923334e-06,
+      "loss": 3.4048,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.905791601092808e-06,
+      "loss": 3.4258,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.903942197949387e-06,
+      "loss": 3.4899,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.902092490099055e-06,
+      "loss": 3.3949,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.900242477837839e-06,
+      "loss": 3.4209,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.898392161461816e-06,
+      "loss": 3.3501,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.89654154126711e-06,
+      "loss": 3.4085,
+      "step": 3454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.894690617549894e-06,
+      "loss": 3.4567,
+      "step": 3455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8928393906063895e-06,
+      "loss": 3.3873,
+      "step": 3456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.890987860732869e-06,
+      "loss": 3.3952,
+      "step": 3457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.889136028225649e-06,
+      "loss": 3.4077,
+      "step": 3458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8872838933811e-06,
+      "loss": 3.5871,
+      "step": 3459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.885431456495631e-06,
+      "loss": 3.3973,
+      "step": 3460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.883578717865713e-06,
+      "loss": 3.4729,
+      "step": 3461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.881725677787856e-06,
+      "loss": 3.4734,
+      "step": 3462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8798723365586195e-06,
+      "loss": 3.3989,
+      "step": 3463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8780186944746115e-06,
+      "loss": 3.5012,
+      "step": 3464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.876164751832493e-06,
+      "loss": 3.3925,
+      "step": 3465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.874310508928966e-06,
+      "loss": 3.4431,
+      "step": 3466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.872455966060782e-06,
+      "loss": 3.4261,
+      "step": 3467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.870601123524743e-06,
+      "loss": 3.4349,
+      "step": 3468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8687459816177e-06,
+      "loss": 3.4723,
+      "step": 3469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.86689054063655e-06,
+      "loss": 3.3697,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8650348008782355e-06,
+      "loss": 3.5503,
+      "step": 3471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.863178762639749e-06,
+      "loss": 3.4408,
+      "step": 3472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8613224262181335e-06,
+      "loss": 3.4587,
+      "step": 3473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.859465791910476e-06,
+      "loss": 3.3906,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.857608860013911e-06,
+      "loss": 3.4457,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.855751630825621e-06,
+      "loss": 3.4631,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.853894104642841e-06,
+      "loss": 3.5097,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.852036281762846e-06,
+      "loss": 3.4216,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.850178162482964e-06,
+      "loss": 3.5584,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.848319747100567e-06,
+      "loss": 3.4312,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.846461035913077e-06,
+      "loss": 3.3835,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.84460202921796e-06,
+      "loss": 3.4718,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.842742727312733e-06,
+      "loss": 3.4514,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.84088313049496e-06,
+      "loss": 3.4987,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.839023239062248e-06,
+      "loss": 3.4372,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.837163053312255e-06,
+      "loss": 3.4266,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8353025735426884e-06,
+      "loss": 3.3483,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.833441800051295e-06,
+      "loss": 3.4116,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8315807331358745e-06,
+      "loss": 3.4123,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.829719373094271e-06,
+      "loss": 3.3386,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.82785772022438e-06,
+      "loss": 3.424,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8259957748241365e-06,
+      "loss": 3.3502,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.824133537191528e-06,
+      "loss": 3.4749,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.822271007624588e-06,
+      "loss": 3.3969,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.820408186421394e-06,
+      "loss": 3.4034,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.818545073880072e-06,
+      "loss": 3.4309,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.816681670298795e-06,
+      "loss": 3.4898,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.814817975975783e-06,
+      "loss": 3.4666,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8129539912093005e-06,
+      "loss": 3.3203,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8110897162976595e-06,
+      "loss": 3.4077,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.809225151539218e-06,
+      "loss": 3.4467,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.807360297232384e-06,
+      "loss": 3.4322,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.805495153675606e-06,
+      "loss": 3.4696,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.803629721167383e-06,
+      "loss": 3.4701,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.801764000006257e-06,
+      "loss": 3.4912,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.799897990490819e-06,
+      "loss": 3.5376,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.798031692919707e-06,
+      "loss": 3.4634,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.796165107591599e-06,
+      "loss": 3.4652,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.794298234805226e-06,
+      "loss": 3.5394,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.792431074859364e-06,
+      "loss": 3.4313,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.790563628052831e-06,
+      "loss": 3.5081,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.788695894684494e-06,
+      "loss": 3.4216,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7868278750532635e-06,
+      "loss": 3.4357,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7849595694581e-06,
+      "loss": 3.4524,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7830909781980045e-06,
+      "loss": 3.435,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7812221015720284e-06,
+      "loss": 3.3715,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.779352939879265e-06,
+      "loss": 3.4627,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.777483493418856e-06,
+      "loss": 3.541,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.775613762489988e-06,
+      "loss": 3.3846,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.773743747391892e-06,
+      "loss": 3.3766,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.771873448423845e-06,
+      "loss": 3.466,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.77000286588517e-06,
+      "loss": 3.5241,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.768132000075235e-06,
+      "loss": 3.3862,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.766260851293452e-06,
+      "loss": 3.3812,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.764389419839283e-06,
+      "loss": 3.4381,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.762517706012231e-06,
+      "loss": 3.3648,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7606457101118415e-06,
+      "loss": 3.4127,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.758773432437713e-06,
+      "loss": 3.4232,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7569008732894826e-06,
+      "loss": 3.3858,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.755028032966835e-06,
+      "loss": 3.4513,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.753154911769499e-06,
+      "loss": 3.506,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7512815099972515e-06,
+      "loss": 3.5052,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7494078279499075e-06,
+      "loss": 3.4991,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.747533865927335e-06,
+      "loss": 3.4445,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7456596242294415e-06,
+      "loss": 3.4198,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.74378510315618e-06,
+      "loss": 3.5149,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.741910303007548e-06,
+      "loss": 3.418,
+      "step": 3537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.740035224083592e-06,
+      "loss": 3.4378,
+      "step": 3538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7381598666843976e-06,
+      "loss": 3.4985,
+      "step": 3539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.736284231110095e-06,
+      "loss": 3.4424,
+      "step": 3540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.734408317660863e-06,
+      "loss": 3.4278,
+      "step": 3541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.732532126636924e-06,
+      "loss": 3.4204,
+      "step": 3542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.730655658338542e-06,
+      "loss": 3.3537,
+      "step": 3543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.728778913066026e-06,
+      "loss": 3.3436,
+      "step": 3544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7269018911197324e-06,
+      "loss": 3.3981,
+      "step": 3545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.725024592800058e-06,
+      "loss": 3.3475,
+      "step": 3546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7231470184074476e-06,
+      "loss": 3.44,
+      "step": 3547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.721269168242387e-06,
+      "loss": 3.4574,
+      "step": 3548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7193910426054065e-06,
+      "loss": 3.4023,
+      "step": 3549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.717512641797083e-06,
+      "loss": 3.4493,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.715633966118034e-06,
+      "loss": 3.4519,
+      "step": 3551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.713755015868925e-06,
+      "loss": 3.3147,
+      "step": 3552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7118757913504624e-06,
+      "loss": 3.4032,
+      "step": 3553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.709996292863397e-06,
+      "loss": 3.4728,
+      "step": 3554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.708116520708522e-06,
+      "loss": 3.3931,
+      "step": 3555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.706236475186679e-06,
+      "loss": 3.4456,
+      "step": 3556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.704356156598749e-06,
+      "loss": 3.5012,
+      "step": 3557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.702475565245657e-06,
+      "loss": 3.4438,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.700594701428377e-06,
+      "loss": 3.3773,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6987135654479185e-06,
+      "loss": 3.3907,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.696832157605339e-06,
+      "loss": 3.4241,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.694950478201741e-06,
+      "loss": 3.4155,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.693068527538267e-06,
+      "loss": 3.4156,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6911863059161044e-06,
+      "loss": 3.5436,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.689303813636483e-06,
+      "loss": 3.3497,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.687421051000678e-06,
+      "loss": 3.3798,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.685538018310009e-06,
+      "loss": 3.4954,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.683654715865833e-06,
+      "loss": 3.3949,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.681771143969554e-06,
+      "loss": 3.4584,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.679887302922622e-06,
+      "loss": 3.4635,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6780031930265235e-06,
+      "loss": 3.4318,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.676118814582793e-06,
+      "loss": 3.3602,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.674234167893007e-06,
+      "loss": 3.519,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.672349253258781e-06,
+      "loss": 3.4571,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.670464070981782e-06,
+      "loss": 3.4773,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.66857862136371e-06,
+      "loss": 3.4081,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.666692904706317e-06,
+      "loss": 3.2787,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.664806921311387e-06,
+      "loss": 3.4155,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6629206714807584e-06,
+      "loss": 3.4302,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.661034155516304e-06,
+      "loss": 3.4717,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.659147373719941e-06,
+      "loss": 3.3861,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.657260326393631e-06,
+      "loss": 3.3478,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.655373013839379e-06,
+      "loss": 3.4891,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6534854363592275e-06,
+      "loss": 3.4096,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.651597594255266e-06,
+      "loss": 3.3619,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.649709487829623e-06,
+      "loss": 3.5106,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.647821117384473e-06,
+      "loss": 3.4174,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.645932483222031e-06,
+      "loss": 3.38,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6440435856445525e-06,
+      "loss": 3.336,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.642154424954337e-06,
+      "loss": 3.5287,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.640265001453727e-06,
+      "loss": 3.4898,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.638375315445105e-06,
+      "loss": 3.4472,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.636485367230896e-06,
+      "loss": 3.4527,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.634595157113567e-06,
+      "loss": 3.4576,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.632704685395628e-06,
+      "loss": 3.4065,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.630813952379631e-06,
+      "loss": 3.3887,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.628922958368167e-06,
+      "loss": 3.4052,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.627031703663871e-06,
+      "loss": 3.4918,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.625140188569421e-06,
+      "loss": 3.4735,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.623248413387535e-06,
+      "loss": 3.49,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.62135637842097e-06,
+      "loss": 3.3188,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.619464083972529e-06,
+      "loss": 3.3849,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.617571530345055e-06,
+      "loss": 3.4486,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.615678717841434e-06,
+      "loss": 3.544,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.613785646764588e-06,
+      "loss": 3.3877,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.611892317417486e-06,
+      "loss": 3.4231,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.609998730103135e-06,
+      "loss": 3.5452,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6081048851245885e-06,
+      "loss": 3.3984,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6062107827849335e-06,
+      "loss": 3.4817,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6043164233873045e-06,
+      "loss": 3.4609,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6024218072348734e-06,
+      "loss": 3.5109,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.600526934630855e-06,
+      "loss": 3.437,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.598631805878506e-06,
+      "loss": 3.4346,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.59673642128112e-06,
+      "loss": 3.5725,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.594840781142037e-06,
+      "loss": 3.4052,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.592944885764634e-06,
+      "loss": 3.4409,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.59104873545233e-06,
+      "loss": 3.3447,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.589152330508586e-06,
+      "loss": 3.3521,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.587255671236901e-06,
+      "loss": 3.4029,
+      "step": 3619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.585358757940817e-06,
+      "loss": 3.4171,
+      "step": 3620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.583461590923918e-06,
+      "loss": 3.3859,
+      "step": 3621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.581564170489822e-06,
+      "loss": 3.5505,
+      "step": 3622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5796664969421966e-06,
+      "loss": 3.4129,
+      "step": 3623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.577768570584744e-06,
+      "loss": 3.4186,
+      "step": 3624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.575870391721209e-06,
+      "loss": 3.493,
+      "step": 3625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.573971960655374e-06,
+      "loss": 3.3779,
+      "step": 3626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.572073277691066e-06,
+      "loss": 3.3912,
+      "step": 3627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.570174343132148e-06,
+      "loss": 3.4203,
+      "step": 3628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5682751572825285e-06,
+      "loss": 3.3282,
+      "step": 3629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.566375720446152e-06,
+      "loss": 3.4307,
+      "step": 3630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.564476032927003e-06,
+      "loss": 3.5083,
+      "step": 3631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.56257609502911e-06,
+      "loss": 3.3382,
+      "step": 3632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5606759070565366e-06,
+      "loss": 3.4286,
+      "step": 3633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.55877546931339e-06,
+      "loss": 3.373,
+      "step": 3634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.556874782103815e-06,
+      "loss": 3.4622,
+      "step": 3635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.554973845732e-06,
+      "loss": 3.3969,
+      "step": 3636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.553072660502169e-06,
+      "loss": 3.3734,
+      "step": 3637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.551171226718589e-06,
+      "loss": 3.4016,
+      "step": 3638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.549269544685564e-06,
+      "loss": 3.4196,
+      "step": 3639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.547367614707441e-06,
+      "loss": 3.3368,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.545465437088603e-06,
+      "loss": 3.2841,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.543563012133476e-06,
+      "loss": 3.472,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5416603401465216e-06,
+      "loss": 3.4138,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.539757421432247e-06,
+      "loss": 3.3816,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.537854256295193e-06,
+      "loss": 3.3544,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.535950845039944e-06,
+      "loss": 3.3733,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.534047187971118e-06,
+      "loss": 3.4551,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.53214328539338e-06,
+      "loss": 3.4363,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5302391376114305e-06,
+      "loss": 3.3932,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.528334744930007e-06,
+      "loss": 3.4107,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.526430107653889e-06,
+      "loss": 3.5553,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.524525226087898e-06,
+      "loss": 3.3892,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.522620100536887e-06,
+      "loss": 3.5228,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5207147313057555e-06,
+      "loss": 3.4157,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.518809118699436e-06,
+      "loss": 3.4938,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.516903263022904e-06,
+      "loss": 3.4382,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.514997164581173e-06,
+      "loss": 3.5883,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.513090823679295e-06,
+      "loss": 3.5432,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.511184240622359e-06,
+      "loss": 3.315,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.509277415715497e-06,
+      "loss": 3.4146,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5073703492638775e-06,
+      "loss": 3.406,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.505463041572705e-06,
+      "loss": 3.4027,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.503555492947226e-06,
+      "loss": 3.4459,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.501647703692725e-06,
+      "loss": 3.3941,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.499739674114524e-06,
+      "loss": 3.3657,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.497831404517986e-06,
+      "loss": 3.4839,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.495922895208507e-06,
+      "loss": 3.289,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.494014146491528e-06,
+      "loss": 3.3863,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.492105158672523e-06,
+      "loss": 3.4407,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49019593205701e-06,
+      "loss": 3.5117,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.488286466950535e-06,
+      "loss": 3.352,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.486376763658695e-06,
+      "loss": 3.439,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.484466822487116e-06,
+      "loss": 3.4961,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.482556643741466e-06,
+      "loss": 3.3816,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.480646227727449e-06,
+      "loss": 3.395,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.47873557475081e-06,
+      "loss": 3.5016,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4768246851173285e-06,
+      "loss": 3.4262,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.474913559132823e-06,
+      "loss": 3.3769,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.473002197103149e-06,
+      "loss": 3.4373,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.471090599334203e-06,
+      "loss": 3.4366,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.469178766131917e-06,
+      "loss": 3.4083,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.46726669780226e-06,
+      "loss": 3.4771,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.46535439465124e-06,
+      "loss": 3.4966,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4634418569849e-06,
+      "loss": 3.4227,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.461529085109326e-06,
+      "loss": 3.3582,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.459616079330636e-06,
+      "loss": 3.4255,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.457702839954987e-06,
+      "loss": 3.5158,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.455789367288574e-06,
+      "loss": 3.3217,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.453875661637629e-06,
+      "loss": 3.4828,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.451961723308425e-06,
+      "loss": 3.4136,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.450047552607264e-06,
+      "loss": 3.4827,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.44813314984049e-06,
+      "loss": 3.4749,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.446218515314488e-06,
+      "loss": 3.4339,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.444303649335673e-06,
+      "loss": 3.5004,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.442388552210499e-06,
+      "loss": 3.459,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.440473224245462e-06,
+      "loss": 3.4297,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4385576657470895e-06,
+      "loss": 3.3419,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4366418770219455e-06,
+      "loss": 3.4584,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.434725858376636e-06,
+      "loss": 3.3513,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.432809610117798e-06,
+      "loss": 3.4153,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.430893132552109e-06,
+      "loss": 3.3959,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.428976425986283e-06,
+      "loss": 3.432,
+      "step": 3702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.427059490727069e-06,
+      "loss": 3.3837,
+      "step": 3703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.42514232708125e-06,
+      "loss": 3.4292,
+      "step": 3704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4232249353556565e-06,
+      "loss": 3.4432,
+      "step": 3705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.421307315857141e-06,
+      "loss": 3.3464,
+      "step": 3706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.419389468892602e-06,
+      "loss": 3.4136,
+      "step": 3707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4174713947689706e-06,
+      "loss": 3.4822,
+      "step": 3708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.415553093793217e-06,
+      "loss": 3.4664,
+      "step": 3709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.413634566272344e-06,
+      "loss": 3.5688,
+      "step": 3710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.411715812513395e-06,
+      "loss": 3.4928,
+      "step": 3711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.409796832823445e-06,
+      "loss": 3.4165,
+      "step": 3712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.407877627509611e-06,
+      "loss": 3.4613,
+      "step": 3713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4059581968790395e-06,
+      "loss": 3.4518,
+      "step": 3714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4040385412389175e-06,
+      "loss": 3.417,
+      "step": 3715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4021186608964645e-06,
+      "loss": 3.4583,
+      "step": 3716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.400198556158942e-06,
+      "loss": 3.44,
+      "step": 3717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3982782273336405e-06,
+      "loss": 3.3964,
+      "step": 3718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3963576747278925e-06,
+      "loss": 3.3945,
+      "step": 3719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.394436898649059e-06,
+      "loss": 3.3871,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.392515899404543e-06,
+      "loss": 3.4228,
+      "step": 3721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.390594677301784e-06,
+      "loss": 3.4698,
+      "step": 3722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.388673232648249e-06,
+      "loss": 3.4186,
+      "step": 3723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.38675156575145e-06,
+      "loss": 3.4477,
+      "step": 3724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.384829676918929e-06,
+      "loss": 3.4409,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.382907566458266e-06,
+      "loss": 3.3907,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.380985234677075e-06,
+      "loss": 3.3807,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.379062681883006e-06,
+      "loss": 3.3268,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.377139908383745e-06,
+      "loss": 3.4227,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3752169144870115e-06,
+      "loss": 3.4715,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.373293700500563e-06,
+      "loss": 3.3889,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.371370266732186e-06,
+      "loss": 3.3465,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.369446613489714e-06,
+      "loss": 3.4752,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.367522741081005e-06,
+      "loss": 3.3996,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.365598649813955e-06,
+      "loss": 3.4873,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.363674339996494e-06,
+      "loss": 3.4332,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.361749811936593e-06,
+      "loss": 3.3752,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.35982506594225e-06,
+      "loss": 3.5011,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.357900102321502e-06,
+      "loss": 3.321,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.355974921382422e-06,
+      "loss": 3.3859,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.354049523433115e-06,
+      "loss": 3.4358,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.352123908781719e-06,
+      "loss": 3.3629,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3501980777364125e-06,
+      "loss": 3.3811,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.348272030605406e-06,
+      "loss": 3.3846,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.346345767696941e-06,
+      "loss": 3.4505,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.344419289319299e-06,
+      "loss": 3.3842,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3424925957807935e-06,
+      "loss": 3.3744,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.340565687389773e-06,
+      "loss": 3.4362,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.338638564454616e-06,
+      "loss": 3.4799,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.336711227283744e-06,
+      "loss": 3.5681,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3347836761856065e-06,
+      "loss": 3.4372,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.332855911468687e-06,
+      "loss": 3.4348,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.330927933441508e-06,
+      "loss": 3.4477,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.32899974241262e-06,
+      "loss": 3.4581,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.327071338690614e-06,
+      "loss": 3.3266,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.32514272258411e-06,
+      "loss": 3.4064,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.323213894401762e-06,
+      "loss": 3.4336,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.321284854452262e-06,
+      "loss": 3.4259,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.319355603044334e-06,
+      "loss": 3.433,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.317426140486733e-06,
+      "loss": 3.591,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3154964670882514e-06,
+      "loss": 3.3627,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3135665831577154e-06,
+      "loss": 3.3967,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.311636489003983e-06,
+      "loss": 3.3594,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.309706184935945e-06,
+      "loss": 3.4912,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.307775671262526e-06,
+      "loss": 3.5225,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305844948292691e-06,
+      "loss": 3.4638,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.303914016335428e-06,
+      "loss": 3.4119,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.301982875699766e-06,
+      "loss": 3.4609,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.300051526694762e-06,
+      "loss": 3.3592,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.298119969629511e-06,
+      "loss": 3.3897,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.29618820481314e-06,
+      "loss": 3.4289,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.294256232554807e-06,
+      "loss": 3.3782,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.292324053163705e-06,
+      "loss": 3.3444,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.290391666949063e-06,
+      "loss": 3.4483,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.288459074220136e-06,
+      "loss": 3.4673,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.286526275286218e-06,
+      "loss": 3.3599,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.284593270456634e-06,
+      "loss": 3.4833,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.282660060040742e-06,
+      "loss": 3.3844,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.280726644347932e-06,
+      "loss": 3.379,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.27879302368763e-06,
+      "loss": 3.4295,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2768591983692915e-06,
+      "loss": 3.3852,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.274925168702406e-06,
+      "loss": 3.4242,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.272990934996495e-06,
+      "loss": 3.4769,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2710564975611145e-06,
+      "loss": 3.4333,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.269121856705851e-06,
+      "loss": 3.4367,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.267187012740324e-06,
+      "loss": 3.4284,
+      "step": 3786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.265251965974189e-06,
+      "loss": 3.4261,
+      "step": 3787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.263316716717126e-06,
+      "loss": 3.3768,
+      "step": 3788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.261381265278854e-06,
+      "loss": 3.3941,
+      "step": 3789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.259445611969126e-06,
+      "loss": 3.4934,
+      "step": 3790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.257509757097719e-06,
+      "loss": 3.4138,
+      "step": 3791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.255573700974452e-06,
+      "loss": 3.3556,
+      "step": 3792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.253637443909166e-06,
+      "loss": 3.4467,
+      "step": 3793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.251700986211745e-06,
+      "loss": 3.3974,
+      "step": 3794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.249764328192097e-06,
+      "loss": 3.4614,
+      "step": 3795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.247827470160165e-06,
+      "loss": 3.3647,
+      "step": 3796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.245890412425922e-06,
+      "loss": 3.3779,
+      "step": 3797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.243953155299378e-06,
+      "loss": 3.4217,
+      "step": 3798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24201569909057e-06,
+      "loss": 3.3289,
+      "step": 3799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.240078044109569e-06,
+      "loss": 3.3669,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.238140190666475e-06,
+      "loss": 3.3955,
+      "step": 3801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.236202139071427e-06,
+      "loss": 3.4813,
+      "step": 3802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.234263889634586e-06,
+      "loss": 3.4106,
+      "step": 3803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.232325442666152e-06,
+      "loss": 3.4254,
+      "step": 3804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.230386798476351e-06,
+      "loss": 3.464,
+      "step": 3805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.228447957375447e-06,
+      "loss": 3.4076,
+      "step": 3806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.226508919673733e-06,
+      "loss": 3.4069,
+      "step": 3807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.224569685681528e-06,
+      "loss": 3.4252,
+      "step": 3808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.22263025570919e-06,
+      "loss": 3.3838,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.220690630067107e-06,
+      "loss": 3.4642,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.218750809065694e-06,
+      "loss": 3.398,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.216810793015399e-06,
+      "loss": 3.4509,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.214870582226706e-06,
+      "loss": 3.424,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.212930177010121e-06,
+      "loss": 3.3722,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.210989577676192e-06,
+      "loss": 3.4338,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.209048784535489e-06,
+      "loss": 3.3473,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2071077978986194e-06,
+      "loss": 3.5154,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.205166618076215e-06,
+      "loss": 3.4005,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.203225245378944e-06,
+      "loss": 3.4169,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.201283680117506e-06,
+      "loss": 3.5047,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.199341922602626e-06,
+      "loss": 3.389,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.197399973145063e-06,
+      "loss": 3.3939,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.195457832055608e-06,
+      "loss": 3.2948,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.19351549964508e-06,
+      "loss": 3.3652,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.191572976224331e-06,
+      "loss": 3.3125,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1896302621042425e-06,
+      "loss": 3.3738,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.187687357595725e-06,
+      "loss": 3.5769,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.185744263009723e-06,
+      "loss": 3.3578,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1838009786572086e-06,
+      "loss": 3.4268,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.181857504849183e-06,
+      "loss": 3.4071,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.179913841896682e-06,
+      "loss": 3.3812,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.17796999011077e-06,
+      "loss": 3.3693,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.176025949802539e-06,
+      "loss": 3.3876,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.174081721283114e-06,
+      "loss": 3.4976,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.17213730486365e-06,
+      "loss": 3.439,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1701927008553305e-06,
+      "loss": 3.4336,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.168247909569372e-06,
+      "loss": 3.5033,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.166302931317014e-06,
+      "loss": 3.4642,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.164357766409536e-06,
+      "loss": 3.4041,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.162412415158241e-06,
+      "loss": 3.358,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1604668778744605e-06,
+      "loss": 3.4543,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.15852115486956e-06,
+      "loss": 3.4018,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1565752464549345e-06,
+      "loss": 3.5084,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.154629152942004e-06,
+      "loss": 3.5086,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.152682874642225e-06,
+      "loss": 3.4042,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.150736411867077e-06,
+      "loss": 3.5616,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.148789764928073e-06,
+      "loss": 3.4128,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.146842934136753e-06,
+      "loss": 3.5049,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.14489591980469e-06,
+      "loss": 3.3708,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.142948722243483e-06,
+      "loss": 3.4544,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.141001341764762e-06,
+      "loss": 3.445,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.139053778680185e-06,
+      "loss": 3.4776,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.137106033301441e-06,
+      "loss": 3.5389,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.135158105940244e-06,
+      "loss": 3.3736,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.133209996908346e-06,
+      "loss": 3.5305,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.131261706517518e-06,
+      "loss": 3.4484,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.129313235079567e-06,
+      "loss": 3.4216,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.127364582906323e-06,
+      "loss": 3.4917,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.125415750309651e-06,
+      "loss": 3.4444,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.123466737601442e-06,
+      "loss": 3.3368,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.121517545093616e-06,
+      "loss": 3.5524,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.11956817309812e-06,
+      "loss": 3.4349,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.117618621926933e-06,
+      "loss": 3.3351,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.115668891892062e-06,
+      "loss": 3.4128,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1137189833055396e-06,
+      "loss": 3.4519,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.111768896479431e-06,
+      "loss": 3.4185,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.109818631725827e-06,
+      "loss": 3.4761,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.107868189356848e-06,
+      "loss": 3.5089,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.105917569684643e-06,
+      "loss": 3.4602,
+      "step": 3869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.10396677302139e-06,
+      "loss": 3.4033,
+      "step": 3870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.102015799679293e-06,
+      "loss": 3.5089,
+      "step": 3871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.100064649970587e-06,
+      "loss": 3.405,
+      "step": 3872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0981133242075355e-06,
+      "loss": 3.3718,
+      "step": 3873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.096161822702424e-06,
+      "loss": 3.4137,
+      "step": 3874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.094210145767575e-06,
+      "loss": 3.393,
+      "step": 3875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.092258293715334e-06,
+      "loss": 3.4663,
+      "step": 3876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.090306266858074e-06,
+      "loss": 3.4271,
+      "step": 3877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.088354065508198e-06,
+      "loss": 3.416,
+      "step": 3878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.086401689978136e-06,
+      "loss": 3.4742,
+      "step": 3879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0844491405803485e-06,
+      "loss": 3.4326,
+      "step": 3880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.082496417627319e-06,
+      "loss": 3.3925,
+      "step": 3881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0805435214315636e-06,
+      "loss": 3.4383,
+      "step": 3882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.078590452305619e-06,
+      "loss": 3.5154,
+      "step": 3883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.076637210562059e-06,
+      "loss": 3.3634,
+      "step": 3884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.074683796513481e-06,
+      "loss": 3.4804,
+      "step": 3885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.072730210472506e-06,
+      "loss": 3.4329,
+      "step": 3886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.070776452751785e-06,
+      "loss": 3.4623,
+      "step": 3887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.068822523664003e-06,
+      "loss": 3.4628,
+      "step": 3888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.066868423521862e-06,
+      "loss": 3.3918,
+      "step": 3889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.064914152638099e-06,
+      "loss": 3.3967,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.062959711325473e-06,
+      "loss": 3.36,
+      "step": 3891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.061005099896774e-06,
+      "loss": 3.4411,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.059050318664819e-06,
+      "loss": 3.455,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.057095367942449e-06,
+      "loss": 3.3894,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.055140248042536e-06,
+      "loss": 3.3919,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.053184959277977e-06,
+      "loss": 3.4165,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.051229501961697e-06,
+      "loss": 3.4414,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0492738764066465e-06,
+      "loss": 3.5039,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.047318082925803e-06,
+      "loss": 3.4724,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.045362121832173e-06,
+      "loss": 3.3273,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.043405993438789e-06,
+      "loss": 3.3779,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.041449698058709e-06,
+      "loss": 3.469,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.039493236005017e-06,
+      "loss": 3.4562,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.037536607590829e-06,
+      "loss": 3.522,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.035579813129281e-06,
+      "loss": 3.4272,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.033622852933538e-06,
+      "loss": 3.4398,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.031665727316793e-06,
+      "loss": 3.4172,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0297084365922655e-06,
+      "loss": 3.3979,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.027750981073198e-06,
+      "loss": 3.4332,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0257933610728625e-06,
+      "loss": 3.4887,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.023835576904558e-06,
+      "loss": 3.4086,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.021877628881606e-06,
+      "loss": 3.4187,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.019919517317356e-06,
+      "loss": 3.4389,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.017961242525189e-06,
+      "loss": 3.4038,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.016002804818502e-06,
+      "loss": 3.4595,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.014044204510725e-06,
+      "loss": 3.395,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.012085441915314e-06,
+      "loss": 3.4295,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.010126517345748e-06,
+      "loss": 3.3921,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.008167431115532e-06,
+      "loss": 3.4195,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.006208183538201e-06,
+      "loss": 3.4115,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0042487749273115e-06,
+      "loss": 3.4219,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.002289205596447e-06,
+      "loss": 3.4381,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.000329475859217e-06,
+      "loss": 3.4107,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.998369586029258e-06,
+      "loss": 3.3846,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.996409536420229e-06,
+      "loss": 3.3787,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.994449327345819e-06,
+      "loss": 3.4121,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.992488959119734e-06,
+      "loss": 3.5011,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.990528432055718e-06,
+      "loss": 3.4404,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.988567746467531e-06,
+      "loss": 3.4873,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.98660690266896e-06,
+      "loss": 3.4121,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.98464590097382e-06,
+      "loss": 3.438,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9826847416959485e-06,
+      "loss": 3.4787,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9807234251492116e-06,
+      "loss": 3.4583,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.978761951647497e-06,
+      "loss": 3.3876,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.976800321504718e-06,
+      "loss": 3.479,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.974838535034814e-06,
+      "loss": 3.4793,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.972876592551753e-06,
+      "loss": 3.3569,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.97091449436952e-06,
+      "loss": 3.42,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9689522408021295e-06,
+      "loss": 3.4168,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.966989832163623e-06,
+      "loss": 3.468,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.965027268768063e-06,
+      "loss": 3.3956,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.963064550929538e-06,
+      "loss": 3.3627,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9611016789621624e-06,
+      "loss": 3.4046,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.959138653180071e-06,
+      "loss": 3.4122,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.957175473897431e-06,
+      "loss": 3.3511,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.955212141428427e-06,
+      "loss": 3.3973,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.95324865608727e-06,
+      "loss": 3.442,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.951285018188199e-06,
+      "loss": 3.4128,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9493212280454716e-06,
+      "loss": 3.4466,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9473572859733755e-06,
+      "loss": 3.4957,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.945393192286217e-06,
+      "loss": 3.4992,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943428947298333e-06,
+      "loss": 3.3663,
+      "step": 3952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.941464551324079e-06,
+      "loss": 3.3997,
+      "step": 3953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9395000046778365e-06,
+      "loss": 3.4712,
+      "step": 3954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9375353076740135e-06,
+      "loss": 3.533,
+      "step": 3955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.93557046062704e-06,
+      "loss": 3.388,
+      "step": 3956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9336054638513675e-06,
+      "loss": 3.3673,
+      "step": 3957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9316403176614765e-06,
+      "loss": 3.5185,
+      "step": 3958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.92967502237187e-06,
+      "loss": 3.428,
+      "step": 3959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9277095782970695e-06,
+      "loss": 3.5703,
+      "step": 3960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.92574398575163e-06,
+      "loss": 3.4868,
+      "step": 3961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.923778245050121e-06,
+      "loss": 3.3981,
+      "step": 3962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.921812356507142e-06,
+      "loss": 3.3174,
+      "step": 3963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.91984632043731e-06,
+      "loss": 3.5034,
+      "step": 3964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9178801371552755e-06,
+      "loss": 3.3855,
+      "step": 3965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.915913806975701e-06,
+      "loss": 3.4218,
+      "step": 3966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.913947330213282e-06,
+      "loss": 3.4757,
+      "step": 3967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9119807071827295e-06,
+      "loss": 3.4492,
+      "step": 3968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9100139381987865e-06,
+      "loss": 3.4474,
+      "step": 3969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.90804702357621e-06,
+      "loss": 3.4254,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.906079963629788e-06,
+      "loss": 3.4144,
+      "step": 3971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.904112758674327e-06,
+      "loss": 3.4942,
+      "step": 3972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.902145409024661e-06,
+      "loss": 3.5044,
+      "step": 3973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.900177914995643e-06,
+      "loss": 3.3731,
+      "step": 3974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.898210276902151e-06,
+      "loss": 3.3507,
+      "step": 3975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.896242495059083e-06,
+      "loss": 3.4727,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8942745697813684e-06,
+      "loss": 3.432,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.89230650138395e-06,
+      "loss": 3.4183,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.890338290181799e-06,
+      "loss": 3.4172,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.888369936489906e-06,
+      "loss": 3.4028,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8864014406232886e-06,
+      "loss": 3.4033,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.884432802896984e-06,
+      "loss": 3.3973,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8824640236260535e-06,
+      "loss": 3.4512,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.880495103125579e-06,
+      "loss": 3.3719,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.878526041710667e-06,
+      "loss": 3.3552,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.876556839696448e-06,
+      "loss": 3.4162,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.874587497398072e-06,
+      "loss": 3.4965,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.872618015130711e-06,
+      "loss": 3.3927,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.870648393209563e-06,
+      "loss": 3.4568,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.868678631949845e-06,
+      "loss": 3.3661,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.866708731666799e-06,
+      "loss": 3.3325,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.864738692675686e-06,
+      "loss": 3.3287,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.862768515291792e-06,
+      "loss": 3.3718,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.860798199830425e-06,
+      "loss": 3.4484,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.858827746606912e-06,
+      "loss": 3.4475,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.856857155936607e-06,
+      "loss": 3.4556,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.854886428134882e-06,
+      "loss": 3.4558,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.852915563517132e-06,
+      "loss": 3.4705,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.850944562398776e-06,
+      "loss": 3.3541,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84897342509525e-06,
+      "loss": 3.4248,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8470021519220165e-06,
+      "loss": 3.443,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8450307431945586e-06,
+      "loss": 3.3494,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84305919922838e-06,
+      "loss": 3.4133,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.841087520339006e-06,
+      "loss": 3.4129,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.839115706841984e-06,
+      "loss": 3.4384,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.837143759052886e-06,
+      "loss": 3.4828,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.835171677287299e-06,
+      "loss": 3.3182,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.833199461860837e-06,
+      "loss": 3.4036,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8312271130891305e-06,
+      "loss": 3.4015,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8292546312878376e-06,
+      "loss": 3.3901,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.827282016772633e-06,
+      "loss": 3.4128,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.825309269859213e-06,
+      "loss": 3.4396,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.823336390863297e-06,
+      "loss": 3.3372,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8213633801006246e-06,
+      "loss": 3.4363,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.819390237886956e-06,
+      "loss": 3.3901,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.817416964538074e-06,
+      "loss": 3.4396,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.815443560369779e-06,
+      "loss": 3.4187,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.813470025697897e-06,
+      "loss": 3.3123,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.811496360838271e-06,
+      "loss": 3.4692,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.809522566106768e-06,
+      "loss": 3.4154,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.807548641819272e-06,
+      "loss": 3.3352,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.805574588291691e-06,
+      "loss": 3.4533,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.803600405839952e-06,
+      "loss": 3.3929,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.801626094780004e-06,
+      "loss": 3.3271,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.799651655427814e-06,
+      "loss": 3.3899,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7976770880993725e-06,
+      "loss": 3.4093,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795702393110692e-06,
+      "loss": 3.3682,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793727570777797e-06,
+      "loss": 3.4544,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79175262141674e-06,
+      "loss": 3.442,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789777545343597e-06,
+      "loss": 3.4654,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787802342874452e-06,
+      "loss": 3.482,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.785827014325421e-06,
+      "loss": 3.5099,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.783851560012632e-06,
+      "loss": 3.3807,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.781875980252241e-06,
+      "loss": 3.4382,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.779900275360417e-06,
+      "loss": 3.4452,
+      "step": 4035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777924445653352e-06,
+      "loss": 3.4336,
+      "step": 4036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.775948491447258e-06,
+      "loss": 3.3752,
+      "step": 4037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773972413058367e-06,
+      "loss": 3.43,
+      "step": 4038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.77199621080293e-06,
+      "loss": 3.4485,
+      "step": 4039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.77001988499722e-06,
+      "loss": 3.3479,
+      "step": 4040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7680434359575265e-06,
+      "loss": 3.3536,
+      "step": 4041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.766066864000159e-06,
+      "loss": 3.399,
+      "step": 4042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.764090169441453e-06,
+      "loss": 3.4044,
+      "step": 4043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.762113352597755e-06,
+      "loss": 3.3696,
+      "step": 4044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.760136413785434e-06,
+      "loss": 3.5103,
+      "step": 4045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.758159353320883e-06,
+      "loss": 3.4132,
+      "step": 4046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.756182171520507e-06,
+      "loss": 3.3984,
+      "step": 4047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.754204868700737e-06,
+      "loss": 3.4039,
+      "step": 4048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.752227445178017e-06,
+      "loss": 3.4668,
+      "step": 4049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.750249901268819e-06,
+      "loss": 3.3876,
+      "step": 4050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.748272237289624e-06,
+      "loss": 3.3875,
+      "step": 4051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.746294453556941e-06,
+      "loss": 3.4852,
+      "step": 4052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.744316550387291e-06,
+      "loss": 3.3741,
+      "step": 4053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7423385280972225e-06,
+      "loss": 3.3706,
+      "step": 4054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.740360387003295e-06,
+      "loss": 3.5075,
+      "step": 4055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7383821274220884e-06,
+      "loss": 3.3567,
+      "step": 4056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.736403749670206e-06,
+      "loss": 3.4466,
+      "step": 4057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.734425254064269e-06,
+      "loss": 3.3834,
+      "step": 4058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.732446640920912e-06,
+      "loss": 3.425,
+      "step": 4059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7304679105567966e-06,
+      "loss": 3.4189,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.728489063288593e-06,
+      "loss": 3.4687,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.726510099433e-06,
+      "loss": 3.4691,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.724531019306732e-06,
+      "loss": 3.4856,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.722551823226517e-06,
+      "loss": 3.5017,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.720572511509108e-06,
+      "loss": 3.3319,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.718593084471275e-06,
+      "loss": 3.5295,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.716613542429804e-06,
+      "loss": 3.4804,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.714633885701502e-06,
+      "loss": 3.5218,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.712654114603189e-06,
+      "loss": 3.3331,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.710674229451714e-06,
+      "loss": 3.3984,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.708694230563934e-06,
+      "loss": 3.51,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.706714118256729e-06,
+      "loss": 3.4712,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.704733892846996e-06,
+      "loss": 3.3637,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.702753554651648e-06,
+      "loss": 3.4069,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7007731039876235e-06,
+      "loss": 3.4129,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.69879254117187e-06,
+      "loss": 3.3847,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.696811866521359e-06,
+      "loss": 3.4639,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.694831080353074e-06,
+      "loss": 3.3644,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.692850182984024e-06,
+      "loss": 3.3602,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.690869174731229e-06,
+      "loss": 3.4112,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.688888055911731e-06,
+      "loss": 3.3463,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.686906826842588e-06,
+      "loss": 3.3796,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.684925487840878e-06,
+      "loss": 3.4801,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.68294403922369e-06,
+      "loss": 3.4329,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.680962481308139e-06,
+      "loss": 3.487,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.678980814411351e-06,
+      "loss": 3.4562,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.676999038850473e-06,
+      "loss": 3.3927,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.675017154942668e-06,
+      "loss": 3.4301,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.673035163005118e-06,
+      "loss": 3.3622,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.671053063355019e-06,
+      "loss": 3.5363,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.669070856309588e-06,
+      "loss": 3.3371,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.667088542186056e-06,
+      "loss": 3.428,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.665106121301675e-06,
+      "loss": 3.3956,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.663123593973707e-06,
+      "loss": 3.4173,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6611409605194415e-06,
+      "loss": 3.4139,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.659158221256175e-06,
+      "loss": 3.4078,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.657175376501226e-06,
+      "loss": 3.3304,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.655192426571929e-06,
+      "loss": 3.4746,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.653209371785636e-06,
+      "loss": 3.4625,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.651226212459715e-06,
+      "loss": 3.4687,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.649242948911551e-06,
+      "loss": 3.4471,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.647259581458543e-06,
+      "loss": 3.4491,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.645276110418112e-06,
+      "loss": 3.4368,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.643292536107693e-06,
+      "loss": 3.3563,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.641308858844735e-06,
+      "loss": 3.4395,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.639325078946706e-06,
+      "loss": 3.444,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.637341196731092e-06,
+      "loss": 3.4207,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.635357212515392e-06,
+      "loss": 3.3747,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.633373126617124e-06,
+      "loss": 3.3388,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.631388939353819e-06,
+      "loss": 3.4271,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.629404651043029e-06,
+      "loss": 3.4804,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.627420262002318e-06,
+      "loss": 3.4308,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.625435772549267e-06,
+      "loss": 3.2969,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.623451183001475e-06,
+      "loss": 3.3339,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.621466493676555e-06,
+      "loss": 3.3815,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6194817048921365e-06,
+      "loss": 3.408,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.617496816965867e-06,
+      "loss": 3.4134,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6155118302154045e-06,
+      "loss": 3.4009,
+      "step": 4118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.61352674495843e-06,
+      "loss": 3.5508,
+      "step": 4119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.611541561512634e-06,
+      "loss": 3.3375,
+      "step": 4120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.609556280195726e-06,
+      "loss": 3.4422,
+      "step": 4121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.60757090132543e-06,
+      "loss": 3.4251,
+      "step": 4122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6055854252194875e-06,
+      "loss": 3.4289,
+      "step": 4123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.603599852195653e-06,
+      "loss": 3.3734,
+      "step": 4124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.601614182571696e-06,
+      "loss": 3.4157,
+      "step": 4125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5996284166654065e-06,
+      "loss": 3.4551,
+      "step": 4126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.597642554794586e-06,
+      "loss": 3.4508,
+      "step": 4127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.595656597277047e-06,
+      "loss": 3.4796,
+      "step": 4128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.593670544430628e-06,
+      "loss": 3.3617,
+      "step": 4129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.591684396573175e-06,
+      "loss": 3.5174,
+      "step": 4130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.58969815402255e-06,
+      "loss": 3.4303,
+      "step": 4131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.587711817096632e-06,
+      "loss": 3.3643,
+      "step": 4132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5857253861133145e-06,
+      "loss": 3.3779,
+      "step": 4133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.583738861390505e-06,
+      "loss": 3.528,
+      "step": 4134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.581752243246129e-06,
+      "loss": 3.4238,
+      "step": 4135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.579765531998122e-06,
+      "loss": 3.4219,
+      "step": 4136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.577778727964442e-06,
+      "loss": 3.3627,
+      "step": 4137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.575791831463051e-06,
+      "loss": 3.4899,
+      "step": 4138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.573804842811935e-06,
+      "loss": 3.5616,
+      "step": 4139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.571817762329092e-06,
+      "loss": 3.3396,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.569830590332535e-06,
+      "loss": 3.5094,
+      "step": 4141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.56784332714029e-06,
+      "loss": 3.395,
+      "step": 4142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.565855973070397e-06,
+      "loss": 3.4002,
+      "step": 4143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.563868528440913e-06,
+      "loss": 3.3735,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.561880993569912e-06,
+      "loss": 3.4769,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.559893368775474e-06,
+      "loss": 3.3813,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.557905654375702e-06,
+      "loss": 3.4432,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.555917850688708e-06,
+      "loss": 3.2576,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.553929958032621e-06,
+      "loss": 3.3274,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5519419767255845e-06,
+      "loss": 3.4414,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.549953907085751e-06,
+      "loss": 3.4259,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.547965749431296e-06,
+      "loss": 3.4737,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.545977504080401e-06,
+      "loss": 3.3512,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.543989171351266e-06,
+      "loss": 3.4284,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.542000751562103e-06,
+      "loss": 3.4034,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.540012245031141e-06,
+      "loss": 3.4441,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.538023652076618e-06,
+      "loss": 3.426,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.53603497301679e-06,
+      "loss": 3.3781,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.534046208169923e-06,
+      "loss": 3.4466,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5320573578543015e-06,
+      "loss": 3.4284,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5300684223882196e-06,
+      "loss": 3.4182,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.528079402089987e-06,
+      "loss": 3.4115,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.526090297277926e-06,
+      "loss": 3.4163,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.524101108270376e-06,
+      "loss": 3.348,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.522111835385682e-06,
+      "loss": 3.4031,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.520122478942209e-06,
+      "loss": 3.4323,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5181330392583356e-06,
+      "loss": 3.4167,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5161435166524495e-06,
+      "loss": 3.5148,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.514153911442954e-06,
+      "loss": 3.4285,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.512164223948268e-06,
+      "loss": 3.3716,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.510174454486817e-06,
+      "loss": 3.3378,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.508184603377047e-06,
+      "loss": 3.4996,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5061946709374125e-06,
+      "loss": 3.4004,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5042046574863824e-06,
+      "loss": 3.4306,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5022145633424385e-06,
+      "loss": 3.5328,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.500224388824077e-06,
+      "loss": 3.3634,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.498234134249802e-06,
+      "loss": 3.4473,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.496243799938138e-06,
+      "loss": 3.4377,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.494253386207613e-06,
+      "loss": 3.4883,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.492262893376777e-06,
+      "loss": 3.3856,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4902723217641885e-06,
+      "loss": 3.379,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4882816716884164e-06,
+      "loss": 3.4219,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.486290943468044e-06,
+      "loss": 3.4029,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.48430013742167e-06,
+      "loss": 3.4334,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.482309253867901e-06,
+      "loss": 3.4078,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.480318293125359e-06,
+      "loss": 3.484,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.478327255512677e-06,
+      "loss": 3.4085,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.476336141348501e-06,
+      "loss": 3.4279,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.474344950951489e-06,
+      "loss": 3.4414,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.47235368464031e-06,
+      "loss": 3.4081,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.470362342733646e-06,
+      "loss": 3.4154,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.468370925550194e-06,
+      "loss": 3.4114,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.466379433408657e-06,
+      "loss": 3.3595,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.464387866627757e-06,
+      "loss": 3.4375,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.462396225526221e-06,
+      "loss": 3.4954,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.460404510422793e-06,
+      "loss": 3.4612,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4584127216362285e-06,
+      "loss": 3.4234,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.456420859485289e-06,
+      "loss": 3.5496,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.454428924288756e-06,
+      "loss": 3.4663,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.452436916365418e-06,
+      "loss": 3.4596,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.450444836034075e-06,
+      "loss": 3.3397,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.448452683613541e-06,
+      "loss": 3.2104,
+      "step": 4202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.44646045942264e-06,
+      "loss": 3.4762,
+      "step": 4203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4444681637802045e-06,
+      "loss": 3.4586,
+      "step": 4204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.442475797005085e-06,
+      "loss": 3.4252,
+      "step": 4205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.440483359416138e-06,
+      "loss": 3.391,
+      "step": 4206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.438490851332234e-06,
+      "loss": 3.5088,
+      "step": 4207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.436498273072254e-06,
+      "loss": 3.4267,
+      "step": 4208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.43450562495509e-06,
+      "loss": 3.4952,
+      "step": 4209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.432512907299645e-06,
+      "loss": 3.5055,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.430520120424834e-06,
+      "loss": 3.3838,
+      "step": 4211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.428527264649583e-06,
+      "loss": 3.3113,
+      "step": 4212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.426534340292829e-06,
+      "loss": 3.381,
+      "step": 4213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.424541347673518e-06,
+      "loss": 3.4303,
+      "step": 4214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4225482871106095e-06,
+      "loss": 3.4044,
+      "step": 4215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.420555158923072e-06,
+      "loss": 3.3903,
+      "step": 4216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.418561963429887e-06,
+      "loss": 3.3225,
+      "step": 4217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.416568700950047e-06,
+      "loss": 3.5081,
+      "step": 4218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.41457537180255e-06,
+      "loss": 3.4193,
+      "step": 4219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.412581976306409e-06,
+      "loss": 3.4011,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.41058851478065e-06,
+      "loss": 3.4283,
+      "step": 4221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.408594987544306e-06,
+      "loss": 3.5257,
+      "step": 4222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4066013949164175e-06,
+      "loss": 3.4139,
+      "step": 4223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.404607737216042e-06,
+      "loss": 3.4787,
+      "step": 4224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.402614014762244e-06,
+      "loss": 3.3985,
+      "step": 4225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4006202278741e-06,
+      "loss": 3.4993,
+      "step": 4226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.398626376870693e-06,
+      "loss": 3.423,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.396632462071121e-06,
+      "loss": 3.3843,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.39463848379449e-06,
+      "loss": 3.4219,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.392644442359914e-06,
+      "loss": 3.353,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3906503380865235e-06,
+      "loss": 3.4584,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3886561712934504e-06,
+      "loss": 3.4851,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.386661942299844e-06,
+      "loss": 3.3766,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3846676514248615e-06,
+      "loss": 3.416,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.382673298987667e-06,
+      "loss": 3.4799,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.380678885307437e-06,
+      "loss": 3.3776,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.378684410703357e-06,
+      "loss": 3.4713,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.376689875494626e-06,
+      "loss": 3.4841,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.374695280000446e-06,
+      "loss": 3.5035,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.372700624540032e-06,
+      "loss": 3.3795,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3707059094326105e-06,
+      "loss": 3.382,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.368711134997415e-06,
+      "loss": 3.483,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3667163015536906e-06,
+      "loss": 3.4152,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.364721409420687e-06,
+      "loss": 3.4189,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.36272645891767e-06,
+      "loss": 3.5604,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.36073145036391e-06,
+      "loss": 3.4478,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3587363840786876e-06,
+      "loss": 3.3998,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.356741260381294e-06,
+      "loss": 3.3849,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.35474607959103e-06,
+      "loss": 3.4557,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3527508420272025e-06,
+      "loss": 3.4253,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.350755548009131e-06,
+      "loss": 3.4285,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.34876019785614e-06,
+      "loss": 3.3399,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.346764791887566e-06,
+      "loss": 3.4793,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.344769330422756e-06,
+      "loss": 3.425,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3427738137810615e-06,
+      "loss": 3.3125,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.340778242281845e-06,
+      "loss": 3.4606,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.338782616244479e-06,
+      "loss": 3.5485,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.336786935988344e-06,
+      "loss": 3.3753,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.334791201832826e-06,
+      "loss": 3.4149,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.332795414097324e-06,
+      "loss": 3.3293,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.330799573101244e-06,
+      "loss": 3.4426,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.328803679164001e-06,
+      "loss": 3.4419,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.326807732605016e-06,
+      "loss": 3.4283,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.32481173374372e-06,
+      "loss": 3.4661,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.322815682899556e-06,
+      "loss": 3.4133,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.320819580391969e-06,
+      "loss": 3.3746,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.318823426540416e-06,
+      "loss": 3.3464,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.316827221664363e-06,
+      "loss": 3.349,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.314830966083279e-06,
+      "loss": 3.4925,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3128346601166484e-06,
+      "loss": 3.3513,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.310838304083957e-06,
+      "loss": 3.6425,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.308841898304705e-06,
+      "loss": 3.4744,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.306845443098394e-06,
+      "loss": 3.4411,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3048489387845396e-06,
+      "loss": 3.3732,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3028523856826595e-06,
+      "loss": 3.3358,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3008557841122844e-06,
+      "loss": 3.3759,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2988591343929475e-06,
+      "loss": 3.3392,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.296862436844196e-06,
+      "loss": 3.358,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.29486569178558e-06,
+      "loss": 3.458,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.292868899536659e-06,
+      "loss": 3.3481,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.290872060416997e-06,
+      "loss": 3.4366,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.288875174746172e-06,
+      "loss": 3.4579,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.286878242843762e-06,
+      "loss": 3.3876,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.28488126502936e-06,
+      "loss": 3.4696,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.282884241622559e-06,
+      "loss": 3.4706,
+      "step": 4285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.280887172942965e-06,
+      "loss": 3.3331,
+      "step": 4286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.278890059310187e-06,
+      "loss": 3.3742,
+      "step": 4287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.276892901043843e-06,
+      "loss": 3.3688,
+      "step": 4288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.274895698463558e-06,
+      "loss": 3.4019,
+      "step": 4289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.272898451888967e-06,
+      "loss": 3.4296,
+      "step": 4290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.270901161639707e-06,
+      "loss": 3.3809,
+      "step": 4291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.268903828035425e-06,
+      "loss": 3.5389,
+      "step": 4292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.266906451395774e-06,
+      "loss": 3.5241,
+      "step": 4293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2649090320404136e-06,
+      "loss": 3.4572,
+      "step": 4294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.262911570289013e-06,
+      "loss": 3.4367,
+      "step": 4295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.260914066461243e-06,
+      "loss": 3.4036,
+      "step": 4296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.258916520876787e-06,
+      "loss": 3.4523,
+      "step": 4297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.256918933855331e-06,
+      "loss": 3.4072,
+      "step": 4298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.254921305716567e-06,
+      "loss": 3.4646,
+      "step": 4299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2529236367801986e-06,
+      "loss": 3.4321,
+      "step": 4300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.25092592736593e-06,
+      "loss": 3.4426,
+      "step": 4301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.248928177793475e-06,
+      "loss": 3.4254,
+      "step": 4302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.246930388382556e-06,
+      "loss": 3.3221,
+      "step": 4303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.244932559452898e-06,
+      "loss": 3.3996,
+      "step": 4304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.242934691324231e-06,
+      "loss": 3.3138,
+      "step": 4305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.240936784316297e-06,
+      "loss": 3.4286,
+      "step": 4306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.23893883874884e-06,
+      "loss": 3.4362,
+      "step": 4307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.236940854941611e-06,
+      "loss": 3.4937,
+      "step": 4308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.234942833214367e-06,
+      "loss": 3.3647,
+      "step": 4309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.232944773886872e-06,
+      "loss": 3.4366,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.230946677278895e-06,
+      "loss": 3.4257,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.228948543710213e-06,
+      "loss": 3.429,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.226950373500604e-06,
+      "loss": 3.3725,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.224952166969858e-06,
+      "loss": 3.3928,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.222953924437767e-06,
+      "loss": 3.4903,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22095564622413e-06,
+      "loss": 3.4295,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2189573326487495e-06,
+      "loss": 3.4808,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.216958984031438e-06,
+      "loss": 3.4018,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.214960600692012e-06,
+      "loss": 3.4643,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.212962182950291e-06,
+      "loss": 3.3674,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.210963731126101e-06,
+      "loss": 3.3211,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.208965245539276e-06,
+      "loss": 3.4241,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.206966726509652e-06,
+      "loss": 3.4673,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2049681743570745e-06,
+      "loss": 3.3625,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.202969589401389e-06,
+      "loss": 3.3658,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.200970971962451e-06,
+      "loss": 3.4031,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1989723223601185e-06,
+      "loss": 3.368,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.196973640914256e-06,
+      "loss": 3.384,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.194974927944729e-06,
+      "loss": 3.4309,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.192976183771417e-06,
+      "loss": 3.3492,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.190977408714196e-06,
+      "loss": 3.4056,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.188978603092951e-06,
+      "loss": 3.4473,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.186979767227569e-06,
+      "loss": 3.3854,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.184980901437945e-06,
+      "loss": 3.4888,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1829820060439774e-06,
+      "loss": 3.3712,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.18098308136557e-06,
+      "loss": 3.5064,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.17898412772263e-06,
+      "loss": 3.4361,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1769851454350695e-06,
+      "loss": 3.3815,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.174986134822807e-06,
+      "loss": 3.385,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.172987096205761e-06,
+      "loss": 3.4746,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.170988029903863e-06,
+      "loss": 3.3808,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.16898893623704e-06,
+      "loss": 3.3832,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.166989815525226e-06,
+      "loss": 3.3555,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1649906680883635e-06,
+      "loss": 3.3999,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1629914942463945e-06,
+      "loss": 3.4007,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.160992294319265e-06,
+      "loss": 3.4818,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.15899306862693e-06,
+      "loss": 3.5023,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.156993817489343e-06,
+      "loss": 3.3462,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.154994541226467e-06,
+      "loss": 3.4132,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1529952401582615e-06,
+      "loss": 3.4492,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.150995914604701e-06,
+      "loss": 3.4117,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.148996564885752e-06,
+      "loss": 3.4406,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1469971913213945e-06,
+      "loss": 3.4067,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.144997794231604e-06,
+      "loss": 3.4137,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.142998373936368e-06,
+      "loss": 3.3938,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1409989307556725e-06,
+      "loss": 3.3911,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1389994650095074e-06,
+      "loss": 3.4097,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.136999977017867e-06,
+      "loss": 3.5114,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.135000467100751e-06,
+      "loss": 3.3299,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1330009355781605e-06,
+      "loss": 3.451,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.131001382770101e-06,
+      "loss": 3.4271,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.129001808996578e-06,
+      "loss": 3.3311,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.127002214577607e-06,
+      "loss": 3.4237,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.125002599833201e-06,
+      "loss": 3.4267,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.123002965083379e-06,
+      "loss": 3.343,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.121003310648161e-06,
+      "loss": 3.3081,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.119003636847574e-06,
+      "loss": 3.3754,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.117003944001645e-06,
+      "loss": 3.42,
+      "step": 4368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.115004232430405e-06,
+      "loss": 3.4173,
+      "step": 4369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1130045024538864e-06,
+      "loss": 3.3142,
+      "step": 4370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.111004754392128e-06,
+      "loss": 3.3909,
+      "step": 4371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.10900498856517e-06,
+      "loss": 3.4572,
+      "step": 4372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.107005205293052e-06,
+      "loss": 3.524,
+      "step": 4373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1050054048958195e-06,
+      "loss": 3.419,
+      "step": 4374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1030055876935245e-06,
+      "loss": 3.4817,
+      "step": 4375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.101005754006214e-06,
+      "loss": 3.3521,
+      "step": 4376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.099005904153943e-06,
+      "loss": 3.4781,
+      "step": 4377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0970060384567665e-06,
+      "loss": 3.3541,
+      "step": 4378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.095006157234744e-06,
+      "loss": 3.3795,
+      "step": 4379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.093006260807935e-06,
+      "loss": 3.3203,
+      "step": 4380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.091006349496405e-06,
+      "loss": 3.3773,
+      "step": 4381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.089006423620219e-06,
+      "loss": 3.4426,
+      "step": 4382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.087006483499445e-06,
+      "loss": 3.3835,
+      "step": 4383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.085006529454151e-06,
+      "loss": 3.4057,
+      "step": 4384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.083006561804414e-06,
+      "loss": 3.4424,
+      "step": 4385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.081006580870307e-06,
+      "loss": 3.3836,
+      "step": 4386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.079006586971905e-06,
+      "loss": 3.3435,
+      "step": 4387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.07700658042929e-06,
+      "loss": 3.421,
+      "step": 4388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.075006561562542e-06,
+      "loss": 3.4132,
+      "step": 4389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.073006530691743e-06,
+      "loss": 3.4192,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0710064881369805e-06,
+      "loss": 3.3171,
+      "step": 4391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.069006434218338e-06,
+      "loss": 3.3982,
+      "step": 4392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.067006369255907e-06,
+      "loss": 3.4142,
+      "step": 4393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.065006293569775e-06,
+      "loss": 3.4757,
+      "step": 4394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.063006207480037e-06,
+      "loss": 3.3846,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0610061113067845e-06,
+      "loss": 3.3433,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.059006005370115e-06,
+      "loss": 3.4148,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0570058899901245e-06,
+      "loss": 3.394,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.055005765486909e-06,
+      "loss": 3.4092,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.053005632180573e-06,
+      "loss": 3.4003,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0510054903912155e-06,
+      "loss": 3.4682,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.04900534043894e-06,
+      "loss": 3.4086,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.047005182643847e-06,
+      "loss": 3.432,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.045005017326046e-06,
+      "loss": 3.3877,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.043004844805643e-06,
+      "loss": 3.4019,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.041004665402743e-06,
+      "loss": 3.3931,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.039004479437456e-06,
+      "loss": 3.3783,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.037004287229894e-06,
+      "loss": 3.4175,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.035004089100165e-06,
+      "loss": 3.4639,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.033003885368382e-06,
+      "loss": 3.4631,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.031003676354655e-06,
+      "loss": 3.3588,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.029003462379102e-06,
+      "loss": 3.4593,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0270032437618345e-06,
+      "loss": 3.354,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.025003020822968e-06,
+      "loss": 3.3963,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.023002793882616e-06,
+      "loss": 3.4215,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.021002563260898e-06,
+      "loss": 3.4619,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.019002329277931e-06,
+      "loss": 3.4292,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.017002092253831e-06,
+      "loss": 3.3329,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.015001852508714e-06,
+      "loss": 3.4275,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.013001610362702e-06,
+      "loss": 3.3404,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.011001366135913e-06,
+      "loss": 3.3824,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.009001120148464e-06,
+      "loss": 3.4653,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.007000872720476e-06,
+      "loss": 3.4554,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.00500062417207e-06,
+      "loss": 3.3098,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.003000374823363e-06,
+      "loss": 3.3879,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.001000124994474e-06,
+      "loss": 3.3933,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.998999875005527e-06,
+      "loss": 3.4975,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.99699962517664e-06,
+      "loss": 3.4334,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.994999375827932e-06,
+      "loss": 3.4581,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.992999127279525e-06,
+      "loss": 3.3868,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9909988798515375e-06,
+      "loss": 3.3883,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.988998633864088e-06,
+      "loss": 3.3526,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.986998389637299e-06,
+      "loss": 3.45,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9849981474912874e-06,
+      "loss": 3.3615,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9829979077461705e-06,
+      "loss": 3.3836,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9809976707220715e-06,
+      "loss": 3.4451,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.978997436739103e-06,
+      "loss": 3.4168,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9769972061173846e-06,
+      "loss": 3.4662,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.974996979177035e-06,
+      "loss": 3.4841,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.972996756238167e-06,
+      "loss": 3.4267,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.970996537620899e-06,
+      "loss": 3.3591,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.968996323645346e-06,
+      "loss": 3.4624,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.96699611463162e-06,
+      "loss": 3.5196,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9649959108998375e-06,
+      "loss": 3.4565,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.962995712770108e-06,
+      "loss": 3.4224,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.960995520562544e-06,
+      "loss": 3.4087,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9589953345972595e-06,
+      "loss": 3.5087,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.956995155194359e-06,
+      "loss": 3.3122,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.954994982673954e-06,
+      "loss": 3.3651,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9529948173561545e-06,
+      "loss": 3.4808,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.950994659561063e-06,
+      "loss": 3.3934,
+      "step": 4451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.948994509608787e-06,
+      "loss": 3.5274,
+      "step": 4452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.946994367819428e-06,
+      "loss": 3.3914,
+      "step": 4453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.944994234513091e-06,
+      "loss": 3.508,
+      "step": 4454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.942994110009878e-06,
+      "loss": 3.4752,
+      "step": 4455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9409939946298865e-06,
+      "loss": 3.4169,
+      "step": 4456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9389938886932155e-06,
+      "loss": 3.3807,
+      "step": 4457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.936993792519965e-06,
+      "loss": 3.4046,
+      "step": 4458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.934993706430226e-06,
+      "loss": 3.4123,
+      "step": 4459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.932993630744096e-06,
+      "loss": 3.4026,
+      "step": 4460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.930993565781663e-06,
+      "loss": 3.3675,
+      "step": 4461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92899351186302e-06,
+      "loss": 3.4558,
+      "step": 4462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9269934693082585e-06,
+      "loss": 3.5077,
+      "step": 4463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924993438437459e-06,
+      "loss": 3.4496,
+      "step": 4464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9229934195707106e-06,
+      "loss": 3.4201,
+      "step": 4465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.920993413028096e-06,
+      "loss": 3.537,
+      "step": 4466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.918993419129695e-06,
+      "loss": 3.3748,
+      "step": 4467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.916993438195587e-06,
+      "loss": 3.4093,
+      "step": 4468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91499347054585e-06,
+      "loss": 3.395,
+      "step": 4469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9129935165005575e-06,
+      "loss": 3.3709,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910993576379783e-06,
+      "loss": 3.5013,
+      "step": 4471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.908993650503597e-06,
+      "loss": 3.4385,
+      "step": 4472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9069937391920665e-06,
+      "loss": 3.3632,
+      "step": 4473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.904993842765258e-06,
+      "loss": 3.3813,
+      "step": 4474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.902993961543234e-06,
+      "loss": 3.4571,
+      "step": 4475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.900994095846059e-06,
+      "loss": 3.4133,
+      "step": 4476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.898994245993787e-06,
+      "loss": 3.4378,
+      "step": 4477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8969944123064755e-06,
+      "loss": 3.4453,
+      "step": 4478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.894994595104181e-06,
+      "loss": 3.3973,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.89299479470695e-06,
+      "loss": 3.3825,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.890995011434833e-06,
+      "loss": 3.456,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.888995245607873e-06,
+      "loss": 3.3957,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8869954975461135e-06,
+      "loss": 3.4351,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.884995767569597e-06,
+      "loss": 3.4077,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.882996055998356e-06,
+      "loss": 3.3258,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.880996363152426e-06,
+      "loss": 3.4051,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.878996689351841e-06,
+      "loss": 3.3904,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.876997034916623e-06,
+      "loss": 3.4466,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.874997400166802e-06,
+      "loss": 3.3553,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.872997785422395e-06,
+      "loss": 3.412,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.870998191003422e-06,
+      "loss": 3.4199,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.868998617229902e-06,
+      "loss": 3.3391,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.86699906442184e-06,
+      "loss": 3.4895,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.864999532899249e-06,
+      "loss": 3.4201,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.863000022982134e-06,
+      "loss": 3.4412,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.861000534990493e-06,
+      "loss": 3.468,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.85900106924433e-06,
+      "loss": 3.4108,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.857001626063633e-06,
+      "loss": 3.458,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.855002205768396e-06,
+      "loss": 3.4641,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.853002808678609e-06,
+      "loss": 3.4305,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.851003435114249e-06,
+      "loss": 3.3931,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8490040853953e-06,
+      "loss": 3.4264,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.847004759841739e-06,
+      "loss": 3.4196,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.845005458773535e-06,
+      "loss": 3.3824,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.843006182510658e-06,
+      "loss": 3.4712,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.841006931373072e-06,
+      "loss": 3.3693,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.839007705680736e-06,
+      "loss": 3.455,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.837008505753607e-06,
+      "loss": 3.3375,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.835009331911638e-06,
+      "loss": 3.4785,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.833010184474774e-06,
+      "loss": 3.429,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8310110637629615e-06,
+      "loss": 3.4593,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.829011970096138e-06,
+      "loss": 3.3893,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8270129037942395e-06,
+      "loss": 3.4063,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.825013865177194e-06,
+      "loss": 3.4007,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.823014854564932e-06,
+      "loss": 3.4651,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.821015872277372e-06,
+      "loss": 3.4222,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819016918634431e-06,
+      "loss": 3.3861,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.817017993956023e-06,
+      "loss": 3.3661,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8150190985620564e-06,
+      "loss": 3.5279,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.813020232772431e-06,
+      "loss": 3.3817,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8110213969070515e-06,
+      "loss": 3.4329,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809022591285805e-06,
+      "loss": 3.4482,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807023816228584e-06,
+      "loss": 3.428,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8050250720552715e-06,
+      "loss": 3.3538,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.803026359085746e-06,
+      "loss": 3.4351,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801027677639884e-06,
+      "loss": 3.4501,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799029028037551e-06,
+      "loss": 3.3625,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.797030410598612e-06,
+      "loss": 3.39,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.795031825642928e-06,
+      "loss": 3.4519,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.793033273490349e-06,
+      "loss": 3.4614,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.791034754460724e-06,
+      "loss": 3.5175,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.789036268873901e-06,
+      "loss": 3.3056,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.78703781704971e-06,
+      "loss": 3.418,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.78503939930799e-06,
+      "loss": 3.3531,
+      "step": 4534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.783041015968563e-06,
+      "loss": 3.3621,
+      "step": 4535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.781042667351251e-06,
+      "loss": 3.4547,
+      "step": 4536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.779044353775873e-06,
+      "loss": 3.4877,
+      "step": 4537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.777046075562235e-06,
+      "loss": 3.3576,
+      "step": 4538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.775047833030143e-06,
+      "loss": 3.4565,
+      "step": 4539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.773049626499398e-06,
+      "loss": 3.3898,
+      "step": 4540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7710514562897895e-06,
+      "loss": 3.4463,
+      "step": 4541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.769053322721108e-06,
+      "loss": 3.4141,
+      "step": 4542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.767055226113129e-06,
+      "loss": 3.463,
+      "step": 4543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.765057166785633e-06,
+      "loss": 3.434,
+      "step": 4544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.763059145058392e-06,
+      "loss": 3.4662,
+      "step": 4545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7610611612511614e-06,
+      "loss": 3.4878,
+      "step": 4546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.759063215683703e-06,
+      "loss": 3.5071,
+      "step": 4547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.757065308675771e-06,
+      "loss": 3.3909,
+      "step": 4548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.755067440547104e-06,
+      "loss": 3.4693,
+      "step": 4549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.753069611617445e-06,
+      "loss": 3.4961,
+      "step": 4550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.751071822206526e-06,
+      "loss": 3.4101,
+      "step": 4551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.749074072634071e-06,
+      "loss": 3.3531,
+      "step": 4552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.747076363219804e-06,
+      "loss": 3.3979,
+      "step": 4553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.745078694283435e-06,
+      "loss": 3.3295,
+      "step": 4554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.743081066144671e-06,
+      "loss": 3.5403,
+      "step": 4555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.741083479123214e-06,
+      "loss": 3.4818,
+      "step": 4556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.739085933538758e-06,
+      "loss": 3.4285,
+      "step": 4557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7370884297109895e-06,
+      "loss": 3.4804,
+      "step": 4558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.735090967959587e-06,
+      "loss": 3.4856,
+      "step": 4559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.733093548604228e-06,
+      "loss": 3.4507,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.731096171964577e-06,
+      "loss": 3.355,
+      "step": 4561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729098838360294e-06,
+      "loss": 3.5056,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727101548111034e-06,
+      "loss": 3.4307,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.725104301536443e-06,
+      "loss": 3.3609,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723107098956159e-06,
+      "loss": 3.4012,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7211099406898165e-06,
+      "loss": 3.3077,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.719112827057038e-06,
+      "loss": 3.3829,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.717115758377441e-06,
+      "loss": 3.5303,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7151187349706425e-06,
+      "loss": 3.5484,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7131217571562385e-06,
+      "loss": 3.4586,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.711124825253829e-06,
+      "loss": 3.3756,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7091279395830045e-06,
+      "loss": 3.4611,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.707131100463343e-06,
+      "loss": 3.3615,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.705134308214423e-06,
+      "loss": 3.3549,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.703137563155805e-06,
+      "loss": 3.2957,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7011408656070525e-06,
+      "loss": 3.4313,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.699144215887719e-06,
+      "loss": 3.453,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.697147614317342e-06,
+      "loss": 3.5707,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.695151061215461e-06,
+      "loss": 3.4309,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.693154556901607e-06,
+      "loss": 3.4387,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.691158101695296e-06,
+      "loss": 3.4281,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.689161695916045e-06,
+      "loss": 3.4574,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.687165339883354e-06,
+      "loss": 3.4277,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.685169033916722e-06,
+      "loss": 3.3739,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.68317277833564e-06,
+      "loss": 3.4005,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.681176573459585e-06,
+      "loss": 3.4325,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.679180419608032e-06,
+      "loss": 3.4036,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.677184317100446e-06,
+      "loss": 3.5177,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.67518826625628e-06,
+      "loss": 3.4883,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.673192267394987e-06,
+      "loss": 3.3129,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.671196320836001e-06,
+      "loss": 3.3933,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.669200426898756e-06,
+      "loss": 3.4986,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.667204585902678e-06,
+      "loss": 3.3749,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.665208798167175e-06,
+      "loss": 3.4024,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.663213064011658e-06,
+      "loss": 3.4539,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.661217383755522e-06,
+      "loss": 3.3329,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6592217577181554e-06,
+      "loss": 3.5052,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.65722618621894e-06,
+      "loss": 3.3777,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655230669577245e-06,
+      "loss": 3.4813,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.653235208112435e-06,
+      "loss": 3.4298,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.651239802143862e-06,
+      "loss": 3.4437,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.649244451990871e-06,
+      "loss": 3.4994,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6472491579728e-06,
+      "loss": 3.486,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.645253920408971e-06,
+      "loss": 3.3877,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.643258739618706e-06,
+      "loss": 3.4274,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.641263615921315e-06,
+      "loss": 3.4599,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.639268549636092e-06,
+      "loss": 3.4821,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.637273541082331e-06,
+      "loss": 3.2714,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.635278590579315e-06,
+      "loss": 3.5013,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.633283698446311e-06,
+      "loss": 3.541,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.631288865002587e-06,
+      "loss": 3.4395,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.62929409056739e-06,
+      "loss": 3.3064,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.627299375459968e-06,
+      "loss": 3.4569,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.625304719999557e-06,
+      "loss": 3.3893,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.623310124505376e-06,
+      "loss": 3.391,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.621315589296643e-06,
+      "loss": 3.4662,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.619321114692565e-06,
+      "loss": 3.4006,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6173267010123345e-06,
+      "loss": 3.3912,
+      "step": 4618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.615332348575141e-06,
+      "loss": 3.4977,
+      "step": 4619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.613338057700156e-06,
+      "loss": 3.3374,
+      "step": 4620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6113438287065495e-06,
+      "loss": 3.4178,
+      "step": 4621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.609349661913479e-06,
+      "loss": 3.2924,
+      "step": 4622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6073555576400865e-06,
+      "loss": 3.3928,
+      "step": 4623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.605361516205511e-06,
+      "loss": 3.4307,
+      "step": 4624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.60336753792888e-06,
+      "loss": 3.4099,
+      "step": 4625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6013736231293074e-06,
+      "loss": 3.4344,
+      "step": 4626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.599379772125902e-06,
+      "loss": 3.4114,
+      "step": 4627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597385985237757e-06,
+      "loss": 3.4118,
+      "step": 4628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595392262783958e-06,
+      "loss": 3.4574,
+      "step": 4629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593398605083584e-06,
+      "loss": 3.3931,
+      "step": 4630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.591405012455697e-06,
+      "loss": 3.4251,
+      "step": 4631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.58941148521935e-06,
+      "loss": 3.3385,
+      "step": 4632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.587418023693591e-06,
+      "loss": 3.4381,
+      "step": 4633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.585424628197452e-06,
+      "loss": 3.4767,
+      "step": 4634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.583431299049956e-06,
+      "loss": 3.32,
+      "step": 4635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5814380365701135e-06,
+      "loss": 3.3471,
+      "step": 4636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5794448410769285e-06,
+      "loss": 3.3835,
+      "step": 4637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.577451712889392e-06,
+      "loss": 3.3481,
+      "step": 4638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.575458652326485e-06,
+      "loss": 3.5097,
+      "step": 4639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.573465659707172e-06,
+      "loss": 3.4494,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.571472735350418e-06,
+      "loss": 3.5277,
+      "step": 4641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.569479879575168e-06,
+      "loss": 3.441,
+      "step": 4642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.567487092700357e-06,
+      "loss": 3.3337,
+      "step": 4643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.565494375044912e-06,
+      "loss": 3.3581,
+      "step": 4644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5635017269277485e-06,
+      "loss": 3.3991,
+      "step": 4645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.561509148667768e-06,
+      "loss": 3.3865,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5595166405838636e-06,
+      "loss": 3.4591,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.557524202994917e-06,
+      "loss": 3.348,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.555531836219798e-06,
+      "loss": 3.4451,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.553539540577362e-06,
+      "loss": 3.4887,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.551547316386461e-06,
+      "loss": 3.3429,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5495551639659256e-06,
+      "loss": 3.4061,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.547563083634582e-06,
+      "loss": 3.4179,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5455710757112455e-06,
+      "loss": 3.5102,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5435791405147116e-06,
+      "loss": 3.4107,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.541587278363775e-06,
+      "loss": 3.3435,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.539595489577208e-06,
+      "loss": 3.4695,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.537603774473779e-06,
+      "loss": 3.4741,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.535612133372246e-06,
+      "loss": 3.391,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.533620566591344e-06,
+      "loss": 3.5329,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.531629074449807e-06,
+      "loss": 3.399,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.529637657266356e-06,
+      "loss": 3.4486,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.527646315359692e-06,
+      "loss": 3.4408,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.525655049048514e-06,
+      "loss": 3.3503,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5236638586515e-06,
+      "loss": 3.4704,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.521672744487323e-06,
+      "loss": 3.4328,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5196817068746424e-06,
+      "loss": 3.4447,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5176907461321e-06,
+      "loss": 3.4343,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.51569986257833e-06,
+      "loss": 3.4,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.513709056531957e-06,
+      "loss": 3.357,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.511718328311585e-06,
+      "loss": 3.5133,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.509727678235814e-06,
+      "loss": 3.4221,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.507737106623224e-06,
+      "loss": 3.4699,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.505746613792387e-06,
+      "loss": 3.3692,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.503756200061865e-06,
+      "loss": 3.466,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.501765865750199e-06,
+      "loss": 3.4636,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.499775611175924e-06,
+      "loss": 3.4688,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.497785436657562e-06,
+      "loss": 3.3522,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.495795342513618e-06,
+      "loss": 3.397,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.493805329062589e-06,
+      "loss": 3.328,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.491815396622955e-06,
+      "loss": 3.3496,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.489825545513184e-06,
+      "loss": 3.3266,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.487835776051734e-06,
+      "loss": 3.4839,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.485846088557048e-06,
+      "loss": 3.3658,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.483856483347551e-06,
+      "loss": 3.3399,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.481866960741666e-06,
+      "loss": 3.4354,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.479877521057792e-06,
+      "loss": 3.427,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.477888164614321e-06,
+      "loss": 3.327,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.475898891729627e-06,
+      "loss": 3.4392,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.473909702722074e-06,
+      "loss": 3.4619,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.471920597910015e-06,
+      "loss": 3.4089,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.469931577611781e-06,
+      "loss": 3.4752,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.467942642145699e-06,
+      "loss": 3.4455,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.465953791830079e-06,
+      "loss": 3.5028,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.463965026983212e-06,
+      "loss": 3.4126,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.461976347923385e-06,
+      "loss": 3.4236,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.459987754968861e-06,
+      "loss": 3.4803,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.457999248437897e-06,
+      "loss": 3.4138,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456010828648736e-06,
+      "loss": 3.4897,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.454022495919601e-06,
+      "loss": 3.3091,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452034250568705e-06,
+      "loss": 3.3688,
+      "step": 4701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.450046092914251e-06,
+      "loss": 3.4317,
+      "step": 4702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448058023274417e-06,
+      "loss": 3.4212,
+      "step": 4703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446070041967381e-06,
+      "loss": 3.3182,
+      "step": 4704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444082149311294e-06,
+      "loss": 3.3952,
+      "step": 4705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.442094345624299e-06,
+      "loss": 3.4541,
+      "step": 4706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.440106631224528e-06,
+      "loss": 3.3545,
+      "step": 4707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.43811900643009e-06,
+      "loss": 3.4662,
+      "step": 4708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.436131471559087e-06,
+      "loss": 3.4733,
+      "step": 4709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.434144026929606e-06,
+      "loss": 3.4856,
+      "step": 4710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.432156672859712e-06,
+      "loss": 3.4782,
+      "step": 4711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.430169409667468e-06,
+      "loss": 3.4282,
+      "step": 4712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.428182237670909e-06,
+      "loss": 3.3912,
+      "step": 4713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.426195157188065e-06,
+      "loss": 3.4167,
+      "step": 4714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.424208168536951e-06,
+      "loss": 3.4506,
+      "step": 4715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.42222127203556e-06,
+      "loss": 3.3783,
+      "step": 4716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.420234468001877e-06,
+      "loss": 3.4776,
+      "step": 4717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.418247756753873e-06,
+      "loss": 3.4114,
+      "step": 4718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.416261138609496e-06,
+      "loss": 3.3611,
+      "step": 4719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.414274613886688e-06,
+      "loss": 3.4338,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.41228818290337e-06,
+      "loss": 3.4325,
+      "step": 4721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.410301845977451e-06,
+      "loss": 3.4388,
+      "step": 4722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.408315603426826e-06,
+      "loss": 3.3908,
+      "step": 4723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.406329455569373e-06,
+      "loss": 3.4044,
+      "step": 4724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4043434027229545e-06,
+      "loss": 3.3043,
+      "step": 4725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.402357445205417e-06,
+      "loss": 3.4108,
+      "step": 4726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.400371583334594e-06,
+      "loss": 3.4031,
+      "step": 4727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.398385817428305e-06,
+      "loss": 3.4474,
+      "step": 4728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.396400147804348e-06,
+      "loss": 3.3927,
+      "step": 4729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.394414574780514e-06,
+      "loss": 3.5145,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.392429098674572e-06,
+      "loss": 3.4358,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.390443719804275e-06,
+      "loss": 3.4436,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.388458438487368e-06,
+      "loss": 3.488,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.386473255041572e-06,
+      "loss": 3.4118,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.384488169784596e-06,
+      "loss": 3.3829,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.382503183034136e-06,
+      "loss": 3.3654,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.380518295107864e-06,
+      "loss": 3.4163,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.378533506323446e-06,
+      "loss": 3.4585,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.376548816998527e-06,
+      "loss": 3.3367,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.374564227450734e-06,
+      "loss": 3.4776,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.372579737997685e-06,
+      "loss": 3.3655,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.370595348956973e-06,
+      "loss": 3.3849,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.368611060646181e-06,
+      "loss": 3.3733,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.366626873382879e-06,
+      "loss": 3.4249,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3646427874846096e-06,
+      "loss": 3.4419,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.362658803268909e-06,
+      "loss": 3.34,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.360674921053296e-06,
+      "loss": 3.4472,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3586911411552665e-06,
+      "loss": 3.3958,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.35670746389231e-06,
+      "loss": 3.3155,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.354723889581889e-06,
+      "loss": 3.5447,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.352740418541458e-06,
+      "loss": 3.4739,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.350757051088452e-06,
+      "loss": 3.375,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.348773787540286e-06,
+      "loss": 3.3992,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3467906282143645e-06,
+      "loss": 3.4457,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.344807573428072e-06,
+      "loss": 3.4147,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.342824623498776e-06,
+      "loss": 3.4474,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.340841778743828e-06,
+      "loss": 3.359,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.33885903948056e-06,
+      "loss": 3.3449,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.336876406026293e-06,
+      "loss": 3.3488,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.334893878698328e-06,
+      "loss": 3.412,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3329114578139445e-06,
+      "loss": 3.4746,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.330929143690412e-06,
+      "loss": 3.383,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.328946936644983e-06,
+      "loss": 3.4734,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3269648369948836e-06,
+      "loss": 3.4069,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3249828450573336e-06,
+      "loss": 3.4027,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.32300096114953e-06,
+      "loss": 3.3573,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.321019185588651e-06,
+      "loss": 3.4076,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.319037518691863e-06,
+      "loss": 3.4522,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317055960776312e-06,
+      "loss": 3.4175,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.315074512159124e-06,
+      "loss": 3.4955,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3130931731574126e-06,
+      "loss": 3.3286,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.31111194408827e-06,
+      "loss": 3.3808,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.309130825268773e-06,
+      "loss": 3.5057,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.307149817015977e-06,
+      "loss": 3.465,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.305168919646928e-06,
+      "loss": 3.4896,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.303188133478644e-06,
+      "loss": 3.4564,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.301207458828131e-06,
+      "loss": 3.4404,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.299226896012377e-06,
+      "loss": 3.3477,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.297246445348352e-06,
+      "loss": 3.4354,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.295266107153006e-06,
+      "loss": 3.4641,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.293285881743274e-06,
+      "loss": 3.4255,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.291305769436068e-06,
+      "loss": 3.3577,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.289325770548287e-06,
+      "loss": 3.4145,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.287345885396812e-06,
+      "loss": 3.4322,
+      "step": 4784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2853661142985005e-06,
+      "loss": 3.4507,
+      "step": 4785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.283386457570199e-06,
+      "loss": 3.4377,
+      "step": 4786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.281406915528726e-06,
+      "loss": 3.4256,
+      "step": 4787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.279427488490892e-06,
+      "loss": 3.4399,
+      "step": 4788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.277448176773484e-06,
+      "loss": 3.4288,
+      "step": 4789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27546898069327e-06,
+      "loss": 3.3903,
+      "step": 4790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.273489900567e-06,
+      "loss": 3.396,
+      "step": 4791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.271510936711409e-06,
+      "loss": 3.4554,
+      "step": 4792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.269532089443206e-06,
+      "loss": 3.4896,
+      "step": 4793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2675533590790895e-06,
+      "loss": 3.4332,
+      "step": 4794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.265574745935733e-06,
+      "loss": 3.5434,
+      "step": 4795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.263596250329794e-06,
+      "loss": 3.4423,
+      "step": 4796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.261617872577914e-06,
+      "loss": 3.4809,
+      "step": 4797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.259639612996708e-06,
+      "loss": 3.4191,
+      "step": 4798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.257661471902778e-06,
+      "loss": 3.445,
+      "step": 4799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.25568344961271e-06,
+      "loss": 3.4037,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.253705546443061e-06,
+      "loss": 3.4565,
+      "step": 4801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.251727762710379e-06,
+      "loss": 3.4413,
+      "step": 4802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.249750098731184e-06,
+      "loss": 3.4038,
+      "step": 4803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.247772554821983e-06,
+      "loss": 3.5171,
+      "step": 4804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.245795131299266e-06,
+      "loss": 3.3538,
+      "step": 4805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.243817828479494e-06,
+      "loss": 3.397,
+      "step": 4806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.241840646679118e-06,
+      "loss": 3.3647,
+      "step": 4807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2398635862145675e-06,
+      "loss": 3.3603,
+      "step": 4808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.237886647402247e-06,
+      "loss": 3.3889,
+      "step": 4809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.235909830558549e-06,
+      "loss": 3.4987,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.233933135999842e-06,
+      "loss": 3.3829,
+      "step": 4811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.231956564042475e-06,
+      "loss": 3.4567,
+      "step": 4812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.229980115002782e-06,
+      "loss": 3.4483,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.228003789197071e-06,
+      "loss": 3.3642,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2260275869416344e-06,
+      "loss": 3.3367,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.224051508552743e-06,
+      "loss": 3.3598,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.222075554346649e-06,
+      "loss": 3.4787,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.220099724639585e-06,
+      "loss": 3.3055,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.21812401974776e-06,
+      "loss": 3.4358,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.216148439987368e-06,
+      "loss": 3.4609,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.214172985674581e-06,
+      "loss": 3.4285,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.212197657125549e-06,
+      "loss": 3.3426,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.210222454656404e-06,
+      "loss": 3.3868,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2082473785832605e-06,
+      "loss": 3.3851,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.206272429222204e-06,
+      "loss": 3.4639,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2042976068893115e-06,
+      "loss": 3.3573,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.202322911900628e-06,
+      "loss": 3.4448,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.200348344572186e-06,
+      "loss": 3.441,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.198373905219999e-06,
+      "loss": 3.3993,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.19639959416005e-06,
+      "loss": 3.3911,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.19442541170831e-06,
+      "loss": 3.4962,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.192451358180731e-06,
+      "loss": 3.4257,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.190477433893233e-06,
+      "loss": 3.3969,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.188503639161731e-06,
+      "loss": 3.4914,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.186529974302105e-06,
+      "loss": 3.4125,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.184556439630221e-06,
+      "loss": 3.4834,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1825830354619285e-06,
+      "loss": 3.4511,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.180609762113045e-06,
+      "loss": 3.3376,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.178636619899376e-06,
+      "loss": 3.3766,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.176663609136705e-06,
+      "loss": 3.4568,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.174690730140788e-06,
+      "loss": 3.4818,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.17271798322737e-06,
+      "loss": 3.3929,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.170745368712164e-06,
+      "loss": 3.4085,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16877288691087e-06,
+      "loss": 3.3632,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.166800538139166e-06,
+      "loss": 3.4022,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.164828322712702e-06,
+      "loss": 3.4007,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.162856240947114e-06,
+      "loss": 3.4086,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1608842931580165e-06,
+      "loss": 3.3919,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.158912479660995e-06,
+      "loss": 3.4338,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.156940800771621e-06,
+      "loss": 3.5204,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.154969256805442e-06,
+      "loss": 3.3975,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1529978480779835e-06,
+      "loss": 3.3438,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.151026574904751e-06,
+      "loss": 3.3909,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.149055437601226e-06,
+      "loss": 3.3882,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1470844364828684e-06,
+      "loss": 3.3907,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1451135718651185e-06,
+      "loss": 3.4826,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.143142844063394e-06,
+      "loss": 3.4577,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.14117225339309e-06,
+      "loss": 3.4168,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.139201800169577e-06,
+      "loss": 3.3713,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.137231484708209e-06,
+      "loss": 3.3984,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.135261307324316e-06,
+      "loss": 3.4338,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.133291268333202e-06,
+      "loss": 3.3804,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.131321368050157e-06,
+      "loss": 3.4391,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.129351606790439e-06,
+      "loss": 3.4231,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.12738198486929e-06,
+      "loss": 3.3806,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.125412502601931e-06,
+      "loss": 3.406,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.123443160303553e-06,
+      "loss": 3.3456,
+      "step": 4867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.121473958289333e-06,
+      "loss": 3.3683,
+      "step": 4868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1195048968744235e-06,
+      "loss": 3.3738,
+      "step": 4869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.117535976373948e-06,
+      "loss": 3.3438,
+      "step": 4870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.115567197103019e-06,
+      "loss": 3.3856,
+      "step": 4871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.113598559376713e-06,
+      "loss": 3.4072,
+      "step": 4872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.111630063510095e-06,
+      "loss": 3.3911,
+      "step": 4873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.109661709818203e-06,
+      "loss": 3.4231,
+      "step": 4874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.107693498616051e-06,
+      "loss": 3.4618,
+      "step": 4875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.105725430218632e-06,
+      "loss": 3.3671,
+      "step": 4876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.103757504940918e-06,
+      "loss": 3.3464,
+      "step": 4877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.101789723097851e-06,
+      "loss": 3.4058,
+      "step": 4878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.099822085004359e-06,
+      "loss": 3.3875,
+      "step": 4879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.09785459097534e-06,
+      "loss": 3.4857,
+      "step": 4880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.095887241325672e-06,
+      "loss": 3.4277,
+      "step": 4881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.093920036370214e-06,
+      "loss": 3.379,
+      "step": 4882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0919529764237905e-06,
+      "loss": 3.4134,
+      "step": 4883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.089986061801214e-06,
+      "loss": 3.4546,
+      "step": 4884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.088019292817271e-06,
+      "loss": 3.4818,
+      "step": 4885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.086052669786719e-06,
+      "loss": 3.4997,
+      "step": 4886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.084086193024301e-06,
+      "loss": 3.3953,
+      "step": 4887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.082119862844726e-06,
+      "loss": 3.3669,
+      "step": 4888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.08015367956269e-06,
+      "loss": 3.4095,
+      "step": 4889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.078187643492862e-06,
+      "loss": 3.3977,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0762217549498805e-06,
+      "loss": 3.3569,
+      "step": 4891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.074256014248371e-06,
+      "loss": 3.3664,
+      "step": 4892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.072290421702931e-06,
+      "loss": 3.4662,
+      "step": 4893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.070324977628132e-06,
+      "loss": 3.4288,
+      "step": 4894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.068359682338524e-06,
+      "loss": 3.3363,
+      "step": 4895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.066394536148634e-06,
+      "loss": 3.3875,
+      "step": 4896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.064429539372961e-06,
+      "loss": 3.4653,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.062464692325987e-06,
+      "loss": 3.4138,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.060499995322164e-06,
+      "loss": 3.4106,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.058535448675923e-06,
+      "loss": 3.4088,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.056571052701669e-06,
+      "loss": 3.436,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.054606807713784e-06,
+      "loss": 3.5796,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.052642714026627e-06,
+      "loss": 3.3506,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.050678771954529e-06,
+      "loss": 3.3697,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.048714981811803e-06,
+      "loss": 3.3325,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0467513439127315e-06,
+      "loss": 3.4692,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.044787858571575e-06,
+      "loss": 3.4382,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.04282452610257e-06,
+      "loss": 3.369,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.04086134681993e-06,
+      "loss": 3.4435,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.038898321037839e-06,
+      "loss": 3.45,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036935449070464e-06,
+      "loss": 3.4368,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0349727312319386e-06,
+      "loss": 3.355,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033010167836378e-06,
+      "loss": 3.2986,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031047759197872e-06,
+      "loss": 3.3486,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029085505630481e-06,
+      "loss": 3.5228,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02712340744825e-06,
+      "loss": 3.3983,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025161464965186e-06,
+      "loss": 3.4684,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023199678495283e-06,
+      "loss": 3.3639,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021238048352505e-06,
+      "loss": 3.417,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019276574850789e-06,
+      "loss": 3.4536,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0173152583040515e-06,
+      "loss": 3.3598,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015354099026182e-06,
+      "loss": 3.4577,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013393097331041e-06,
+      "loss": 3.3479,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011432253532472e-06,
+      "loss": 3.4685,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0094715679442835e-06,
+      "loss": 3.4252,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.007511040880267e-06,
+      "loss": 3.464,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.005550672654185e-06,
+      "loss": 3.4519,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.003590463579772e-06,
+      "loss": 3.4938,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.001630413970742e-06,
+      "loss": 3.4333,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.999670524140784e-06,
+      "loss": 3.4726,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9977107944035534e-06,
+      "loss": 3.4351,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.99575122507269e-06,
+      "loss": 3.4583,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9937918164618e-06,
+      "loss": 3.3779,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9918325688844684e-06,
+      "loss": 3.4745,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989873482654255e-06,
+      "loss": 3.3177,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987914558084688e-06,
+      "loss": 3.3213,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.985955795489275e-06,
+      "loss": 3.4054,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.983997195181499e-06,
+      "loss": 3.3356,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.982038757474813e-06,
+      "loss": 3.4058,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.980080482682645e-06,
+      "loss": 3.4051,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.978122371118396e-06,
+      "loss": 3.4618,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.976164423095444e-06,
+      "loss": 3.4268,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.974206638927139e-06,
+      "loss": 3.421,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9722490189268035e-06,
+      "loss": 3.3757,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.970291563407736e-06,
+      "loss": 3.4656,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.968334272683209e-06,
+      "loss": 3.4278,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.966377147066463e-06,
+      "loss": 3.3748,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.964420186870722e-06,
+      "loss": 3.3923,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.962463392409173e-06,
+      "loss": 3.358,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.960506763994983e-06,
+      "loss": 3.4375,
+      "step": 4950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.958550301941294e-06,
+      "loss": 3.3423,
+      "step": 4951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.956594006561213e-06,
+      "loss": 3.3225,
+      "step": 4952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9546378781678275e-06,
+      "loss": 3.4092,
+      "step": 4953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.952681917074199e-06,
+      "loss": 3.5328,
+      "step": 4954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.950726123593355e-06,
+      "loss": 3.368,
+      "step": 4955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.948770498038306e-06,
+      "loss": 3.3611,
+      "step": 4956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.946815040722025e-06,
+      "loss": 3.3384,
+      "step": 4957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.944859751957465e-06,
+      "loss": 3.3769,
+      "step": 4958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.942904632057553e-06,
+      "loss": 3.4564,
+      "step": 4959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.940949681335183e-06,
+      "loss": 3.5006,
+      "step": 4960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.938994900103227e-06,
+      "loss": 3.4281,
+      "step": 4961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.937040288674529e-06,
+      "loss": 3.3906,
+      "step": 4962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.935085847361902e-06,
+      "loss": 3.5009,
+      "step": 4963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.93313157647814e-06,
+      "loss": 3.3637,
+      "step": 4964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.931177476335998e-06,
+      "loss": 3.5026,
+      "step": 4965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.929223547248214e-06,
+      "loss": 3.3111,
+      "step": 4966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.927269789527497e-06,
+      "loss": 3.249,
+      "step": 4967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.925316203486521e-06,
+      "loss": 3.3084,
+      "step": 4968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.92336278943794e-06,
+      "loss": 3.3807,
+      "step": 4969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.921409547694382e-06,
+      "loss": 3.3974,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.919456478568439e-06,
+      "loss": 3.4723,
+      "step": 4971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9175035823726825e-06,
+      "loss": 3.3933,
+      "step": 4972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.915550859419652e-06,
+      "loss": 3.4342,
+      "step": 4973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.913598310021863e-06,
+      "loss": 3.3155,
+      "step": 4974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.911645934491803e-06,
+      "loss": 3.3404,
+      "step": 4975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.909693733141927e-06,
+      "loss": 3.3315,
+      "step": 4976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.907741706284667e-06,
+      "loss": 3.4248,
+      "step": 4977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9057898542324265e-06,
+      "loss": 3.4433,
+      "step": 4978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.903838177297577e-06,
+      "loss": 3.4193,
+      "step": 4979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.901886675792467e-06,
+      "loss": 3.3894,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8999353500294135e-06,
+      "loss": 3.4235,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.897984200320707e-06,
+      "loss": 3.4373,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.896033226978611e-06,
+      "loss": 3.4324,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8940824303153585e-06,
+      "loss": 3.376,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.892131810643154e-06,
+      "loss": 3.4693,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.890181368274175e-06,
+      "loss": 3.46,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.888231103520571e-06,
+      "loss": 3.4583,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.886281016694463e-06,
+      "loss": 3.3331,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.884331108107939e-06,
+      "loss": 3.3719,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.882381378073068e-06,
+      "loss": 3.3957,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880431826901881e-06,
+      "loss": 3.4637,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.878482454906386e-06,
+      "loss": 3.3394,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.876533262398559e-06,
+      "loss": 3.518,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.87458424969035e-06,
+      "loss": 3.5307,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.872635417093678e-06,
+      "loss": 3.3533,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.870686764920436e-06,
+      "loss": 3.3346,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.868738293482483e-06,
+      "loss": 3.4706,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.866790003091654e-06,
+      "loss": 3.4303,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.864841894059757e-06,
+      "loss": 3.4103,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.862893966698561e-06,
+      "loss": 3.5004,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.860946221319818e-06,
+      "loss": 3.4257,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.85899865823524e-06,
+      "loss": 3.3662,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.857051277756517e-06,
+      "loss": 3.3715,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.855104080195312e-06,
+      "loss": 3.41,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.853157065863248e-06,
+      "loss": 3.4475,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.851210235071928e-06,
+      "loss": 3.4081,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.849263588132925e-06,
+      "loss": 3.3999,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.847317125357776e-06,
+      "loss": 3.4501,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8453708470579975e-06,
+      "loss": 3.4037,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.843424753545067e-06,
+      "loss": 3.4873,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.84147884513044e-06,
+      "loss": 3.3553,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.839533122125542e-06,
+      "loss": 3.452,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.837587584841761e-06,
+      "loss": 3.5397,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8356422335904645e-06,
+      "loss": 3.4447,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8336970686829874e-06,
+      "loss": 3.3707,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.831752090430631e-06,
+      "loss": 3.4631,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.829807299144672e-06,
+      "loss": 3.3387,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.827862695136352e-06,
+      "loss": 3.3242,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.825918278716886e-06,
+      "loss": 3.5218,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8239740501974635e-06,
+      "loss": 3.3865,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.822030009889231e-06,
+      "loss": 3.4684,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.820086158103319e-06,
+      "loss": 3.3884,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.818142495150819e-06,
+      "loss": 3.3424,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.816199021342793e-06,
+      "loss": 3.3958,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.814255736990279e-06,
+      "loss": 3.4174,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8123126424042768e-06,
+      "loss": 3.3903,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8103697378957587e-06,
+      "loss": 3.4305,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8084270237756705e-06,
+      "loss": 3.4296,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.806484500354922e-06,
+      "loss": 3.3266,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.804542167944394e-06,
+      "loss": 3.432,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.802600026854939e-06,
+      "loss": 3.4348,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8006580773973755e-06,
+      "loss": 3.4236,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7987163198824967e-06,
+      "loss": 3.3808,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7967747546210563e-06,
+      "loss": 3.4239,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.794833381923786e-06,
+      "loss": 3.3601,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7928922021013835e-06,
+      "loss": 3.4292,
+      "step": 5036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7909512154645116e-06,
+      "loss": 3.4224,
+      "step": 5037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7890104223238084e-06,
+      "loss": 3.4642,
+      "step": 5038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7870698229898805e-06,
+      "loss": 3.3923,
+      "step": 5039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.785129417773296e-06,
+      "loss": 3.3463,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.783189206984603e-06,
+      "loss": 3.397,
+      "step": 5041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7812491909343085e-06,
+      "loss": 3.4351,
+      "step": 5042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.779309369932894e-06,
+      "loss": 3.3151,
+      "step": 5043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.777369744290811e-06,
+      "loss": 3.4856,
+      "step": 5044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.775430314318473e-06,
+      "loss": 3.392,
+      "step": 5045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7734910803262705e-06,
+      "loss": 3.4366,
+      "step": 5046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7715520426245537e-06,
+      "loss": 3.4385,
+      "step": 5047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7696132015236495e-06,
+      "loss": 3.3832,
+      "step": 5048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7676745573338516e-06,
+      "loss": 3.4517,
+      "step": 5049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7657361103654156e-06,
+      "loss": 3.4378,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.763797860928574e-06,
+      "loss": 3.3924,
+      "step": 5051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7618598093335257e-06,
+      "loss": 3.5564,
+      "step": 5052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7599219558904323e-06,
+      "loss": 3.4074,
+      "step": 5053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.757984300909432e-06,
+      "loss": 3.3106,
+      "step": 5054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.756046844700623e-06,
+      "loss": 3.5244,
+      "step": 5055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7541095875740778e-06,
+      "loss": 3.4203,
+      "step": 5056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7521725298398374e-06,
+      "loss": 3.3634,
+      "step": 5057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7502356718079047e-06,
+      "loss": 3.3879,
+      "step": 5058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7482990137882547e-06,
+      "loss": 3.4082,
+      "step": 5059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7463625560908345e-06,
+      "loss": 3.4767,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.74442629902555e-06,
+      "loss": 3.4135,
+      "step": 5061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7424902429022812e-06,
+      "loss": 3.3389,
+      "step": 5062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.740554388030876e-06,
+      "loss": 3.4443,
+      "step": 5063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7386187347211457e-06,
+      "loss": 3.4485,
+      "step": 5064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7366832832828757e-06,
+      "loss": 3.389,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7347480340258137e-06,
+      "loss": 3.3567,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7328129872596764e-06,
+      "loss": 3.4153,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7308781432941497e-06,
+      "loss": 3.449,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7289435024388863e-06,
+      "loss": 3.3858,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.727009065003506e-06,
+      "loss": 3.4421,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.725074831297595e-06,
+      "loss": 3.402,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.72314080163071e-06,
+      "loss": 3.3791,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7212069763123713e-06,
+      "loss": 3.3858,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.719273355652069e-06,
+      "loss": 3.3956,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7173399399592603e-06,
+      "loss": 3.3956,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.715406729543368e-06,
+      "loss": 3.3746,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.713473724713783e-06,
+      "loss": 3.4955,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7115409257798667e-06,
+      "loss": 3.3674,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.709608333050939e-06,
+      "loss": 3.4264,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.707675946836295e-06,
+      "loss": 3.3602,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.705743767445195e-06,
+      "loss": 3.4025,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.703811795186861e-06,
+      "loss": 3.3751,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7018800303704895e-06,
+      "loss": 3.4279,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6999484733052405e-06,
+      "loss": 3.3101,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.698017124300236e-06,
+      "loss": 3.3096,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6960859836645746e-06,
+      "loss": 3.4025,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.694155051707311e-06,
+      "loss": 3.3746,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6922243287374736e-06,
+      "loss": 3.3817,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6902938150640583e-06,
+      "loss": 3.4346,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6883635109960194e-06,
+      "loss": 3.4483,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.686433416842285e-06,
+      "loss": 3.4572,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6845035329117494e-06,
+      "loss": 3.3695,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6825738595132675e-06,
+      "loss": 3.4119,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.680644396955669e-06,
+      "loss": 3.376,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.678715145547739e-06,
+      "loss": 3.408,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.676786105598239e-06,
+      "loss": 3.5172,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.674857277415893e-06,
+      "loss": 3.4649,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6729286613093874e-06,
+      "loss": 3.3935,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6710002575873793e-06,
+      "loss": 3.3554,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.669072066558494e-06,
+      "loss": 3.4061,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6671440885313135e-06,
+      "loss": 3.3858,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.665216323814396e-06,
+      "loss": 3.431,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6632887727162575e-06,
+      "loss": 3.3541,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.661361435545384e-06,
+      "loss": 3.5195,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6594343126102305e-06,
+      "loss": 3.4545,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6575074042192078e-06,
+      "loss": 3.3884,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6555807106807006e-06,
+      "loss": 3.4547,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6536542323030606e-06,
+      "loss": 3.4744,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6517279693945956e-06,
+      "loss": 3.4458,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6498019222635884e-06,
+      "loss": 3.4135,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.647876091218283e-06,
+      "loss": 3.474,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6459504765668872e-06,
+      "loss": 3.4522,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6440250786175796e-06,
+      "loss": 3.3673,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6420998976784994e-06,
+      "loss": 3.4383,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6401749340577526e-06,
+      "loss": 3.5302,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.638250188063409e-06,
+      "loss": 3.4663,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.636325660003508e-06,
+      "loss": 3.4817,
+      "step": 5117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6344013501860486e-06,
+      "loss": 3.4395,
+      "step": 5118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.632477258918997e-06,
+      "loss": 3.3827,
+      "step": 5119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6305533865102875e-06,
+      "loss": 3.3607,
+      "step": 5120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.628629733267815e-06,
+      "loss": 3.3554,
+      "step": 5121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6267062994994394e-06,
+      "loss": 3.5051,
+      "step": 5122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.624783085512992e-06,
+      "loss": 3.4466,
+      "step": 5123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.622860091616257e-06,
+      "loss": 3.4422,
+      "step": 5124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6209373181169943e-06,
+      "loss": 3.4974,
+      "step": 5125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6190147653229267e-06,
+      "loss": 3.5308,
+      "step": 5126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6170924335417344e-06,
+      "loss": 3.5309,
+      "step": 5127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.615170323081071e-06,
+      "loss": 3.3592,
+      "step": 5128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6132484342485518e-06,
+      "loss": 3.3932,
+      "step": 5129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6113267673517515e-06,
+      "loss": 3.4157,
+      "step": 5130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6094053226982196e-06,
+      "loss": 3.388,
+      "step": 5131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6074841005954575e-06,
+      "loss": 3.4807,
+      "step": 5132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.605563101350942e-06,
+      "loss": 3.3921,
+      "step": 5133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6036423252721104e-06,
+      "loss": 3.3969,
+      "step": 5134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.60172177266636e-06,
+      "loss": 3.4827,
+      "step": 5135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5998014438410576e-06,
+      "loss": 3.3589,
+      "step": 5136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5978813391035363e-06,
+      "loss": 3.4204,
+      "step": 5137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.595961458761084e-06,
+      "loss": 3.4549,
+      "step": 5138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.594041803120963e-06,
+      "loss": 3.4296,
+      "step": 5139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5921223724903907e-06,
+      "loss": 3.4152,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.590203167176555e-06,
+      "loss": 3.4235,
+      "step": 5141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.588284187486607e-06,
+      "loss": 3.4137,
+      "step": 5142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.586365433727657e-06,
+      "loss": 3.3022,
+      "step": 5143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5844469062067845e-06,
+      "loss": 3.4151,
+      "step": 5144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5825286052310315e-06,
+      "loss": 3.3876,
+      "step": 5145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5806105311074002e-06,
+      "loss": 3.4322,
+      "step": 5146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5786926841428617e-06,
+      "loss": 3.4977,
+      "step": 5147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5767750646443456e-06,
+      "loss": 3.4041,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.574857672918749e-06,
+      "loss": 3.4082,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.572940509272934e-06,
+      "loss": 3.3156,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5710235740137177e-06,
+      "loss": 3.4858,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5691068674478914e-06,
+      "loss": 3.4009,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5671903898822035e-06,
+      "loss": 3.4135,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5652741416233655e-06,
+      "loss": 3.3922,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5633581229780558e-06,
+      "loss": 3.3942,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.561442334252912e-06,
+      "loss": 3.3756,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5595267757545393e-06,
+      "loss": 3.4887,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5576114477895018e-06,
+      "loss": 3.4984,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5556963506643294e-06,
+      "loss": 3.3499,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5537814846855147e-06,
+      "loss": 3.4524,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5518668501595116e-06,
+      "loss": 3.34,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.549952447392738e-06,
+      "loss": 3.3527,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5480382766915784e-06,
+      "loss": 3.3376,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.546124338362371e-06,
+      "loss": 3.4248,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5442106327114267e-06,
+      "loss": 3.3813,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.542297160045015e-06,
+      "loss": 3.4358,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.540383920669366e-06,
+      "loss": 3.4117,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5384709148906744e-06,
+      "loss": 3.4114,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.536558143015101e-06,
+      "loss": 3.4976,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.534645605348761e-06,
+      "loss": 3.5251,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5327333021977415e-06,
+      "loss": 3.4668,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5308212338680842e-06,
+      "loss": 3.4647,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.528909400665797e-06,
+      "loss": 3.411,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5269978028968526e-06,
+      "loss": 3.4382,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5250864408671786e-06,
+      "loss": 3.5914,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.523175314882674e-06,
+      "loss": 3.4337,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5212644252491912e-06,
+      "loss": 3.4818,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.51935377227255e-06,
+      "loss": 3.4009,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5174433562585353e-06,
+      "loss": 3.3668,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5155331775128844e-06,
+      "loss": 3.4176,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.513623236341305e-06,
+      "loss": 3.3761,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5117135330494663e-06,
+      "loss": 3.4815,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5098040679429923e-06,
+      "loss": 3.4474,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5078948413274783e-06,
+      "loss": 3.457,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5059858535084733e-06,
+      "loss": 3.4308,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5040771047914934e-06,
+      "loss": 3.4282,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.502168595482016e-06,
+      "loss": 3.4347,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5002603258854765e-06,
+      "loss": 3.3595,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4983522963072756e-06,
+      "loss": 3.5134,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4964445070527764e-06,
+      "loss": 3.4815,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.494536958427297e-06,
+      "loss": 3.421,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4926296507361246e-06,
+      "loss": 3.3535,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4907225842845045e-06,
+      "loss": 3.4062,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4888157593776416e-06,
+      "loss": 3.4457,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4869091763207073e-06,
+      "loss": 3.2062,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4850028354188293e-06,
+      "loss": 3.4138,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4830967369770976e-06,
+      "loss": 3.3742,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.481190881300566e-06,
+      "loss": 3.3635,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.479285268694247e-06,
+      "loss": 3.33,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.477379899463115e-06,
+      "loss": 3.4853,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.475474773912103e-06,
+      "loss": 3.4293,
+      "step": 5201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4735698923461113e-06,
+      "loss": 3.3664,
+      "step": 5202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4716652550699946e-06,
+      "loss": 3.3939,
+      "step": 5203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4697608623885716e-06,
+      "loss": 3.42,
+      "step": 5204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.467856714606621e-06,
+      "loss": 3.3332,
+      "step": 5205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4659528120288837e-06,
+      "loss": 3.4626,
+      "step": 5206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4640491549600585e-06,
+      "loss": 3.4429,
+      "step": 5207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4621457437048096e-06,
+      "loss": 3.3634,
+      "step": 5208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4602425785677548e-06,
+      "loss": 3.3456,
+      "step": 5209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4583396598534793e-06,
+      "loss": 3.4241,
+      "step": 5210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.456436987866527e-06,
+      "loss": 3.3055,
+      "step": 5211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4545345629113985e-06,
+      "loss": 3.5001,
+      "step": 5212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.45263238529256e-06,
+      "loss": 3.3615,
+      "step": 5213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.450730455314437e-06,
+      "loss": 3.4452,
+      "step": 5214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.448828773281412e-06,
+      "loss": 3.4307,
+      "step": 5215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.446927339497833e-06,
+      "loss": 3.3417,
+      "step": 5216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4450261542680015e-06,
+      "loss": 3.3962,
+      "step": 5217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4431252178961854e-06,
+      "loss": 3.4394,
+      "step": 5218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4412245306866126e-06,
+      "loss": 3.4449,
+      "step": 5219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.439324092943465e-06,
+      "loss": 3.4973,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4374239049708914e-06,
+      "loss": 3.3951,
+      "step": 5221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4355239670729983e-06,
+      "loss": 3.4891,
+      "step": 5222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4336242795538487e-06,
+      "loss": 3.4609,
+      "step": 5223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.431724842717473e-06,
+      "loss": 3.352,
+      "step": 5224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4298256568678522e-06,
+      "loss": 3.4418,
+      "step": 5225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4279267223089353e-06,
+      "loss": 3.4661,
+      "step": 5226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4260280393446277e-06,
+      "loss": 3.3474,
+      "step": 5227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4241296082787922e-06,
+      "loss": 3.4428,
+      "step": 5228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.422231429415256e-06,
+      "loss": 3.3355,
+      "step": 5229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4203335030578043e-06,
+      "loss": 3.4329,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.418435829510178e-06,
+      "loss": 3.3619,
+      "step": 5231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4165384090760857e-06,
+      "loss": 3.4007,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4146412420591845e-06,
+      "loss": 3.4262,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4127443287631003e-06,
+      "loss": 3.4094,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.410847669491417e-06,
+      "loss": 3.4301,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4089512645476713e-06,
+      "loss": 3.384,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.407055114235367e-06,
+      "loss": 3.3876,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4051592188579653e-06,
+      "loss": 3.3564,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.403263578718881e-06,
+      "loss": 3.3423,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.401368194121496e-06,
+      "loss": 3.3763,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3994730653691465e-06,
+      "loss": 3.2961,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3975781927651274e-06,
+      "loss": 3.478,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.395683576612697e-06,
+      "loss": 3.3633,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3937892172150678e-06,
+      "loss": 3.3842,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3918951148754136e-06,
+      "loss": 3.3612,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3900012698968656e-06,
+      "loss": 3.4209,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3881076825825155e-06,
+      "loss": 3.397,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3862143532354148e-06,
+      "loss": 3.4214,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3843212821585682e-06,
+      "loss": 3.389,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3824284696549448e-06,
+      "loss": 3.4225,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.380535916027472e-06,
+      "loss": 3.4303,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3786436215790308e-06,
+      "loss": 3.3998,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.376751586612468e-06,
+      "loss": 3.4228,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3748598114305795e-06,
+      "loss": 3.4407,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.372968296336129e-06,
+      "loss": 3.6102,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.371077041631835e-06,
+      "loss": 3.4455,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.36918604762037e-06,
+      "loss": 3.4381,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.367295314604372e-06,
+      "loss": 3.4292,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.365404842886435e-06,
+      "loss": 3.4971,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3635146327691053e-06,
+      "loss": 3.3762,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.361624684554897e-06,
+      "loss": 3.3819,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3597349985462747e-06,
+      "loss": 3.456,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.357845575045663e-06,
+      "loss": 3.4578,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3559564143554495e-06,
+      "loss": 3.317,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3540675167779702e-06,
+      "loss": 3.3466,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3521788826155267e-06,
+      "loss": 3.4561,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.350290512170379e-06,
+      "loss": 3.3477,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3484024057447356e-06,
+      "loss": 3.4294,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3465145636407746e-06,
+      "loss": 3.3971,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.344626986160623e-06,
+      "loss": 3.4476,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.342739673606369e-06,
+      "loss": 3.3623,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3408526262800612e-06,
+      "loss": 3.4193,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.338965844483698e-06,
+      "loss": 3.4433,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.337079328519243e-06,
+      "loss": 3.3492,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3351930786886144e-06,
+      "loss": 3.4018,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3333070952936853e-06,
+      "loss": 3.4812,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3314213786362905e-06,
+      "loss": 3.3703,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3295359290182195e-06,
+      "loss": 3.3577,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.327650746741219e-06,
+      "loss": 3.301,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3257658321069952e-06,
+      "loss": 3.3752,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.323881185417209e-06,
+      "loss": 3.4086,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.321996806973478e-06,
+      "loss": 3.4713,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3201126970773794e-06,
+      "loss": 3.3687,
+      "step": 5283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3182288560304465e-06,
+      "loss": 3.4056,
+      "step": 5284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.316345284134169e-06,
+      "loss": 3.4816,
+      "step": 5285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3144619816899924e-06,
+      "loss": 3.4333,
+      "step": 5286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.312578948999323e-06,
+      "loss": 3.418,
+      "step": 5287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.310696186363519e-06,
+      "loss": 3.4083,
+      "step": 5288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3088136940838977e-06,
+      "loss": 3.3839,
+      "step": 5289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3069314724617353e-06,
+      "loss": 3.4428,
+      "step": 5290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3050495217982613e-06,
+      "loss": 3.431,
+      "step": 5291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.303167842394662e-06,
+      "loss": 3.4289,
+      "step": 5292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.301286434552084e-06,
+      "loss": 3.4201,
+      "step": 5293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2994052985716245e-06,
+      "loss": 3.4554,
+      "step": 5294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2975244347543422e-06,
+      "loss": 3.4121,
+      "step": 5295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.295643843401253e-06,
+      "loss": 3.4248,
+      "step": 5296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2937635248133227e-06,
+      "loss": 3.4477,
+      "step": 5297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2918834792914777e-06,
+      "loss": 3.474,
+      "step": 5298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2900037071366055e-06,
+      "loss": 3.3484,
+      "step": 5299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.288124208649538e-06,
+      "loss": 3.4306,
+      "step": 5300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2862449841310763e-06,
+      "loss": 3.4807,
+      "step": 5301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.284366033881966e-06,
+      "loss": 3.4212,
+      "step": 5302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.282487358202917e-06,
+      "loss": 3.4204,
+      "step": 5303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2806089573945956e-06,
+      "loss": 3.4759,
+      "step": 5304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2787308317576144e-06,
+      "loss": 3.4666,
+      "step": 5305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.276852981592554e-06,
+      "loss": 3.409,
+      "step": 5306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.274975407199943e-06,
+      "loss": 3.5012,
+      "step": 5307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2730981088802684e-06,
+      "loss": 3.2959,
+      "step": 5308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2712210869339755e-06,
+      "loss": 3.3926,
+      "step": 5309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2693443416614595e-06,
+      "loss": 3.3701,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.267467873363076e-06,
+      "loss": 3.3759,
+      "step": 5311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2655916823391374e-06,
+      "loss": 3.3852,
+      "step": 5312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2637157688899057e-06,
+      "loss": 3.4607,
+      "step": 5313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2618401333156054e-06,
+      "loss": 3.4501,
+      "step": 5314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2599647759164087e-06,
+      "loss": 3.4561,
+      "step": 5315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2580896969924512e-06,
+      "loss": 3.4408,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2562148968438224e-06,
+      "loss": 3.3419,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.25434037577056e-06,
+      "loss": 3.4068,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.252466134072665e-06,
+      "loss": 3.3737,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2505921720500938e-06,
+      "loss": 3.3501,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.24871849000275e-06,
+      "loss": 3.4196,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.246845088230502e-06,
+      "loss": 3.3288,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.244971967033167e-06,
+      "loss": 3.4208,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2430991267105187e-06,
+      "loss": 3.3847,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2412265675622887e-06,
+      "loss": 3.3179,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2393542898881597e-06,
+      "loss": 3.4777,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2374822939877704e-06,
+      "loss": 3.3927,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2356105801607176e-06,
+      "loss": 3.4778,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2337391487065485e-06,
+      "loss": 3.424,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2318679999247673e-06,
+      "loss": 3.4273,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.229997134114832e-06,
+      "loss": 3.288,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2281265515761572e-06,
+      "loss": 3.3739,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.226256252608111e-06,
+      "loss": 3.3694,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2243862375100137e-06,
+      "loss": 3.379,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.222516506581146e-06,
+      "loss": 3.3784,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2206470601207375e-06,
+      "loss": 3.4703,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2187778984279732e-06,
+      "loss": 3.3869,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2169090218019984e-06,
+      "loss": 3.4608,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.215040430541902e-06,
+      "loss": 3.4703,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2131721249467373e-06,
+      "loss": 3.4298,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.211304105315508e-06,
+      "loss": 3.4598,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.20943637194717e-06,
+      "loss": 3.3434,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.207568925140636e-06,
+      "loss": 3.363,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.205701765194775e-06,
+      "loss": 3.3467,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2038348924084017e-06,
+      "loss": 3.4217,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2019683070802965e-06,
+      "loss": 3.3877,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.200102009509182e-06,
+      "loss": 3.4587,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.198235999993744e-06,
+      "loss": 3.3993,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1963702788326194e-06,
+      "loss": 3.3956,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.194504846324395e-06,
+      "loss": 3.4591,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.192639702767616e-06,
+      "loss": 3.4047,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1907748484607824e-06,
+      "loss": 3.5004,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1889102837023413e-06,
+      "loss": 3.4512,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1870460087907016e-06,
+      "loss": 3.3806,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.185182024024219e-06,
+      "loss": 3.4103,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1833183297012056e-06,
+      "loss": 3.3848,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1814549261199303e-06,
+      "loss": 3.4926,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1795918135786076e-06,
+      "loss": 3.3887,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.177728992375413e-06,
+      "loss": 3.487,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1758664628084735e-06,
+      "loss": 3.4713,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.174004225175864e-06,
+      "loss": 3.3446,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.172142279775623e-06,
+      "loss": 3.2557,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1702806269057296e-06,
+      "loss": 3.3868,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1684192668641267e-06,
+      "loss": 3.4144,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1665581999487074e-06,
+      "loss": 3.4271,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1646974264573132e-06,
+      "loss": 3.3723,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.162836946687744e-06,
+      "loss": 3.437,
+      "step": 5367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.160976760937754e-06,
+      "loss": 3.3523,
+      "step": 5368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1591168695050418e-06,
+      "loss": 3.3341,
+      "step": 5369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1572572726872684e-06,
+      "loss": 3.3557,
+      "step": 5370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1553979707820413e-06,
+      "loss": 3.3484,
+      "step": 5371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1535389640869253e-06,
+      "loss": 3.4384,
+      "step": 5372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.151680252899435e-06,
+      "loss": 3.3916,
+      "step": 5373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1498218375170368e-06,
+      "loss": 3.4007,
+      "step": 5374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1479637182371557e-06,
+      "loss": 3.4144,
+      "step": 5375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1461058953571605e-06,
+      "loss": 3.4474,
+      "step": 5376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1442483691743786e-06,
+      "loss": 3.408,
+      "step": 5377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.142391139986092e-06,
+      "loss": 3.4695,
+      "step": 5378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.140534208089526e-06,
+      "loss": 3.3176,
+      "step": 5379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1386775737818664e-06,
+      "loss": 3.4836,
+      "step": 5380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1368212373602525e-06,
+      "loss": 3.4279,
+      "step": 5381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1349651991217653e-06,
+      "loss": 3.4353,
+      "step": 5382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1331094593634528e-06,
+      "loss": 3.5562,
+      "step": 5383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.131254018382301e-06,
+      "loss": 3.3554,
+      "step": 5384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.129398876475257e-06,
+      "loss": 3.367,
+      "step": 5385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.127544033939221e-06,
+      "loss": 3.5343,
+      "step": 5386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1256894910710365e-06,
+      "loss": 3.3651,
+      "step": 5387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1238352481675084e-06,
+      "loss": 3.4205,
+      "step": 5388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.121981305525389e-06,
+      "loss": 3.4975,
+      "step": 5389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1201276634413817e-06,
+      "loss": 3.418,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1182743222121463e-06,
+      "loss": 3.4425,
+      "step": 5391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1164212821342876e-06,
+      "loss": 3.3383,
+      "step": 5392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.114568543504369e-06,
+      "loss": 3.4563,
+      "step": 5393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1127161066189037e-06,
+      "loss": 3.3934,
+      "step": 5394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1108639717743516e-06,
+      "loss": 3.5136,
+      "step": 5395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.109012139267131e-06,
+      "loss": 3.4319,
+      "step": 5396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1071606093936114e-06,
+      "loss": 3.3591,
+      "step": 5397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1053093824501066e-06,
+      "loss": 3.3183,
+      "step": 5398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.103458458732893e-06,
+      "loss": 3.4456,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.101607838538186e-06,
+      "loss": 3.4651,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.099757522162162e-06,
+      "loss": 3.4326,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0979075099009475e-06,
+      "loss": 3.4335,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.096057802050615e-06,
+      "loss": 3.3765,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0942083989071923e-06,
+      "loss": 3.3315,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.092359300766662e-06,
+      "loss": 3.3685,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0905105079249477e-06,
+      "loss": 3.3213,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.088662020677934e-06,
+      "loss": 3.4577,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0868138393214525e-06,
+      "loss": 3.5689,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.084965964151284e-06,
+      "loss": 3.385,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.083118395463166e-06,
+      "loss": 3.4447,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0812711335527807e-06,
+      "loss": 3.4268,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.079424178715763e-06,
+      "loss": 3.4134,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.077577531247703e-06,
+      "loss": 3.422,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0757311914441364e-06,
+      "loss": 3.4341,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.073885159600551e-06,
+      "loss": 3.3194,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.072039436012385e-06,
+      "loss": 3.3638,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0701940209750304e-06,
+      "loss": 3.3985,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0683489147838253e-06,
+      "loss": 3.4323,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.066504117734061e-06,
+      "loss": 3.4479,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0646596301209795e-06,
+      "loss": 3.3784,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0628154522397737e-06,
+      "loss": 3.471,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.060971584385582e-06,
+      "loss": 3.4717,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0591280268535016e-06,
+      "loss": 3.4105,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.057284779938571e-06,
+      "loss": 3.3562,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0554418439357858e-06,
+      "loss": 3.4427,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.053599219140092e-06,
+      "loss": 3.4204,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0517569058463773e-06,
+      "loss": 3.4884,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.04991490434949e-06,
+      "loss": 3.3668,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0480732149442237e-06,
+      "loss": 3.4133,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0462318379253207e-06,
+      "loss": 3.4246,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0443907735874774e-06,
+      "loss": 3.3974,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0425500222253333e-06,
+      "loss": 3.447,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0407095841334865e-06,
+      "loss": 3.4438,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0388694596064804e-06,
+      "loss": 3.4212,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.037029648938805e-06,
+      "loss": 3.397,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.035190152424909e-06,
+      "loss": 3.3813,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.03335097035918e-06,
+      "loss": 3.4089,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.031512103035963e-06,
+      "loss": 3.4575,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.029673550749553e-06,
+      "loss": 3.4554,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0278353137941867e-06,
+      "loss": 3.4096,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0259973924640586e-06,
+      "loss": 3.3482,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.024159787053311e-06,
+      "loss": 3.4957,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0223224978560306e-06,
+      "loss": 3.3852,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.020485525166261e-06,
+      "loss": 3.4182,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.018648869277988e-06,
+      "loss": 3.4395,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0168125304851512e-06,
+      "loss": 3.4241,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.01497650908164e-06,
+      "loss": 3.3591,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.013140805361288e-06,
+      "loss": 3.3752,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.011305419617883e-06,
+      "loss": 3.5222,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0094703521451626e-06,
+      "loss": 3.4358,
+      "step": 5450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0076356032368057e-06,
+      "loss": 3.435,
+      "step": 5451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0058011731864493e-06,
+      "loss": 3.3526,
+      "step": 5452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.003967062287675e-06,
+      "loss": 3.4482,
+      "step": 5453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.002133270834012e-06,
+      "loss": 3.3275,
+      "step": 5454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0002997991189434e-06,
+      "loss": 3.4259,
+      "step": 5455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9984666474358963e-06,
+      "loss": 3.3928,
+      "step": 5456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9966338160782477e-06,
+      "loss": 3.3528,
+      "step": 5457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9948013053393253e-06,
+      "loss": 3.3341,
+      "step": 5458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.992969115512403e-06,
+      "loss": 3.4583,
+      "step": 5459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.991137246890706e-06,
+      "loss": 3.4512,
+      "step": 5460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.989305699767403e-06,
+      "loss": 3.4097,
+      "step": 5461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.98747447443562e-06,
+      "loss": 3.4464,
+      "step": 5462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9856435711884225e-06,
+      "loss": 3.4181,
+      "step": 5463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9838129903188283e-06,
+      "loss": 3.3303,
+      "step": 5464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.981982732119805e-06,
+      "loss": 3.4777,
+      "step": 5465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9801527968842682e-06,
+      "loss": 3.4417,
+      "step": 5466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9783231849050777e-06,
+      "loss": 3.3723,
+      "step": 5467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.976493896475048e-06,
+      "loss": 3.4718,
+      "step": 5468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9746649318869346e-06,
+      "loss": 3.2979,
+      "step": 5469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9728362914334473e-06,
+      "loss": 3.3194,
+      "step": 5470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.971007975407243e-06,
+      "loss": 3.4682,
+      "step": 5471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9691799841009216e-06,
+      "loss": 3.4099,
+      "step": 5472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.967352317807038e-06,
+      "loss": 3.3772,
+      "step": 5473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.965524976818092e-06,
+      "loss": 3.425,
+      "step": 5474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9636979614265287e-06,
+      "loss": 3.493,
+      "step": 5475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9618712719247466e-06,
+      "loss": 3.4659,
+      "step": 5476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9600449086050863e-06,
+      "loss": 3.4143,
+      "step": 5477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9582188717598405e-06,
+      "loss": 3.4552,
+      "step": 5478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.956393161681249e-06,
+      "loss": 3.3956,
+      "step": 5479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9545677786614957e-06,
+      "loss": 3.4477,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.952742722992717e-06,
+      "loss": 3.464,
+      "step": 5481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9509179949669964e-06,
+      "loss": 3.346,
+      "step": 5482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9490935948763586e-06,
+      "loss": 3.3009,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.947269523012786e-06,
+      "loss": 3.401,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.945445779668199e-06,
+      "loss": 3.4549,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9436223651344697e-06,
+      "loss": 3.3461,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.941799279703421e-06,
+      "loss": 3.3544,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9399765236668155e-06,
+      "loss": 3.3809,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9381540973163684e-06,
+      "loss": 3.4076,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9363320009437436e-06,
+      "loss": 3.2837,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.934510234840544e-06,
+      "loss": 3.4479,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9326887992983305e-06,
+      "loss": 3.416,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9308676946086013e-06,
+      "loss": 3.4711,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.929046921062807e-06,
+      "loss": 3.3341,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9272264789523465e-06,
+      "loss": 3.4603,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9254063685685613e-06,
+      "loss": 3.4744,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9235865902027417e-06,
+      "loss": 3.469,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.921767144146127e-06,
+      "loss": 3.362,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9199480306898996e-06,
+      "loss": 3.4054,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9181292501251924e-06,
+      "loss": 3.4413,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9163108027430782e-06,
+      "loss": 3.4547,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.914492688834587e-06,
+      "loss": 3.3803,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.912674908690687e-06,
+      "loss": 3.4058,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9108574626022955e-06,
+      "loss": 3.4433,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.909040350860278e-06,
+      "loss": 3.3705,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9072235737554443e-06,
+      "loss": 3.3475,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9054071315785496e-06,
+      "loss": 3.4369,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.903591024620303e-06,
+      "loss": 3.3713,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.901775253171346e-06,
+      "loss": 3.4057,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.899959817522281e-06,
+      "loss": 3.3904,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.898144717963649e-06,
+      "loss": 3.3328,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.896329954785937e-06,
+      "loss": 3.3994,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.894515528279581e-06,
+      "loss": 3.4162,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.892701438734962e-06,
+      "loss": 3.296,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.890887686442403e-06,
+      "loss": 3.415,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8890742716921834e-06,
+      "loss": 3.3307,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8872611947745175e-06,
+      "loss": 3.4262,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8854484559795733e-06,
+      "loss": 3.3856,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8836360555974583e-06,
+      "loss": 3.4422,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8818239939182313e-06,
+      "loss": 3.3032,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8800122712318936e-06,
+      "loss": 3.4437,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8782008878283916e-06,
+      "loss": 3.4471,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8763898439976233e-06,
+      "loss": 3.3843,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.874579140029425e-06,
+      "loss": 3.4385,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8727687762135837e-06,
+      "loss": 3.4598,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8709587528398285e-06,
+      "loss": 3.4595,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8691490701978365e-06,
+      "loss": 3.3873,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8673397285772264e-06,
+      "loss": 3.3046,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.865530728267572e-06,
+      "loss": 3.344,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8637220695583774e-06,
+      "loss": 3.4373,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.861913752739106e-06,
+      "loss": 3.3786,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.860105778099159e-06,
+      "loss": 3.4661,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.858298145927886e-06,
+      "loss": 3.4161,
+      "step": 5533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8564908565145767e-06,
+      "loss": 3.4311,
+      "step": 5534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.854683910148476e-06,
+      "loss": 3.3358,
+      "step": 5535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8528773071187597e-06,
+      "loss": 3.4601,
+      "step": 5536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8510710477145653e-06,
+      "loss": 3.3826,
+      "step": 5537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8492651322249577e-06,
+      "loss": 3.416,
+      "step": 5538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.847459560938962e-06,
+      "loss": 3.4313,
+      "step": 5539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.84565433414554e-06,
+      "loss": 3.4552,
+      "step": 5540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.843849452133599e-06,
+      "loss": 3.3656,
+      "step": 5541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.842044915191992e-06,
+      "loss": 3.3039,
+      "step": 5542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8402407236095223e-06,
+      "loss": 3.3443,
+      "step": 5543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.838436877674925e-06,
+      "loss": 3.4098,
+      "step": 5544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.836633377676895e-06,
+      "loss": 3.3526,
+      "step": 5545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.834830223904057e-06,
+      "loss": 3.3869,
+      "step": 5546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8330274166449933e-06,
+      "loss": 3.4326,
+      "step": 5547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.831224956188224e-06,
+      "loss": 3.4223,
+      "step": 5548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8294228428222136e-06,
+      "loss": 3.4032,
+      "step": 5549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.827621076835371e-06,
+      "loss": 3.4919,
+      "step": 5550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8258196585160574e-06,
+      "loss": 3.3891,
+      "step": 5551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.824018588152563e-06,
+      "loss": 3.4419,
+      "step": 5552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8222178660331394e-06,
+      "loss": 3.311,
+      "step": 5553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.820417492445966e-06,
+      "loss": 3.3937,
+      "step": 5554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.818617467679181e-06,
+      "loss": 3.3658,
+      "step": 5555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.816817792020858e-06,
+      "loss": 3.3745,
+      "step": 5556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.815018465759017e-06,
+      "loss": 3.4338,
+      "step": 5557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.81321948918162e-06,
+      "loss": 3.4195,
+      "step": 5558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.811420862576582e-06,
+      "loss": 3.4586,
+      "step": 5559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8096225862317467e-06,
+      "loss": 3.3109,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8078246604349164e-06,
+      "loss": 3.4422,
+      "step": 5561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8060270854738287e-06,
+      "loss": 3.4197,
+      "step": 5562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.804229861636168e-06,
+      "loss": 3.4767,
+      "step": 5563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.802432989209562e-06,
+      "loss": 3.379,
+      "step": 5564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.800636468481582e-06,
+      "loss": 3.388,
+      "step": 5565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.798840299739743e-06,
+      "loss": 3.4071,
+      "step": 5566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.797044483271502e-06,
+      "loss": 3.5156,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7952490193642657e-06,
+      "loss": 3.4179,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.793453908305378e-06,
+      "loss": 3.4683,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7916591503821285e-06,
+      "loss": 3.3832,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7898647458817474e-06,
+      "loss": 3.4608,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7880706950914196e-06,
+      "loss": 3.473,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.786276998298254e-06,
+      "loss": 3.3951,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7844836557893236e-06,
+      "loss": 3.4602,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.782690667851626e-06,
+      "loss": 3.3709,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.780898034772118e-06,
+      "loss": 3.4117,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7791057568376905e-06,
+      "loss": 3.4003,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.777313834335179e-06,
+      "loss": 3.2915,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7755222675513616e-06,
+      "loss": 3.3362,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7737310567729652e-06,
+      "loss": 3.331,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7719402022866483e-06,
+      "loss": 3.3699,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.770149704379027e-06,
+      "loss": 3.5099,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7683595633366456e-06,
+      "loss": 3.5185,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7665697794460023e-06,
+      "loss": 3.4332,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7647803529935337e-06,
+      "loss": 3.3615,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7629912842656187e-06,
+      "loss": 3.4069,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.761202573548578e-06,
+      "loss": 3.3344,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7594142211286834e-06,
+      "loss": 3.4359,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.757626227292135e-06,
+      "loss": 3.4101,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7558385923250897e-06,
+      "loss": 3.42,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7540513165136345e-06,
+      "loss": 3.375,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7522644001438104e-06,
+      "loss": 3.3416,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7504778435015925e-06,
+      "loss": 3.342,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.748691646872903e-06,
+      "loss": 3.3128,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.746905810543602e-06,
+      "loss": 3.419,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7451203347995005e-06,
+      "loss": 3.4226,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7433352199263373e-06,
+      "loss": 3.4995,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7415504662098124e-06,
+      "loss": 3.2569,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.739766073935548e-06,
+      "loss": 3.4978,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7379820433891245e-06,
+      "loss": 3.4194,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7361983748560572e-06,
+      "loss": 3.4532,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7344150686218036e-06,
+      "loss": 3.4811,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7326321249717615e-06,
+      "loss": 3.451,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7308495441912786e-06,
+      "loss": 3.3133,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.729067326565635e-06,
+      "loss": 3.361,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.727285472380059e-06,
+      "loss": 3.4933,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.725503981919718e-06,
+      "loss": 3.3867,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7237228554697203e-06,
+      "loss": 3.4515,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.72194209331512e-06,
+      "loss": 3.3558,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7201616957409055e-06,
+      "loss": 3.4155,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.718381663032017e-06,
+      "loss": 3.3943,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7166019954733293e-06,
+      "loss": 3.3433,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7148226933496597e-06,
+      "loss": 3.4373,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7130437569457678e-06,
+      "loss": 3.4758,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.711265186546355e-06,
+      "loss": 3.4629,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7094869824360615e-06,
+      "loss": 3.4235,
+      "step": 5616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.707709144899476e-06,
+      "loss": 3.5101,
+      "step": 5617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7059316742211177e-06,
+      "loss": 3.4082,
+      "step": 5618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7041545706854576e-06,
+      "loss": 3.362,
+      "step": 5619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7023778345769014e-06,
+      "loss": 3.4597,
+      "step": 5620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7006014661797985e-06,
+      "loss": 3.426,
+      "step": 5621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.698825465778439e-06,
+      "loss": 3.4382,
+      "step": 5622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.697049833657053e-06,
+      "loss": 3.3946,
+      "step": 5623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.695274570099812e-06,
+      "loss": 3.3991,
+      "step": 5624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6934996753908336e-06,
+      "loss": 3.4194,
+      "step": 5625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.691725149814166e-06,
+      "loss": 3.4418,
+      "step": 5626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6899509936538083e-06,
+      "loss": 3.3955,
+      "step": 5627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6881772071936952e-06,
+      "loss": 3.5197,
+      "step": 5628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.686403790717704e-06,
+      "loss": 3.4461,
+      "step": 5629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6846307445096515e-06,
+      "loss": 3.3969,
+      "step": 5630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6828580688532964e-06,
+      "loss": 3.3562,
+      "step": 5631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6810857640323356e-06,
+      "loss": 3.4519,
+      "step": 5632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6793138303304143e-06,
+      "loss": 3.4643,
+      "step": 5633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6775422680311062e-06,
+      "loss": 3.4978,
+      "step": 5634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.675771077417939e-06,
+      "loss": 3.4619,
+      "step": 5635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6740002587743664e-06,
+      "loss": 3.3931,
+      "step": 5636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.672229812383796e-06,
+      "loss": 3.4296,
+      "step": 5637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.670459738529568e-06,
+      "loss": 3.2801,
+      "step": 5638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6686900374949655e-06,
+      "loss": 3.3839,
+      "step": 5639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.666920709563208e-06,
+      "loss": 3.4143,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6651517550174666e-06,
+      "loss": 3.4027,
+      "step": 5641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6633831741408365e-06,
+      "loss": 3.4534,
+      "step": 5642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.661614967216366e-06,
+      "loss": 3.4057,
+      "step": 5643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6598471345270383e-06,
+      "loss": 3.5032,
+      "step": 5644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.658079676355777e-06,
+      "loss": 3.3414,
+      "step": 5645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.656312592985446e-06,
+      "loss": 3.4442,
+      "step": 5646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6545458846988494e-06,
+      "loss": 3.414,
+      "step": 5647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6527795517787293e-06,
+      "loss": 3.3741,
+      "step": 5648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.651013594507773e-06,
+      "loss": 3.4676,
+      "step": 5649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6492480131686027e-06,
+      "loss": 3.4281,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.647482808043782e-06,
+      "loss": 3.394,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.645717979415815e-06,
+      "loss": 3.4745,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6439535275671437e-06,
+      "loss": 3.3654,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.642189452780152e-06,
+      "loss": 3.3814,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6404257553371603e-06,
+      "loss": 3.5507,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6386624355204336e-06,
+      "loss": 3.459,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6368994936121726e-06,
+      "loss": 3.3028,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.635136929894519e-06,
+      "loss": 3.3256,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6333747446495526e-06,
+      "loss": 3.3088,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6316129381592935e-06,
+      "loss": 3.3281,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6298515107057e-06,
+      "loss": 3.3554,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.628090462570676e-06,
+      "loss": 3.3705,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6263297940360523e-06,
+      "loss": 3.4714,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.624569505383612e-06,
+      "loss": 3.4288,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6228095968950697e-06,
+      "loss": 3.3916,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.621050068852081e-06,
+      "loss": 3.3085,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.619290921536242e-06,
+      "loss": 3.4197,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.617532155229085e-06,
+      "loss": 3.3872,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6157737702120817e-06,
+      "loss": 3.3467,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.61401576676665e-06,
+      "loss": 3.4296,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6122581451741323e-06,
+      "loss": 3.4106,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.610500905715826e-06,
+      "loss": 3.4557,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.608744048672955e-06,
+      "loss": 3.3529,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6069875743266894e-06,
+      "loss": 3.4633,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6052314829581337e-06,
+      "loss": 3.5049,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.603475774848333e-06,
+      "loss": 3.2917,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.601720450278269e-06,
+      "loss": 3.4252,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5999655095288702e-06,
+      "loss": 3.4223,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.598210952880989e-06,
+      "loss": 3.3697,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.596456780615431e-06,
+      "loss": 3.4695,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.594702993012931e-06,
+      "loss": 3.5218,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5929495903541655e-06,
+      "loss": 3.3794,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5911965729197496e-06,
+      "loss": 3.4059,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.589443940990236e-06,
+      "loss": 3.3801,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.587691694846114e-06,
+      "loss": 3.4039,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.585939834767819e-06,
+      "loss": 3.4404,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5841883610357103e-06,
+      "loss": 3.4018,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5824372739301007e-06,
+      "loss": 3.3953,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5806865737312304e-06,
+      "loss": 3.4492,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5789362607192837e-06,
+      "loss": 3.5228,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.577186335174379e-06,
+      "loss": 3.4795,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5754367973765747e-06,
+      "loss": 3.4021,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.573687647605866e-06,
+      "loss": 3.4178,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5719388861421894e-06,
+      "loss": 3.4354,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.570190513265416e-06,
+      "loss": 3.4332,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.568442529255354e-06,
+      "loss": 3.3872,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5666949343917514e-06,
+      "loss": 3.3306,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.564947728954291e-06,
+      "loss": 3.3872,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.563200913222602e-06,
+      "loss": 3.5253,
+      "step": 5699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.561454487476237e-06,
+      "loss": 3.4855,
+      "step": 5700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5597084519946993e-06,
+      "loss": 3.3746,
+      "step": 5701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5579628070574227e-06,
+      "loss": 3.417,
+      "step": 5702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5562175529437805e-06,
+      "loss": 3.3661,
+      "step": 5703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5544726899330824e-06,
+      "loss": 3.3903,
+      "step": 5704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.552728218304577e-06,
+      "loss": 3.4119,
+      "step": 5705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.550984138337448e-06,
+      "loss": 3.3748,
+      "step": 5706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5492404503108226e-06,
+      "loss": 3.3674,
+      "step": 5707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.547497154503753e-06,
+      "loss": 3.4422,
+      "step": 5708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5457542511952436e-06,
+      "loss": 3.3903,
+      "step": 5709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.544011740664225e-06,
+      "loss": 3.4348,
+      "step": 5710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.542269623189569e-06,
+      "loss": 3.3647,
+      "step": 5711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.540527899050085e-06,
+      "loss": 3.3948,
+      "step": 5712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5387865685245172e-06,
+      "loss": 3.4473,
+      "step": 5713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.537045631891547e-06,
+      "loss": 3.4978,
+      "step": 5714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5353050894298e-06,
+      "loss": 3.4752,
+      "step": 5715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5335649414178244e-06,
+      "loss": 3.3942,
+      "step": 5716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5318251881341183e-06,
+      "loss": 3.3416,
+      "step": 5717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5300858298571117e-06,
+      "loss": 3.4343,
+      "step": 5718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5283468668651704e-06,
+      "loss": 3.4736,
+      "step": 5719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5266082994365987e-06,
+      "loss": 3.3942,
+      "step": 5720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.524870127849636e-06,
+      "loss": 3.2963,
+      "step": 5721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5231323523824587e-06,
+      "loss": 3.3742,
+      "step": 5722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.521394973313185e-06,
+      "loss": 3.3808,
+      "step": 5723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5196579909198575e-06,
+      "loss": 3.3796,
+      "step": 5724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.517921405480469e-06,
+      "loss": 3.4513,
+      "step": 5725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5161852172729405e-06,
+      "loss": 3.3883,
+      "step": 5726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.514449426575132e-06,
+      "loss": 3.4271,
+      "step": 5727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.512714033664839e-06,
+      "loss": 3.4631,
+      "step": 5728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5109790388197942e-06,
+      "loss": 3.3655,
+      "step": 5729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5092444423176633e-06,
+      "loss": 3.4235,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5075102444360555e-06,
+      "loss": 3.3724,
+      "step": 5731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.50577644545251e-06,
+      "loss": 3.449,
+      "step": 5732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5040430456445047e-06,
+      "loss": 3.3162,
+      "step": 5733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5023100452894516e-06,
+      "loss": 3.4069,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5005774446647012e-06,
+      "loss": 3.3659,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.498845244047538e-06,
+      "loss": 3.2956,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4971134437151816e-06,
+      "loss": 3.4168,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.495382043944794e-06,
+      "loss": 3.4288,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4936510450134654e-06,
+      "loss": 3.4138,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4919204471982256e-06,
+      "loss": 3.4523,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4901902507760396e-06,
+      "loss": 3.4171,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4884604560238072e-06,
+      "loss": 3.3624,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4867310632183634e-06,
+      "loss": 3.3099,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4850020726364858e-06,
+      "loss": 3.3816,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.483273484554875e-06,
+      "loss": 3.363,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.481545299250179e-06,
+      "loss": 3.4625,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.479817516998976e-06,
+      "loss": 3.4579,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4780901380777784e-06,
+      "loss": 3.4195,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4763631627630358e-06,
+      "loss": 3.3349,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4746365913311376e-06,
+      "loss": 3.3718,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.472910424058398e-06,
+      "loss": 3.5,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4711846612210793e-06,
+      "loss": 3.3358,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4694593030953654e-06,
+      "loss": 3.3525,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4677343499573887e-06,
+      "loss": 3.3445,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4660098020832087e-06,
+      "loss": 3.4443,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.464285659748821e-06,
+      "loss": 3.3494,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.462561923230159e-06,
+      "loss": 3.336,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4608385928030885e-06,
+      "loss": 3.4013,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.459115668743409e-06,
+      "loss": 3.4677,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4573931513268643e-06,
+      "loss": 3.3808,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4556710408291174e-06,
+      "loss": 3.4046,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.453949337525781e-06,
+      "loss": 3.3394,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4522280416923943e-06,
+      "loss": 3.4265,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.450507153604434e-06,
+      "loss": 3.3019,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.44878667353731e-06,
+      "loss": 3.4258,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4470666017663693e-06,
+      "loss": 3.319,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4453469385668894e-06,
+      "loss": 3.5336,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4436276842140912e-06,
+      "loss": 3.4314,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4419088389831164e-06,
+      "loss": 3.327,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.440190403149055e-06,
+      "loss": 3.4546,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4384723769869228e-06,
+      "loss": 3.5348,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4367547607716736e-06,
+      "loss": 3.3871,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.435037554778194e-06,
+      "loss": 3.4191,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.433320759281307e-06,
+      "loss": 3.4001,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.431604374555765e-06,
+      "loss": 3.3296,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4298884008762624e-06,
+      "loss": 3.4706,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4281728385174226e-06,
+      "loss": 3.3588,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4264576877538033e-06,
+      "loss": 3.4319,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.424742948859897e-06,
+      "loss": 3.5016,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.423028622110132e-06,
+      "loss": 3.4209,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4213147077788672e-06,
+      "loss": 3.4265,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4196012061403974e-06,
+      "loss": 3.3558,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.417888117468954e-06,
+      "loss": 3.4259,
+      "step": 5783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4161754420386983e-06,
+      "loss": 3.403,
+      "step": 5784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4144631801237267e-06,
+      "loss": 3.4386,
+      "step": 5785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.41275133199807e-06,
+      "loss": 3.3601,
+      "step": 5786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4110398979356915e-06,
+      "loss": 3.4748,
+      "step": 5787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.409328878210488e-06,
+      "loss": 3.4301,
+      "step": 5788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.407618273096296e-06,
+      "loss": 3.3578,
+      "step": 5789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4059080828668747e-06,
+      "loss": 3.3201,
+      "step": 5790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.404198307795927e-06,
+      "loss": 3.4244,
+      "step": 5791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.402488948157084e-06,
+      "loss": 3.4204,
+      "step": 5792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4007800042239117e-06,
+      "loss": 3.4092,
+      "step": 5793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3990714762699074e-06,
+      "loss": 3.4256,
+      "step": 5794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3973633645685097e-06,
+      "loss": 3.454,
+      "step": 5795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3956556693930773e-06,
+      "loss": 3.3572,
+      "step": 5796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3939483910169177e-06,
+      "loss": 3.4812,
+      "step": 5797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.392241529713255e-06,
+      "loss": 3.3889,
+      "step": 5798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3905350857552624e-06,
+      "loss": 3.4072,
+      "step": 5799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.388829059416036e-06,
+      "loss": 3.3556,
+      "step": 5800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3871234509686094e-06,
+      "loss": 3.3996,
+      "step": 5801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.385418260685945e-06,
+      "loss": 3.4145,
+      "step": 5802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383713488840948e-06,
+      "loss": 3.4166,
+      "step": 5803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.382009135706443e-06,
+      "loss": 3.476,
+      "step": 5804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.380305201555201e-06,
+      "loss": 3.5554,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.378601686659913e-06,
+      "loss": 3.4525,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3768985912932146e-06,
+      "loss": 3.4531,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3751959157276676e-06,
+      "loss": 3.4306,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3734936602357688e-06,
+      "loss": 3.4788,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.371791825089944e-06,
+      "loss": 3.3354,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3700904105625617e-06,
+      "loss": 3.3801,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3683894169259076e-06,
+      "loss": 3.4695,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3666888444522172e-06,
+      "loss": 3.4054,
+      "step": 5813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3649886934136424e-06,
+      "loss": 3.3624,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.363288964082281e-06,
+      "loss": 3.4356,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3615896567301557e-06,
+      "loss": 3.4208,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3598907716292236e-06,
+      "loss": 3.3309,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3581923090513718e-06,
+      "loss": 3.3638,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3564942692684274e-06,
+      "loss": 3.514,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3547966525521423e-06,
+      "loss": 3.3493,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3530994591742023e-06,
+      "loss": 3.4127,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.351402689406227e-06,
+      "loss": 3.3756,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.349706343519767e-06,
+      "loss": 3.3504,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.348010421786306e-06,
+      "loss": 3.3535,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.346314924477257e-06,
+      "loss": 3.3703,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3446198518639733e-06,
+      "loss": 3.4609,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.342925204217727e-06,
+      "loss": 3.5311,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3412309818097345e-06,
+      "loss": 3.4474,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3395371849111377e-06,
+      "loss": 3.3576,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3378438137930116e-06,
+      "loss": 3.3425,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.336150868726362e-06,
+      "loss": 3.4844,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.334458349982132e-06,
+      "loss": 3.451,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3327662578311873e-06,
+      "loss": 3.3472,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3310745925443357e-06,
+      "loss": 3.3995,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3293833543923046e-06,
+      "loss": 3.4301,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3276925436457647e-06,
+      "loss": 3.3976,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.326002160575313e-06,
+      "loss": 3.312,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.324312205451477e-06,
+      "loss": 3.375,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3226226785447158e-06,
+      "loss": 3.4585,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.320933580125427e-06,
+      "loss": 3.5112,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3192449104639257e-06,
+      "loss": 3.3802,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.317556669830475e-06,
+      "loss": 3.395,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3158688584952534e-06,
+      "loss": 3.3395,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.314181476728383e-06,
+      "loss": 3.2933,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3124945247999116e-06,
+      "loss": 3.3726,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.310808002979818e-06,
+      "loss": 3.3938,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.309121911538012e-06,
+      "loss": 3.2864,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3074362507443405e-06,
+      "loss": 3.4422,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3057510208685702e-06,
+      "loss": 3.3183,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.304066222180412e-06,
+      "loss": 3.3624,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3023818549494937e-06,
+      "loss": 3.5279,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3006979194453873e-06,
+      "loss": 3.4222,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2990144159375878e-06,
+      "loss": 3.4811,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.297331344695523e-06,
+      "loss": 3.3791,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.29564870598855e-06,
+      "loss": 3.3703,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.293966500085964e-06,
+      "loss": 3.4605,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.292284727256977e-06,
+      "loss": 3.3076,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.290603387770747e-06,
+      "loss": 3.4802,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2889224818963522e-06,
+      "loss": 3.4496,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.287242009902805e-06,
+      "loss": 3.5172,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2855619720590495e-06,
+      "loss": 3.3801,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2838823686339574e-06,
+      "loss": 3.4978,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2822031998963313e-06,
+      "loss": 3.4584,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.280524466114909e-06,
+      "loss": 3.2943,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2788461675583527e-06,
+      "loss": 3.5181,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2771683044952587e-06,
+      "loss": 3.5758,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2754908771941504e-06,
+      "loss": 3.3937,
+      "step": 5867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2738138859234845e-06,
+      "loss": 3.4279,
+      "step": 5868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.272137330951646e-06,
+      "loss": 3.4503,
+      "step": 5869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2704612125469487e-06,
+      "loss": 3.2704,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2687855309776437e-06,
+      "loss": 3.3091,
+      "step": 5871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2671102865119045e-06,
+      "loss": 3.3667,
+      "step": 5872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2654354794178363e-06,
+      "loss": 3.338,
+      "step": 5873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263761109963476e-06,
+      "loss": 3.3779,
+      "step": 5874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2620871784167895e-06,
+      "loss": 3.4028,
+      "step": 5875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260413685045672e-06,
+      "loss": 3.3309,
+      "step": 5876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.258740630117954e-06,
+      "loss": 3.3574,
+      "step": 5877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2570680139013827e-06,
+      "loss": 3.4461,
+      "step": 5878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2553958366636513e-06,
+      "loss": 3.5275,
+      "step": 5879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.253724098672372e-06,
+      "loss": 3.4655,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2520528001950898e-06,
+      "loss": 3.3132,
+      "step": 5881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2503819414992795e-06,
+      "loss": 3.4462,
+      "step": 5882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2487115228523454e-06,
+      "loss": 3.3846,
+      "step": 5883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247041544521619e-06,
+      "loss": 3.3907,
+      "step": 5884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2453720067743702e-06,
+      "loss": 3.4,
+      "step": 5885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2437029098777833e-06,
+      "loss": 3.3307,
+      "step": 5886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2420342540989888e-06,
+      "loss": 3.4718,
+      "step": 5887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2403660397050303e-06,
+      "loss": 3.442,
+      "step": 5888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2386982669628945e-06,
+      "loss": 3.4002,
+      "step": 5889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2370309361394906e-06,
+      "loss": 3.3952,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.235364047501657e-06,
+      "loss": 3.3132,
+      "step": 5891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2336976013161614e-06,
+      "loss": 3.4145,
+      "step": 5892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.232031597849708e-06,
+      "loss": 3.386,
+      "step": 5893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2303660373689157e-06,
+      "loss": 3.4237,
+      "step": 5894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2287009201403486e-06,
+      "loss": 3.4215,
+      "step": 5895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2270362464304846e-06,
+      "loss": 3.4805,
+      "step": 5896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2253720165057435e-06,
+      "loss": 3.3912,
+      "step": 5897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2237082306324668e-06,
+      "loss": 3.2877,
+      "step": 5898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.222044889076927e-06,
+      "loss": 3.4452,
+      "step": 5899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2203819921053237e-06,
+      "loss": 3.3585,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.218719539983792e-06,
+      "loss": 3.4036,
+      "step": 5901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.217057532978383e-06,
+      "loss": 3.4366,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2153959713550905e-06,
+      "loss": 3.5056,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2137348553798286e-06,
+      "loss": 3.4068,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.212074185318443e-06,
+      "loss": 3.4433,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2104139614367067e-06,
+      "loss": 3.4166,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2087541840003216e-06,
+      "loss": 3.4412,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2070948532749175e-06,
+      "loss": 3.3014,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.205435969526057e-06,
+      "loss": 3.4385,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2037775330192256e-06,
+      "loss": 3.3709,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2021195440198403e-06,
+      "loss": 3.4967,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.200462002793245e-06,
+      "loss": 3.4835,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1988049096047107e-06,
+      "loss": 3.448,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1971482647194446e-06,
+      "loss": 3.4681,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.195492068402569e-06,
+      "loss": 3.4227,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.193836320919146e-06,
+      "loss": 3.4286,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1921810225341605e-06,
+      "loss": 3.4147,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1905261735125265e-06,
+      "loss": 3.4764,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.188871774119086e-06,
+      "loss": 3.3678,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1872178246186083e-06,
+      "loss": 3.4862,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1855643252757896e-06,
+      "loss": 3.3352,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1839112763552628e-06,
+      "loss": 3.4296,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.182258678121574e-06,
+      "loss": 3.3998,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1806065308392094e-06,
+      "loss": 3.5374,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.178954834772578e-06,
+      "loss": 3.5078,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1773035901860173e-06,
+      "loss": 3.3648,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.175652797343792e-06,
+      "loss": 3.4265,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.174002456510095e-06,
+      "loss": 3.3792,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1723525679490442e-06,
+      "loss": 3.3888,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.170703131924695e-06,
+      "loss": 3.3109,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1690541487010153e-06,
+      "loss": 3.3574,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.167405618541913e-06,
+      "loss": 3.3829,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1657575417112177e-06,
+      "loss": 3.4148,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.164109918472687e-06,
+      "loss": 3.4211,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.162462749090008e-06,
+      "loss": 3.4805,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.160816033826792e-06,
+      "loss": 3.4411,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.159169772946579e-06,
+      "loss": 3.4143,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.15752396671284e-06,
+      "loss": 3.4271,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1558786153889647e-06,
+      "loss": 3.4159,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1542337192382793e-06,
+      "loss": 3.3644,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.152589278524032e-06,
+      "loss": 3.3975,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1509452935093983e-06,
+      "loss": 3.3904,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.149301764457482e-06,
+      "loss": 3.3248,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1476586916313135e-06,
+      "loss": 3.3616,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.146016075293848e-06,
+      "loss": 3.4168,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.144373915707974e-06,
+      "loss": 3.4236,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1427322131365007e-06,
+      "loss": 3.3852,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.141090967842166e-06,
+      "loss": 3.4197,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1394501800876346e-06,
+      "loss": 3.484,
+      "step": 5949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1378098501354986e-06,
+      "loss": 3.3861,
+      "step": 5950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1361699782482765e-06,
+      "loss": 3.3874,
+      "step": 5951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.134530564688411e-06,
+      "loss": 3.4406,
+      "step": 5952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1328916097182783e-06,
+      "loss": 3.394,
+      "step": 5953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.131253113600174e-06,
+      "loss": 3.3174,
+      "step": 5954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1296150765963234e-06,
+      "loss": 3.443,
+      "step": 5955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1279774989688787e-06,
+      "loss": 3.4938,
+      "step": 5956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1263403809799166e-06,
+      "loss": 3.3376,
+      "step": 5957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1247037228914407e-06,
+      "loss": 3.4056,
+      "step": 5958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1230675249653863e-06,
+      "loss": 3.4071,
+      "step": 5959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1214317874636044e-06,
+      "loss": 3.4552,
+      "step": 5960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1197965106478822e-06,
+      "loss": 3.33,
+      "step": 5961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.118161694779929e-06,
+      "loss": 3.4963,
+      "step": 5962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.11652734012138e-06,
+      "loss": 3.4879,
+      "step": 5963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.114893446933798e-06,
+      "loss": 3.3665,
+      "step": 5964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1132600154786702e-06,
+      "loss": 3.3622,
+      "step": 5965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1116270460174102e-06,
+      "loss": 3.4405,
+      "step": 5966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.109994538811364e-06,
+      "loss": 3.3824,
+      "step": 5967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.10836249412179e-06,
+      "loss": 3.3758,
+      "step": 5968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1067309122098863e-06,
+      "loss": 3.3345,
+      "step": 5969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1050997933367697e-06,
+      "loss": 3.4452,
+      "step": 5970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1034691377634838e-06,
+      "loss": 3.3466,
+      "step": 5971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.101838945750999e-06,
+      "loss": 3.4448,
+      "step": 5972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1002092175602125e-06,
+      "loss": 3.3838,
+      "step": 5973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0985799534519426e-06,
+      "loss": 3.4009,
+      "step": 5974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0969511536869426e-06,
+      "loss": 3.3327,
+      "step": 5975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0953228185258783e-06,
+      "loss": 3.3595,
+      "step": 5976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0936949482293543e-06,
+      "loss": 3.4386,
+      "step": 5977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.092067543057893e-06,
+      "loss": 3.3709,
+      "step": 5978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0904406032719437e-06,
+      "loss": 3.4113,
+      "step": 5979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.088814129131883e-06,
+      "loss": 3.3359,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.087188120898011e-06,
+      "loss": 3.3001,
+      "step": 5981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.085562578830552e-06,
+      "loss": 3.3608,
+      "step": 5982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0839375031896645e-06,
+      "loss": 3.4735,
+      "step": 5983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.082312894235417e-06,
+      "loss": 3.4513,
+      "step": 5984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.080688752227818e-06,
+      "loss": 3.3645,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0790650774267936e-06,
+      "loss": 3.4347,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0774418700921956e-06,
+      "loss": 3.4413,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0758191304838034e-06,
+      "loss": 3.3866,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0741968588613192e-06,
+      "loss": 3.5174,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0725750554843704e-06,
+      "loss": 3.4333,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0709537206125137e-06,
+      "loss": 3.3906,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0693328545052253e-06,
+      "loss": 3.3453,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.067712457421909e-06,
+      "loss": 3.4465,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0660925296218936e-06,
+      "loss": 3.5013,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.064473071364432e-06,
+      "loss": 3.4851,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.062854082908702e-06,
+      "loss": 3.3521,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0612355645138044e-06,
+      "loss": 3.5141,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0596175164387717e-06,
+      "loss": 3.4126,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0579999389425537e-06,
+      "loss": 3.443,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0563828322840274e-06,
+      "loss": 3.392,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0547661967219944e-06,
+      "loss": 3.4838,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.053150032515182e-06,
+      "loss": 3.3557,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.051534339922239e-06,
+      "loss": 3.3053,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0499191192017455e-06,
+      "loss": 3.4389,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0483043706121947e-06,
+      "loss": 3.4111,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0466900944120164e-06,
+      "loss": 3.4085,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.045076290859558e-06,
+      "loss": 3.441,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0434629602130918e-06,
+      "loss": 3.4888,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.041850102730814e-06,
+      "loss": 3.4081,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0402377186708516e-06,
+      "loss": 3.4132,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.038625808291243e-06,
+      "loss": 3.4757,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0370143718499657e-06,
+      "loss": 3.3887,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0354034096049073e-06,
+      "loss": 3.5163,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0337929218138906e-06,
+      "loss": 3.3386,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.032182908734657e-06,
+      "loss": 3.3852,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.030573370624874e-06,
+      "loss": 3.3523,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0289643077421294e-06,
+      "loss": 3.4324,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0273557203439396e-06,
+      "loss": 3.4984,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0257476086877415e-06,
+      "loss": 3.4565,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0241399730309013e-06,
+      "loss": 3.3975,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0225328136306982e-06,
+      "loss": 3.2984,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0209261307443477e-06,
+      "loss": 3.3878,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0193199246289814e-06,
+      "loss": 3.4333,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.017714195541657e-06,
+      "loss": 3.4454,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0161089437393543e-06,
+      "loss": 3.4025,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0145041694789786e-06,
+      "loss": 3.3961,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.012899873017356e-06,
+      "loss": 3.3906,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0112960546112435e-06,
+      "loss": 3.4058,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.009692714517309e-06,
+      "loss": 3.4647,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0080898529921567e-06,
+      "loss": 3.4494,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0064874702923067e-06,
+      "loss": 3.4042,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0048855666742035e-06,
+      "loss": 3.3324,
+      "step": 6032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0032841423942166e-06,
+      "loss": 3.3441,
+      "step": 6033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0016831977086377e-06,
+      "loss": 3.4434,
+      "step": 6034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0000827328736803e-06,
+      "loss": 3.4181,
+      "step": 6035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9984827481454865e-06,
+      "loss": 3.3976,
+      "step": 6036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9968832437801154e-06,
+      "loss": 3.3772,
+      "step": 6037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.995284220033553e-06,
+      "loss": 3.4474,
+      "step": 6038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.993685677161706e-06,
+      "loss": 3.4306,
+      "step": 6039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.992087615420403e-06,
+      "loss": 3.3488,
+      "step": 6040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9904900350654043e-06,
+      "loss": 3.332,
+      "step": 6041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988892936352379e-06,
+      "loss": 3.4941,
+      "step": 6042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9872963195369315e-06,
+      "loss": 3.3787,
+      "step": 6043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9857001848745837e-06,
+      "loss": 3.4521,
+      "step": 6044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.984104532620781e-06,
+      "loss": 3.4099,
+      "step": 6045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982509363030889e-06,
+      "loss": 3.3685,
+      "step": 6046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9809146763602044e-06,
+      "loss": 3.3523,
+      "step": 6047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979320472863933e-06,
+      "loss": 3.4096,
+      "step": 6048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9777267527972196e-06,
+      "loss": 3.4571,
+      "step": 6049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9761335164151152e-06,
+      "loss": 3.5029,
+      "step": 6050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9745407639726065e-06,
+      "loss": 3.3935,
+      "step": 6051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9729484957245965e-06,
+      "loss": 3.4544,
+      "step": 6052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9713567119259113e-06,
+      "loss": 3.4088,
+      "step": 6053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.969765412831298e-06,
+      "loss": 3.4708,
+      "step": 6054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.968174598695435e-06,
+      "loss": 3.3887,
+      "step": 6055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9665842697729065e-06,
+      "loss": 3.3661,
+      "step": 6056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.964994426318238e-06,
+      "loss": 3.5221,
+      "step": 6057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9634050685858595e-06,
+      "loss": 3.4236,
+      "step": 6058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9618161968301388e-06,
+      "loss": 3.4459,
+      "step": 6059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.960227811305356e-06,
+      "loss": 3.4383,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9586399122657156e-06,
+      "loss": 3.4816,
+      "step": 6061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9570524999653453e-06,
+      "loss": 3.3374,
+      "step": 6062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.955465574658298e-06,
+      "loss": 3.3575,
+      "step": 6063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9538791365985386e-06,
+      "loss": 3.5255,
+      "step": 6064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.952293186039968e-06,
+      "loss": 3.399,
+      "step": 6065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9507077232363954e-06,
+      "loss": 3.4023,
+      "step": 6066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.949122748441562e-06,
+      "loss": 3.4599,
+      "step": 6067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.947538261909126e-06,
+      "loss": 3.4256,
+      "step": 6068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9459542638926693e-06,
+      "loss": 3.3792,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9443707546456923e-06,
+      "loss": 3.4139,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9427877344216264e-06,
+      "loss": 3.3848,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9412052034738087e-06,
+      "loss": 3.384,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.939623162055516e-06,
+      "loss": 3.314,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9380416104199334e-06,
+      "loss": 3.4047,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9364605488201736e-06,
+      "loss": 3.3796,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93487997750927e-06,
+      "loss": 3.4501,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.933299896740178e-06,
+      "loss": 3.3475,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93172030676577e-06,
+      "loss": 3.3587,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301412078388485e-06,
+      "loss": 3.3697,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9285626002121304e-06,
+      "loss": 3.4661,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9269844841382557e-06,
+      "loss": 3.4513,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9254068598697866e-06,
+      "loss": 3.4066,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.923829727659206e-06,
+      "loss": 3.4058,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9222530877589178e-06,
+      "loss": 3.4122,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.920676940421246e-06,
+      "loss": 3.4596,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9191012858984426e-06,
+      "loss": 3.3307,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.917526124442668e-06,
+      "loss": 3.4573,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.915951456306016e-06,
+      "loss": 3.3786,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.914377281740496e-06,
+      "loss": 3.4484,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.912803600998038e-06,
+      "loss": 3.3008,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9112304143304915e-06,
+      "loss": 3.3844,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9096577219896356e-06,
+      "loss": 3.4274,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.908085524227157e-06,
+      "loss": 3.4442,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9065138212946765e-06,
+      "loss": 3.3839,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9049426134437227e-06,
+      "loss": 3.4059,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9033719009257572e-06,
+      "loss": 3.4274,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.901801683992155e-06,
+      "loss": 3.3993,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9002319628942135e-06,
+      "loss": 3.4586,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8986627378831491e-06,
+      "loss": 3.4749,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8970940092101059e-06,
+      "loss": 3.4628,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8955257771261365e-06,
+      "loss": 3.4107,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8939580418822273e-06,
+      "loss": 3.4434,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8923908037292721e-06,
+      "loss": 3.4059,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8908240629180963e-06,
+      "loss": 3.3138,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8892578196994398e-06,
+      "loss": 3.445,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8876920743239647e-06,
+      "loss": 3.3243,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.88612682704225e-06,
+      "loss": 3.4027,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8845620781048036e-06,
+      "loss": 3.4434,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8829978277620409e-06,
+      "loss": 3.3963,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8814340762643118e-06,
+      "loss": 3.3542,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8798708238618725e-06,
+      "loss": 3.2475,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8783080708049101e-06,
+      "loss": 3.4079,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8767458173435261e-06,
+      "loss": 3.4476,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.875184063727744e-06,
+      "loss": 3.4401,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.873622810207505e-06,
+      "loss": 3.3321,
+      "step": 6115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.872062057032677e-06,
+      "loss": 3.4938,
+      "step": 6116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8705018044530365e-06,
+      "loss": 3.358,
+      "step": 6117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.868942052718291e-06,
+      "loss": 3.3962,
+      "step": 6118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8673828020780615e-06,
+      "loss": 3.4149,
+      "step": 6119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.865824052781891e-06,
+      "loss": 3.381,
+      "step": 6120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8642658050792412e-06,
+      "loss": 3.4243,
+      "step": 6121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862708059219494e-06,
+      "loss": 3.4302,
+      "step": 6122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8611508154519504e-06,
+      "loss": 3.4564,
+      "step": 6123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8595940740258333e-06,
+      "loss": 3.427,
+      "step": 6124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8580378351902834e-06,
+      "loss": 3.4259,
+      "step": 6125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8564820991943605e-06,
+      "loss": 3.5106,
+      "step": 6126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8549268662870451e-06,
+      "loss": 3.4676,
+      "step": 6127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.853372136717234e-06,
+      "loss": 3.4148,
+      "step": 6128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8518179107337524e-06,
+      "loss": 3.383,
+      "step": 6129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850264188585331e-06,
+      "loss": 3.3823,
+      "step": 6130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8487109705206336e-06,
+      "loss": 3.4542,
+      "step": 6131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.847158256788234e-06,
+      "loss": 3.413,
+      "step": 6132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8456060476366295e-06,
+      "loss": 3.5033,
+      "step": 6133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.844054343314235e-06,
+      "loss": 3.4896,
+      "step": 6134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8425031440693863e-06,
+      "loss": 3.3869,
+      "step": 6135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8409524501503345e-06,
+      "loss": 3.4701,
+      "step": 6136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8394022618052587e-06,
+      "loss": 3.476,
+      "step": 6137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8378525792822437e-06,
+      "loss": 3.3428,
+      "step": 6138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8363034028293058e-06,
+      "loss": 3.4677,
+      "step": 6139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8347547326943744e-06,
+      "loss": 3.3895,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.833206569125297e-06,
+      "loss": 3.3434,
+      "step": 6141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.831658912369843e-06,
+      "loss": 3.3502,
+      "step": 6142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8301117626756987e-06,
+      "loss": 3.3889,
+      "step": 6143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8285651202904686e-06,
+      "loss": 3.4749,
+      "step": 6144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8270189854616827e-06,
+      "loss": 3.3809,
+      "step": 6145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8254733584367768e-06,
+      "loss": 3.5387,
+      "step": 6146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8239282394631202e-06,
+      "loss": 3.5078,
+      "step": 6147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8223836287879876e-06,
+      "loss": 3.4563,
+      "step": 6148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8208395266585832e-06,
+      "loss": 3.3716,
+      "step": 6149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8192959333220227e-06,
+      "loss": 3.4117,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8177528490253438e-06,
+      "loss": 3.3922,
+      "step": 6151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8162102740154997e-06,
+      "loss": 3.4623,
+      "step": 6152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8146682085393686e-06,
+      "loss": 3.561,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8131266528437363e-06,
+      "loss": 3.3402,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8115856071753207e-06,
+      "loss": 3.4327,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8100450717807432e-06,
+      "loss": 3.3896,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8085050469065563e-06,
+      "loss": 3.4052,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8069655327992242e-06,
+      "loss": 3.4014,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8054265297051305e-06,
+      "loss": 3.3937,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.803888037870576e-06,
+      "loss": 3.4108,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8023500575417835e-06,
+      "loss": 3.303,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8008125889648896e-06,
+      "loss": 3.3438,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7992756323859522e-06,
+      "loss": 3.3545,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7977391880509448e-06,
+      "loss": 3.3712,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7962032562057602e-06,
+      "loss": 3.4163,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7946678370962084e-06,
+      "loss": 3.4012,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7931329309680168e-06,
+      "loss": 3.4694,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7915985380668348e-06,
+      "loss": 3.3978,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7900646586382253e-06,
+      "loss": 3.4012,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7885312929276704e-06,
+      "loss": 3.2952,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7869984411805692e-06,
+      "loss": 3.4421,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.78546610364224e-06,
+      "loss": 3.4806,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7839342805579163e-06,
+      "loss": 3.3762,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7824029721727565e-06,
+      "loss": 3.4365,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7808721787318234e-06,
+      "loss": 3.4777,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.779341900480111e-06,
+      "loss": 3.3942,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7778121376625234e-06,
+      "loss": 3.4263,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7762828905238833e-06,
+      "loss": 3.3632,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7747541593089324e-06,
+      "loss": 3.3995,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7732259442623273e-06,
+      "loss": 3.3219,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7716982456286435e-06,
+      "loss": 3.4055,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7701710636523778e-06,
+      "loss": 3.3537,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7686443985779344e-06,
+      "loss": 3.3735,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671182506496454e-06,
+      "loss": 3.3361,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7655926201117535e-06,
+      "loss": 3.3891,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640675072084213e-06,
+      "loss": 3.3931,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625429121837279e-06,
+      "loss": 3.4209,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610188352816687e-06,
+      "loss": 3.3161,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7594952767461565e-06,
+      "loss": 3.4257,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7579722368210257e-06,
+      "loss": 3.407,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7564497157500166e-06,
+      "loss": 3.3883,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7549277137767995e-06,
+      "loss": 3.3155,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7534062311449535e-06,
+      "loss": 3.3822,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7518852680979763e-06,
+      "loss": 3.3826,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7503648248792832e-06,
+      "loss": 3.3785,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7488449017322062e-06,
+      "loss": 3.4863,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7473254988999915e-06,
+      "loss": 3.3881,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7458066166258097e-06,
+      "loss": 3.4747,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7442882551527362e-06,
+      "loss": 3.3361,
+      "step": 6199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7427704147237745e-06,
+      "loss": 3.4146,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.741253095581838e-06,
+      "loss": 3.3968,
+      "step": 6201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7397362979697586e-06,
+      "loss": 3.432,
+      "step": 6202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7382200221302842e-06,
+      "loss": 3.4439,
+      "step": 6203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7367042683060796e-06,
+      "loss": 3.3551,
+      "step": 6204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.735189036739725e-06,
+      "loss": 3.3959,
+      "step": 6205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7336743276737206e-06,
+      "loss": 3.4115,
+      "step": 6206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7321601413504796e-06,
+      "loss": 3.465,
+      "step": 6207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7306464780123322e-06,
+      "loss": 3.3814,
+      "step": 6208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729133337901524e-06,
+      "loss": 3.4503,
+      "step": 6209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7276207212602198e-06,
+      "loss": 3.3997,
+      "step": 6210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7261086283304968e-06,
+      "loss": 3.4613,
+      "step": 6211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7245970593543503e-06,
+      "loss": 3.4916,
+      "step": 6212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7230860145736943e-06,
+      "loss": 3.2642,
+      "step": 6213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7215754942303548e-06,
+      "loss": 3.5145,
+      "step": 6214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7200654985660758e-06,
+      "loss": 3.4227,
+      "step": 6215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7185560278225166e-06,
+      "loss": 3.36,
+      "step": 6216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7170470822412533e-06,
+      "loss": 3.4501,
+      "step": 6217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7155386620637758e-06,
+      "loss": 3.4691,
+      "step": 6218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.714030767531496e-06,
+      "loss": 3.4903,
+      "step": 6219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.712523398885732e-06,
+      "loss": 3.419,
+      "step": 6220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7110165563677266e-06,
+      "loss": 3.4387,
+      "step": 6221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7095102402186342e-06,
+      "loss": 3.3725,
+      "step": 6222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7080044506795257e-06,
+      "loss": 3.4695,
+      "step": 6223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7064991879913873e-06,
+      "loss": 3.3696,
+      "step": 6224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7049944523951207e-06,
+      "loss": 3.3853,
+      "step": 6225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7034902441315432e-06,
+      "loss": 3.4344,
+      "step": 6226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7019865634413923e-06,
+      "loss": 3.4043,
+      "step": 6227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.700483410565311e-06,
+      "loss": 3.432,
+      "step": 6228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.698980785743869e-06,
+      "loss": 3.375,
+      "step": 6229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6974786892175443e-06,
+      "loss": 3.4637,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6959771212267328e-06,
+      "loss": 3.5046,
+      "step": 6231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.694476082011745e-06,
+      "loss": 3.2886,
+      "step": 6232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6929755718128077e-06,
+      "loss": 3.46,
+      "step": 6233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6914755908700604e-06,
+      "loss": 3.394,
+      "step": 6234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6899761394235653e-06,
+      "loss": 3.3929,
+      "step": 6235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.688477217713288e-06,
+      "loss": 3.4139,
+      "step": 6236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6869788259791208e-06,
+      "loss": 3.4337,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6854809644608645e-06,
+      "loss": 3.4698,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6839836333982373e-06,
+      "loss": 3.4474,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.682486833030872e-06,
+      "loss": 3.4848,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6809905635983164e-06,
+      "loss": 3.4057,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6794948253400318e-06,
+      "loss": 3.4551,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6779996184954005e-06,
+      "loss": 3.5017,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.67650494330371e-06,
+      "loss": 3.4326,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6750108000041721e-06,
+      "loss": 3.4995,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.673517188835908e-06,
+      "loss": 3.4304,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6720241100379558e-06,
+      "loss": 3.3703,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6705315638492676e-06,
+      "loss": 3.4227,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6690395505087103e-06,
+      "loss": 3.4217,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6675480702550639e-06,
+      "loss": 3.3254,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6660571233270284e-06,
+      "loss": 3.3908,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6645667099632134e-06,
+      "loss": 3.3966,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.663076830402145e-06,
+      "loss": 3.3578,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6615874848822628e-06,
+      "loss": 3.4499,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.66009867364192e-06,
+      "loss": 3.437,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6586103969193917e-06,
+      "loss": 3.4369,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.657122654952854e-06,
+      "loss": 3.4012,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.655635447980411e-06,
+      "loss": 3.5131,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6541487762400738e-06,
+      "loss": 3.4736,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6526626399697682e-06,
+      "loss": 3.3016,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6511770394073352e-06,
+      "loss": 3.3658,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6496919747905337e-06,
+      "loss": 3.3435,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6482074463570285e-06,
+      "loss": 3.4018,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6467234543444089e-06,
+      "loss": 3.4593,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6452399989901669e-06,
+      "loss": 3.402,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6437570805317198e-06,
+      "loss": 3.4463,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6422746992063914e-06,
+      "loss": 3.4222,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6407928552514236e-06,
+      "loss": 3.472,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6393115489039668e-06,
+      "loss": 3.3732,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6378307804010962e-06,
+      "loss": 3.3866,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6363505499797865e-06,
+      "loss": 3.3862,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6348708578769407e-06,
+      "loss": 3.3242,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6333917043293623e-06,
+      "loss": 3.4118,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.631913089573779e-06,
+      "loss": 3.3481,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6304350138468277e-06,
+      "loss": 3.3645,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.628957477385059e-06,
+      "loss": 3.3622,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6274804804249384e-06,
+      "loss": 3.4279,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6260040232028435e-06,
+      "loss": 3.4137,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6245281059550661e-06,
+      "loss": 3.4216,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6230527289178161e-06,
+      "loss": 3.4243,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6215778923272068e-06,
+      "loss": 3.31,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6201035964192757e-06,
+      "loss": 3.4418,
+      "step": 6282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6186298414299673e-06,
+      "loss": 3.4498,
+      "step": 6283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6171566275951417e-06,
+      "loss": 3.4076,
+      "step": 6284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6156839551505721e-06,
+      "loss": 3.345,
+      "step": 6285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6142118243319444e-06,
+      "loss": 3.3766,
+      "step": 6286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6127402353748578e-06,
+      "loss": 3.3858,
+      "step": 6287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6112691885148274e-06,
+      "loss": 3.4181,
+      "step": 6288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6097986839872787e-06,
+      "loss": 3.4026,
+      "step": 6289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6083287220275511e-06,
+      "loss": 3.3736,
+      "step": 6290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6068593028708972e-06,
+      "loss": 3.4217,
+      "step": 6291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6053904267524834e-06,
+      "loss": 3.3971,
+      "step": 6292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6039220939073879e-06,
+      "loss": 3.516,
+      "step": 6293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6024543045706016e-06,
+      "loss": 3.3882,
+      "step": 6294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600987058977032e-06,
+      "loss": 3.4582,
+      "step": 6295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5995203573614958e-06,
+      "loss": 3.3604,
+      "step": 6296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5980541999587246e-06,
+      "loss": 3.4197,
+      "step": 6297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.596588587003362e-06,
+      "loss": 3.3487,
+      "step": 6298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595123518729964e-06,
+      "loss": 3.4426,
+      "step": 6299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5936589953729991e-06,
+      "loss": 3.3963,
+      "step": 6300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5921950171668539e-06,
+      "loss": 3.3289,
+      "step": 6301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5907315843458176e-06,
+      "loss": 3.4451,
+      "step": 6302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.589268697144103e-06,
+      "loss": 3.4833,
+      "step": 6303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5878063557958285e-06,
+      "loss": 3.3762,
+      "step": 6304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586344560535027e-06,
+      "loss": 3.3638,
+      "step": 6305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584883311595643e-06,
+      "loss": 3.3518,
+      "step": 6306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5834226092115396e-06,
+      "loss": 3.4181,
+      "step": 6307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5819624536164813e-06,
+      "loss": 3.4204,
+      "step": 6308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5805028450441573e-06,
+      "loss": 3.3629,
+      "step": 6309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5790437837281575e-06,
+      "loss": 3.3787,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5775852699019946e-06,
+      "loss": 3.3444,
+      "step": 6311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5761273037990876e-06,
+      "loss": 3.4501,
+      "step": 6312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5746698856527692e-06,
+      "loss": 3.3775,
+      "step": 6313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5732130156962839e-06,
+      "loss": 3.3601,
+      "step": 6314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5717566941627927e-06,
+      "loss": 3.3614,
+      "step": 6315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5703009212853593e-06,
+      "loss": 3.291,
+      "step": 6316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5688456972969723e-06,
+      "loss": 3.3838,
+      "step": 6317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5673910224305194e-06,
+      "loss": 3.5239,
+      "step": 6318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5659368969188115e-06,
+      "loss": 3.4311,
+      "step": 6319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5644833209945648e-06,
+      "loss": 3.3829,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5630302948904102e-06,
+      "loss": 3.3649,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.561577818838888e-06,
+      "loss": 3.4669,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5601258930724572e-06,
+      "loss": 3.4025,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.558674517823478e-06,
+      "loss": 3.3958,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5572236933242346e-06,
+      "loss": 3.3649,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5557734198069107e-06,
+      "loss": 3.2997,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.554323697503613e-06,
+      "loss": 3.3899,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.552874526646353e-06,
+      "loss": 3.3007,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5514259074670562e-06,
+      "loss": 3.2949,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5499778401975584e-06,
+      "loss": 3.3765,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5485303250696126e-06,
+      "loss": 3.2951,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5470833623148728e-06,
+      "loss": 3.3779,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5456369521649161e-06,
+      "loss": 3.391,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.544191094851224e-06,
+      "loss": 3.3492,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5427457906051912e-06,
+      "loss": 3.329,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5413010396581251e-06,
+      "loss": 3.3799,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.539856842241243e-06,
+      "loss": 3.3212,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5384131985856726e-06,
+      "loss": 3.4231,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5369701089224587e-06,
+      "loss": 3.5129,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5355275734825504e-06,
+      "loss": 3.4533,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5340855924968118e-06,
+      "loss": 3.446,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5326441661960179e-06,
+      "loss": 3.4603,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.531203294810854e-06,
+      "loss": 3.473,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.529762978571917e-06,
+      "loss": 3.4341,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5283232177097146e-06,
+      "loss": 3.436,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5268840124546697e-06,
+      "loss": 3.5289,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.525445363037107e-06,
+      "loss": 3.3587,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.524007269687272e-06,
+      "loss": 3.4292,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5225697326353168e-06,
+      "loss": 3.4307,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5211327521113035e-06,
+      "loss": 3.4495,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.519696328345206e-06,
+      "loss": 3.4115,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5182604615669134e-06,
+      "loss": 3.3568,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.516825152006216e-06,
+      "loss": 3.401,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.515390399892827e-06,
+      "loss": 3.2935,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5139562054563584e-06,
+      "loss": 3.4579,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5125225689263424e-06,
+      "loss": 3.3753,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5110894905322164e-06,
+      "loss": 3.3654,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.509656970503332e-06,
+      "loss": 3.3967,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.508225009068946e-06,
+      "loss": 3.3738,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5067936064582362e-06,
+      "loss": 3.3935,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5053627629002764e-06,
+      "loss": 3.4074,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5039324786240655e-06,
+      "loss": 3.5039,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5025027538585007e-06,
+      "loss": 3.3312,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.501073588832399e-06,
+      "loss": 3.3891,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4996449837744831e-06,
+      "loss": 3.369,
+      "step": 6365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.498216938913386e-06,
+      "loss": 3.2923,
+      "step": 6366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4967894544776513e-06,
+      "loss": 3.4318,
+      "step": 6367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4953625306957382e-06,
+      "loss": 3.3449,
+      "step": 6368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4939361677960045e-06,
+      "loss": 3.4493,
+      "step": 6369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4925103660067324e-06,
+      "loss": 3.518,
+      "step": 6370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.491085125556101e-06,
+      "loss": 3.459,
+      "step": 6371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4896604466722104e-06,
+      "loss": 3.4377,
+      "step": 6372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4882363295830642e-06,
+      "loss": 3.3639,
+      "step": 6373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4868127745165784e-06,
+      "loss": 3.4636,
+      "step": 6374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4853897817005768e-06,
+      "loss": 3.424,
+      "step": 6375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4839673513627994e-06,
+      "loss": 3.4538,
+      "step": 6376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4825454837308894e-06,
+      "loss": 3.4244,
+      "step": 6377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4811241790324026e-06,
+      "loss": 3.3537,
+      "step": 6378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.479703437494805e-06,
+      "loss": 3.3505,
+      "step": 6379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4782832593454715e-06,
+      "loss": 3.2869,
+      "step": 6380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.476863644811688e-06,
+      "loss": 3.5014,
+      "step": 6381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.475444594120648e-06,
+      "loss": 3.3949,
+      "step": 6382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4740261074994588e-06,
+      "loss": 3.2444,
+      "step": 6383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4726081851751338e-06,
+      "loss": 3.4396,
+      "step": 6384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4711908273745978e-06,
+      "loss": 3.3739,
+      "step": 6385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4697740343246846e-06,
+      "loss": 3.328,
+      "step": 6386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4683578062521376e-06,
+      "loss": 3.4126,
+      "step": 6387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4669421433836079e-06,
+      "loss": 3.4204,
+      "step": 6388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4655270459456633e-06,
+      "loss": 3.3435,
+      "step": 6389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4641125141647695e-06,
+      "loss": 3.4024,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4626985482673134e-06,
+      "loss": 3.3968,
+      "step": 6391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4612851484795843e-06,
+      "loss": 3.4154,
+      "step": 6392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4598723150277827e-06,
+      "loss": 3.4507,
+      "step": 6393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4584600481380179e-06,
+      "loss": 3.3536,
+      "step": 6394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4570483480363102e-06,
+      "loss": 3.3684,
+      "step": 6395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4556372149485854e-06,
+      "loss": 3.4855,
+      "step": 6396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4542266491006868e-06,
+      "loss": 3.3354,
+      "step": 6397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4528166507183544e-06,
+      "loss": 3.342,
+      "step": 6398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.45140722002725e-06,
+      "loss": 3.2823,
+      "step": 6399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4499983572529359e-06,
+      "loss": 3.3267,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4485900626208876e-06,
+      "loss": 3.4202,
+      "step": 6401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4471823363564886e-06,
+      "loss": 3.3377,
+      "step": 6402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4457751786850304e-06,
+      "loss": 3.3629,
+      "step": 6403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4443685898317145e-06,
+      "loss": 3.3782,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4429625700216548e-06,
+      "loss": 3.5173,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.441557119479865e-06,
+      "loss": 3.4698,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4401522384312795e-06,
+      "loss": 3.3052,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4387479271007287e-06,
+      "loss": 3.5359,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4373441857129644e-06,
+      "loss": 3.4255,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4359410144926388e-06,
+      "loss": 3.33,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4345384136643164e-06,
+      "loss": 3.3461,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.433136383452467e-06,
+      "loss": 3.3579,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4317349240814766e-06,
+      "loss": 3.359,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4303340357756285e-06,
+      "loss": 3.5336,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.428933718759125e-06,
+      "loss": 3.3689,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4275339732560722e-06,
+      "loss": 3.4045,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4261347994904851e-06,
+      "loss": 3.3854,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4247361976862866e-06,
+      "loss": 3.4061,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4233381680673103e-06,
+      "loss": 3.382,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4219407108572947e-06,
+      "loss": 3.3425,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4205438262798916e-06,
+      "loss": 3.4267,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4191475145586569e-06,
+      "loss": 3.366,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4177517759170572e-06,
+      "loss": 3.3225,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4163566105784653e-06,
+      "loss": 3.4102,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4149620187661645e-06,
+      "loss": 3.4773,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4135680007033447e-06,
+      "loss": 3.4474,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4121745566131034e-06,
+      "loss": 3.4652,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.41078168671845e-06,
+      "loss": 3.3754,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409389391242298e-06,
+      "loss": 3.388,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4079976704074704e-06,
+      "loss": 3.4859,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4066065244366984e-06,
+      "loss": 3.2696,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.405215953552621e-06,
+      "loss": 3.344,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.403825957977783e-06,
+      "loss": 3.4253,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.402436537934645e-06,
+      "loss": 3.361,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4010476936455625e-06,
+      "loss": 3.3963,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3996594253328115e-06,
+      "loss": 3.4256,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.398271733218569e-06,
+      "loss": 3.4399,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3968846175249206e-06,
+      "loss": 3.337,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.395498078473861e-06,
+      "loss": 3.3773,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3941121162872912e-06,
+      "loss": 3.3582,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.392726731187019e-06,
+      "loss": 3.3454,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3913419233947672e-06,
+      "loss": 3.4487,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3899576931321534e-06,
+      "loss": 3.4022,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3885740406207144e-06,
+      "loss": 3.3408,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3871909660818883e-06,
+      "loss": 3.3983,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3858084697370233e-06,
+      "loss": 3.5411,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3844265518073729e-06,
+      "loss": 3.4747,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3830452125140997e-06,
+      "loss": 3.3787,
+      "step": 6448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3816644520782725e-06,
+      "loss": 3.381,
+      "step": 6449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3802842707208713e-06,
+      "loss": 3.3638,
+      "step": 6450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3789046686627756e-06,
+      "loss": 3.3899,
+      "step": 6451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.377525646124781e-06,
+      "loss": 3.2876,
+      "step": 6452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3761472033275841e-06,
+      "loss": 3.3705,
+      "step": 6453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3747693404917917e-06,
+      "loss": 3.3959,
+      "step": 6454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3733920578379167e-06,
+      "loss": 3.403,
+      "step": 6455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3720153555863792e-06,
+      "loss": 3.4397,
+      "step": 6456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.370639233957506e-06,
+      "loss": 3.5059,
+      "step": 6457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3692636931715348e-06,
+      "loss": 3.4856,
+      "step": 6458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.367888733448602e-06,
+      "loss": 3.4467,
+      "step": 6459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.36651435500876e-06,
+      "loss": 3.3702,
+      "step": 6460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.365140558071964e-06,
+      "loss": 3.4189,
+      "step": 6461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3637673428580756e-06,
+      "loss": 3.3411,
+      "step": 6462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3623947095868639e-06,
+      "loss": 3.3648,
+      "step": 6463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.361022658478006e-06,
+      "loss": 3.4028,
+      "step": 6464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3596511897510827e-06,
+      "loss": 3.3824,
+      "step": 6465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.358280303625587e-06,
+      "loss": 3.417,
+      "step": 6466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3569100003209147e-06,
+      "loss": 3.3572,
+      "step": 6467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3555402800563683e-06,
+      "loss": 3.4186,
+      "step": 6468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3541711430511584e-06,
+      "loss": 3.3928,
+      "step": 6469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3528025895243995e-06,
+      "loss": 3.4121,
+      "step": 6470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.35143461969512e-06,
+      "loss": 3.4627,
+      "step": 6471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350067233782244e-06,
+      "loss": 3.3651,
+      "step": 6472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3487004320046114e-06,
+      "loss": 3.446,
+      "step": 6473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3473342145809649e-06,
+      "loss": 3.3477,
+      "step": 6474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3459685817299535e-06,
+      "loss": 3.3339,
+      "step": 6475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3446035336701325e-06,
+      "loss": 3.3598,
+      "step": 6476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3432390706199643e-06,
+      "loss": 3.3817,
+      "step": 6477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3418751927978168e-06,
+      "loss": 3.3822,
+      "step": 6478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3405119004219697e-06,
+      "loss": 3.383,
+      "step": 6479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3391491937105972e-06,
+      "loss": 3.4609,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3377870728817916e-06,
+      "loss": 3.4412,
+      "step": 6481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3364255381535462e-06,
+      "loss": 3.4205,
+      "step": 6482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3350645897437598e-06,
+      "loss": 3.3308,
+      "step": 6483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3337042278702396e-06,
+      "loss": 3.3505,
+      "step": 6484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3323444527506974e-06,
+      "loss": 3.4098,
+      "step": 6485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3309852646027499e-06,
+      "loss": 3.3894,
+      "step": 6486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3296266636439265e-06,
+      "loss": 3.3967,
+      "step": 6487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3282686500916513e-06,
+      "loss": 3.5477,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3269112241632654e-06,
+      "loss": 3.3076,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3255543860760106e-06,
+      "loss": 3.2986,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.324198136047034e-06,
+      "loss": 3.306,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3228424742933904e-06,
+      "loss": 3.5014,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.32148740103204e-06,
+      "loss": 3.3867,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3201329164798472e-06,
+      "loss": 3.4565,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3187790208535878e-06,
+      "loss": 3.5365,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3174257143699342e-06,
+      "loss": 3.4632,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3160729972454729e-06,
+      "loss": 3.4173,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3147208696966923e-06,
+      "loss": 3.3959,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.313369331939986e-06,
+      "loss": 3.4403,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3120183841916544e-06,
+      "loss": 3.4765,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.310668026667904e-06,
+      "loss": 3.3919,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.309318259584843e-06,
+      "loss": 3.4006,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3079690831584923e-06,
+      "loss": 3.438,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3066204976047725e-06,
+      "loss": 3.3095,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3052725031395103e-06,
+      "loss": 3.3905,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3039250999784397e-06,
+      "loss": 3.4497,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.302578288337199e-06,
+      "loss": 3.4142,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3012320684313318e-06,
+      "loss": 3.4714,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2998864404762855e-06,
+      "loss": 3.516,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2985414046874172e-06,
+      "loss": 3.3847,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2971969612799856e-06,
+      "loss": 3.3223,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2958531104691547e-06,
+      "loss": 3.3276,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.294509852469995e-06,
+      "loss": 3.4448,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2931671874974811e-06,
+      "loss": 3.3859,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2918251157664913e-06,
+      "loss": 3.507,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2904836374918161e-06,
+      "loss": 3.5502,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2891427528881396e-06,
+      "loss": 3.2946,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2878024621700603e-06,
+      "loss": 3.4782,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2864627655520783e-06,
+      "loss": 3.3539,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.285123663248598e-06,
+      "loss": 3.4442,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2837851554739271e-06,
+      "loss": 3.4008,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2824472424422868e-06,
+      "loss": 3.4384,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2811099243677888e-06,
+      "loss": 3.4493,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2797732014644643e-06,
+      "loss": 3.3786,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2784370739462364e-06,
+      "loss": 3.3825,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2771015420269434e-06,
+      "loss": 3.4773,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2757666059203227e-06,
+      "loss": 3.3501,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2744322658400166e-06,
+      "loss": 3.3586,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2730985219995723e-06,
+      "loss": 3.464,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.271765374612446e-06,
+      "loss": 3.4631,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.270432823891989e-06,
+      "loss": 3.355,
+      "step": 6531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2691008700514685e-06,
+      "loss": 3.3544,
+      "step": 6532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2677695133040446e-06,
+      "loss": 3.3207,
+      "step": 6533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.266438753862792e-06,
+      "loss": 3.4686,
+      "step": 6534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2651085919406836e-06,
+      "loss": 3.3386,
+      "step": 6535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.263779027750599e-06,
+      "loss": 3.3655,
+      "step": 6536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.262450061505322e-06,
+      "loss": 3.4404,
+      "step": 6537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2611216934175397e-06,
+      "loss": 3.5241,
+      "step": 6538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2597939236998424e-06,
+      "loss": 3.4088,
+      "step": 6539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.258466752564732e-06,
+      "loss": 3.379,
+      "step": 6540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2571401802246019e-06,
+      "loss": 3.51,
+      "step": 6541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2558142068917611e-06,
+      "loss": 3.4931,
+      "step": 6542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2544888327784184e-06,
+      "loss": 3.4409,
+      "step": 6543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2531640580966848e-06,
+      "loss": 3.4869,
+      "step": 6544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2518398830585782e-06,
+      "loss": 3.3879,
+      "step": 6545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2505163078760196e-06,
+      "loss": 3.4088,
+      "step": 6546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2491933327608325e-06,
+      "loss": 3.3039,
+      "step": 6547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2478709579247484e-06,
+      "loss": 3.4194,
+      "step": 6548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2465491835793992e-06,
+      "loss": 3.3938,
+      "step": 6549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2452280099363212e-06,
+      "loss": 3.4172,
+      "step": 6550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2439074372069554e-06,
+      "loss": 3.3788,
+      "step": 6551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.242587465602646e-06,
+      "loss": 3.3954,
+      "step": 6552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2412680953346412e-06,
+      "loss": 3.4079,
+      "step": 6553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2399493266140917e-06,
+      "loss": 3.3962,
+      "step": 6554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.238631159652056e-06,
+      "loss": 3.4867,
+      "step": 6555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2373135946594928e-06,
+      "loss": 3.4209,
+      "step": 6556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2359966318472645e-06,
+      "loss": 3.4996,
+      "step": 6557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.234680271426138e-06,
+      "loss": 3.5133,
+      "step": 6558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2333645136067835e-06,
+      "loss": 3.5196,
+      "step": 6559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2320493585997734e-06,
+      "loss": 3.3696,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.23073480661559e-06,
+      "loss": 3.4505,
+      "step": 6561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2294208578646078e-06,
+      "loss": 3.432,
+      "step": 6562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2281075125571157e-06,
+      "loss": 3.3821,
+      "step": 6563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2267947709033002e-06,
+      "loss": 3.3001,
+      "step": 6564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2254826331132519e-06,
+      "loss": 3.4438,
+      "step": 6565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2241710993969647e-06,
+      "loss": 3.3637,
+      "step": 6566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2228601699643405e-06,
+      "loss": 3.2993,
+      "step": 6567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2215498450251744e-06,
+      "loss": 3.4224,
+      "step": 6568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2202401247891766e-06,
+      "loss": 3.2483,
+      "step": 6569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.218931009465949e-06,
+      "loss": 3.3947,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2176224992650072e-06,
+      "loss": 3.4697,
+      "step": 6571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2163145943957633e-06,
+      "loss": 3.434,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2150072950675345e-06,
+      "loss": 3.371,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2137006014895398e-06,
+      "loss": 3.3788,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2123945138709059e-06,
+      "loss": 3.3873,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2110890324206543e-06,
+      "loss": 3.3969,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2097841573477192e-06,
+      "loss": 3.4074,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2084798888609277e-06,
+      "loss": 3.4691,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.207176227169019e-06,
+      "loss": 3.4551,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.20587317248063e-06,
+      "loss": 3.3847,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2045707250043009e-06,
+      "loss": 3.4046,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2032688849484742e-06,
+      "loss": 3.4996,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2019676525215007e-06,
+      "loss": 3.4503,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2006670279316246e-06,
+      "loss": 3.3815,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1993670113870026e-06,
+      "loss": 3.4291,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.198067603095684e-06,
+      "loss": 3.471,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1967688032656305e-06,
+      "loss": 3.3462,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1954706121047004e-06,
+      "loss": 3.4313,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1941730298206566e-06,
+      "loss": 3.3287,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1928760566211623e-06,
+      "loss": 3.4066,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1915796927137879e-06,
+      "loss": 3.4206,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1902839383060021e-06,
+      "loss": 3.3809,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.188988793605178e-06,
+      "loss": 3.4597,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1876942588185902e-06,
+      "loss": 3.4229,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1864003341534158e-06,
+      "loss": 3.414,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1851070198167346e-06,
+      "loss": 3.449,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.183814316015528e-06,
+      "loss": 3.3446,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1825222229566818e-06,
+      "loss": 3.5515,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.181230740846982e-06,
+      "loss": 3.3999,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1799398698931175e-06,
+      "loss": 3.3885,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1786496103016786e-06,
+      "loss": 3.4485,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1773599622791594e-06,
+      "loss": 3.4203,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1760709260319525e-06,
+      "loss": 3.3919,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1747825017663606e-06,
+      "loss": 3.4385,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1734946896885768e-06,
+      "loss": 3.3466,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1722074900047093e-06,
+      "loss": 3.5036,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1709209029207541e-06,
+      "loss": 3.3926,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1696349286426224e-06,
+      "loss": 3.477,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.168349567376119e-06,
+      "loss": 3.4467,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1670648193269545e-06,
+      "loss": 3.3649,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.165780684700737e-06,
+      "loss": 3.3771,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1644971637029844e-06,
+      "loss": 3.4626,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1632142565391057e-06,
+      "loss": 3.3556,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1619319634144232e-06,
+      "loss": 3.4726,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1606502845341494e-06,
+      "loss": 3.3649,
+      "step": 6615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1593692201034085e-06,
+      "loss": 3.3411,
+      "step": 6616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1580887703272197e-06,
+      "loss": 3.4676,
+      "step": 6617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156808935410507e-06,
+      "loss": 3.4423,
+      "step": 6618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.155529715558094e-06,
+      "loss": 3.3887,
+      "step": 6619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1542511109747113e-06,
+      "loss": 3.4251,
+      "step": 6620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1529731218649803e-06,
+      "loss": 3.4253,
+      "step": 6621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1516957484334362e-06,
+      "loss": 3.5815,
+      "step": 6622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1504189908845047e-06,
+      "loss": 3.4606,
+      "step": 6623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1491428494225227e-06,
+      "loss": 3.306,
+      "step": 6624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1478673242517214e-06,
+      "loss": 3.372,
+      "step": 6625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1465924155762364e-06,
+      "loss": 3.3102,
+      "step": 6626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1453181236001026e-06,
+      "loss": 3.3133,
+      "step": 6627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1440444485272622e-06,
+      "loss": 3.5559,
+      "step": 6628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1427713905615478e-06,
+      "loss": 3.4368,
+      "step": 6629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1414989499067048e-06,
+      "loss": 3.5039,
+      "step": 6630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1402271267663722e-06,
+      "loss": 3.3996,
+      "step": 6631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1389559213440937e-06,
+      "loss": 3.3294,
+      "step": 6632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.137685333843312e-06,
+      "loss": 3.2611,
+      "step": 6633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136415364467372e-06,
+      "loss": 3.4784,
+      "step": 6634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1351460134195186e-06,
+      "loss": 3.434,
+      "step": 6635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.133877280902902e-06,
+      "loss": 3.5377,
+      "step": 6636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1326091671205674e-06,
+      "loss": 3.4341,
+      "step": 6637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.131341672275465e-06,
+      "loss": 3.3888,
+      "step": 6638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.130074796570444e-06,
+      "loss": 3.3746,
+      "step": 6639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1288085402082555e-06,
+      "loss": 3.4433,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1275429033915515e-06,
+      "loss": 3.3912,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.126277886322882e-06,
+      "loss": 3.3813,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1250134892047039e-06,
+      "loss": 3.4223,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1237497122393699e-06,
+      "loss": 3.4724,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1224865556291349e-06,
+      "loss": 3.4574,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1212240195761543e-06,
+      "loss": 3.4543,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1199621042824843e-06,
+      "loss": 3.4352,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1187008099500807e-06,
+      "loss": 3.4583,
+      "step": 6648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1174401367808051e-06,
+      "loss": 3.3115,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1161800849764104e-06,
+      "loss": 3.3633,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1149206547385588e-06,
+      "loss": 3.366,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1136618462688081e-06,
+      "loss": 3.3213,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1124036597686188e-06,
+      "loss": 3.4566,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1111460954393506e-06,
+      "loss": 3.474,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1098891534822642e-06,
+      "loss": 3.3672,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1086328340985192e-06,
+      "loss": 3.4357,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1073771374891805e-06,
+      "loss": 3.4391,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1061220638552055e-06,
+      "loss": 3.374,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.104867613397459e-06,
+      "loss": 3.5188,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1036137863167028e-06,
+      "loss": 3.4429,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.102360582813599e-06,
+      "loss": 3.3964,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1011080030887107e-06,
+      "loss": 3.4615,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0998560473425003e-06,
+      "loss": 3.2944,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0986047157753293e-06,
+      "loss": 3.4066,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0973540085874657e-06,
+      "loss": 3.4037,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0961039259790662e-06,
+      "loss": 3.2916,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0948544681502e-06,
+      "loss": 3.3273,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.093605635300825e-06,
+      "loss": 3.4696,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0923574276308085e-06,
+      "loss": 3.4095,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0911098453399116e-06,
+      "loss": 3.4465,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0898628886277984e-06,
+      "loss": 3.3673,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0886165576940299e-06,
+      "loss": 3.4913,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0873708527380733e-06,
+      "loss": 3.3586,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.086125773959285e-06,
+      "loss": 3.402,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0848813215569325e-06,
+      "loss": 3.3241,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.083637495730176e-06,
+      "loss": 3.4116,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0823942966780777e-06,
+      "loss": 3.5082,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0811517245995989e-06,
+      "loss": 3.4517,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0799097796936008e-06,
+      "loss": 3.3959,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0786684621588433e-06,
+      "loss": 3.3701,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0774277721939892e-06,
+      "loss": 3.4379,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0761877099975977e-06,
+      "loss": 3.3619,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0749482757681273e-06,
+      "loss": 3.3725,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0737094697039379e-06,
+      "loss": 3.4395,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.072471292003287e-06,
+      "loss": 3.3759,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0712337428643355e-06,
+      "loss": 3.4546,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0699968224851365e-06,
+      "loss": 3.3761,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.06876053106365e-06,
+      "loss": 3.4604,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0675248687977307e-06,
+      "loss": 3.385,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0662898358851348e-06,
+      "loss": 3.4176,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0650554325235158e-06,
+      "loss": 3.4405,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0638216589104288e-06,
+      "loss": 3.3994,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0625885152433246e-06,
+      "loss": 3.412,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0613560017195607e-06,
+      "loss": 3.3675,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.060124118536382e-06,
+      "loss": 3.36,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0588928658909435e-06,
+      "loss": 3.465,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0576622439802937e-06,
+      "loss": 3.4291,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0564322530013815e-06,
+      "loss": 3.3416,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0552028931510538e-06,
+      "loss": 3.3491,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0539741646260586e-06,
+      "loss": 3.4371,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0527460676230393e-06,
+      "loss": 3.4487,
+      "step": 6701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0515186023385455e-06,
+      "loss": 3.3379,
+      "step": 6702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0502917689690139e-06,
+      "loss": 3.3937,
+      "step": 6703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0490655677107925e-06,
+      "loss": 3.4508,
+      "step": 6704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0478399987601202e-06,
+      "loss": 3.3985,
+      "step": 6705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0466150623131378e-06,
+      "loss": 3.4345,
+      "step": 6706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.045390758565883e-06,
+      "loss": 3.5171,
+      "step": 6707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0441670877142944e-06,
+      "loss": 3.307,
+      "step": 6708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.042944049954207e-06,
+      "loss": 3.401,
+      "step": 6709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0417216454813588e-06,
+      "loss": 3.3658,
+      "step": 6710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.040499874491379e-06,
+      "loss": 3.402,
+      "step": 6711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0392787371798035e-06,
+      "loss": 3.3563,
+      "step": 6712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.038058233742062e-06,
+      "loss": 3.5231,
+      "step": 6713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0368383643734835e-06,
+      "loss": 3.3203,
+      "step": 6714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0356191292692963e-06,
+      "loss": 3.4507,
+      "step": 6715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0344005286246262e-06,
+      "loss": 3.4591,
+      "step": 6716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033182562634497e-06,
+      "loss": 3.279,
+      "step": 6717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.031965231493835e-06,
+      "loss": 3.4315,
+      "step": 6718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03074853539746e-06,
+      "loss": 3.3652,
+      "step": 6719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0295324745400925e-06,
+      "loss": 3.314,
+      "step": 6720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0283170491163509e-06,
+      "loss": 3.3581,
+      "step": 6721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0271022593207508e-06,
+      "loss": 3.363,
+      "step": 6722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0258881053477089e-06,
+      "loss": 3.4345,
+      "step": 6723
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0246745873915358e-06,
+      "loss": 3.4086,
+      "step": 6724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0234617056464464e-06,
+      "loss": 3.4328,
+      "step": 6725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.022249460306548e-06,
+      "loss": 3.5711,
+      "step": 6726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0210378515658492e-06,
+      "loss": 3.3715,
+      "step": 6727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0198268796182552e-06,
+      "loss": 3.3858,
+      "step": 6728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0186165446575707e-06,
+      "loss": 3.3813,
+      "step": 6729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0174068468774955e-06,
+      "loss": 3.3848,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0161977864716344e-06,
+      "loss": 3.3867,
+      "step": 6731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0149893636334801e-06,
+      "loss": 3.3345,
+      "step": 6732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.013781578556432e-06,
+      "loss": 3.485,
+      "step": 6733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0125744314337827e-06,
+      "loss": 3.4265,
+      "step": 6734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0113679224587247e-06,
+      "loss": 3.43,
+      "step": 6735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.010162051824347e-06,
+      "loss": 3.4543,
+      "step": 6736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0089568197236378e-06,
+      "loss": 3.5113,
+      "step": 6737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0077522263494805e-06,
+      "loss": 3.3707,
+      "step": 6738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0065482718946628e-06,
+      "loss": 3.3899,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0053449565518597e-06,
+      "loss": 3.3871,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0041422805136542e-06,
+      "loss": 3.3307,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0029402439725206e-06,
+      "loss": 3.3945,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0017388471208333e-06,
+      "loss": 3.3986,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0005380901508643e-06,
+      "loss": 3.4117,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.99337973254782e-07,
+      "loss": 3.5041,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.981384966246522e-07,
+      "loss": 3.4318,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.969396604524429e-07,
+      "loss": 3.4711,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.957414649300107e-07,
+      "loss": 3.3757,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.945439102491183e-07,
+      "loss": 3.5498,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.93346996601422e-07,
+      "loss": 3.4491,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.921507241784751e-07,
+      "loss": 3.4497,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.909550931717287e-07,
+      "loss": 3.3848,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.89760103772533e-07,
+      "loss": 3.3145,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.885657561721313e-07,
+      "loss": 3.5104,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.873720505616719e-07,
+      "loss": 3.4435,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.861789871321897e-07,
+      "loss": 3.3624,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.849865660746266e-07,
+      "loss": 3.3318,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837947875798165e-07,
+      "loss": 3.335,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.826036518384912e-07,
+      "loss": 3.4711,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.814131590412806e-07,
+      "loss": 3.3085,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.80223309378711e-07,
+      "loss": 3.4995,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.790341030412048e-07,
+      "loss": 3.4485,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.778455402190844e-07,
+      "loss": 3.4846,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.766576211025674e-07,
+      "loss": 3.461,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.754703458817671e-07,
+      "loss": 3.4457,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.742837147466961e-07,
+      "loss": 3.3085,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.73097727887262e-07,
+      "loss": 3.5041,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.719123854932706e-07,
+      "loss": 3.3783,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.707276877544224e-07,
+      "loss": 3.3647,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.695436348603198e-07,
+      "loss": 3.3761,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.683602270004567e-07,
+      "loss": 3.3834,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.671774643642257e-07,
+      "loss": 3.2575,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.659953471409161e-07,
+      "loss": 3.2792,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.648138755197145e-07,
+      "loss": 3.4247,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.636330496897011e-07,
+      "loss": 3.3299,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.624528698398606e-07,
+      "loss": 3.4587,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.612733361590631e-07,
+      "loss": 3.5099,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.600944488360854e-07,
+      "loss": 3.3716,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.589162080595949e-07,
+      "loss": 3.4279,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57738614018157e-07,
+      "loss": 3.4543,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.565616669002331e-07,
+      "loss": 3.405,
+      "step": 6781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.553853668941858e-07,
+      "loss": 3.3339,
+      "step": 6782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.54209714188265e-07,
+      "loss": 3.453,
+      "step": 6783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.530347089706265e-07,
+      "loss": 3.4119,
+      "step": 6784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.51860351429314e-07,
+      "loss": 3.4034,
+      "step": 6785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506866417522747e-07,
+      "loss": 3.3707,
+      "step": 6786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.495135801273487e-07,
+      "loss": 3.3775,
+      "step": 6787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.48341166742272e-07,
+      "loss": 3.3247,
+      "step": 6788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.471694017846767e-07,
+      "loss": 3.3913,
+      "step": 6789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.459982854420957e-07,
+      "loss": 3.439,
+      "step": 6790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.448278179019498e-07,
+      "loss": 3.3966,
+      "step": 6791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.436579993515654e-07,
+      "loss": 3.3631,
+      "step": 6792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.424888299781548e-07,
+      "loss": 3.3545,
+      "step": 6793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.413203099688367e-07,
+      "loss": 3.4826,
+      "step": 6794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.401524395106187e-07,
+      "loss": 3.4208,
+      "step": 6795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.389852187904075e-07,
+      "loss": 3.4619,
+      "step": 6796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.378186479950047e-07,
+      "loss": 3.3709,
+      "step": 6797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.366527273111086e-07,
+      "loss": 3.4522,
+      "step": 6798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.354874569253119e-07,
+      "loss": 3.4989,
+      "step": 6799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.343228370241075e-07,
+      "loss": 3.5079,
+      "step": 6800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.331588677938774e-07,
+      "loss": 3.4635,
+      "step": 6801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31995549420906e-07,
+      "loss": 3.3402,
+      "step": 6802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.308328820913698e-07,
+      "loss": 3.3524,
+      "step": 6803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.296708659913422e-07,
+      "loss": 3.4018,
+      "step": 6804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.285095013067919e-07,
+      "loss": 3.3986,
+      "step": 6805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.273487882235844e-07,
+      "loss": 3.4357,
+      "step": 6806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.261887269274783e-07,
+      "loss": 3.3722,
+      "step": 6807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.250293176041325e-07,
+      "loss": 3.4166,
+      "step": 6808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.23870560439098e-07,
+      "loss": 3.3957,
+      "step": 6809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.227124556178219e-07,
+      "loss": 3.4224,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.21555003325647e-07,
+      "loss": 3.3939,
+      "step": 6811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.203982037478116e-07,
+      "loss": 3.393,
+      "step": 6812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.192420570694532e-07,
+      "loss": 3.3027,
+      "step": 6813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.180865634755964e-07,
+      "loss": 3.4117,
+      "step": 6814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.169317231511698e-07,
+      "loss": 3.3689,
+      "step": 6815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.157775362809935e-07,
+      "loss": 3.5202,
+      "step": 6816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.146240030497832e-07,
+      "loss": 3.411,
+      "step": 6817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.134711236421495e-07,
+      "loss": 3.3275,
+      "step": 6818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.123188982426023e-07,
+      "loss": 3.4247,
+      "step": 6819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.111673270355392e-07,
+      "loss": 3.423,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.100164102052628e-07,
+      "loss": 3.3494,
+      "step": 6821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.088661479359601e-07,
+      "loss": 3.4793,
+      "step": 6822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.077165404117238e-07,
+      "loss": 3.4857,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.065675878165353e-07,
+      "loss": 3.375,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.054192903342729e-07,
+      "loss": 3.4028,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.042716481487091e-07,
+      "loss": 3.4382,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.031246614435163e-07,
+      "loss": 3.4553,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.019783304022528e-07,
+      "loss": 3.3671,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.008326552083829e-07,
+      "loss": 3.4134,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.996876360452556e-07,
+      "loss": 3.4404,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.98543273096123e-07,
+      "loss": 3.3531,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.973995665441282e-07,
+      "loss": 3.4154,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.962565165723097e-07,
+      "loss": 3.334,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.951141233636001e-07,
+      "loss": 3.4089,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.939723871008321e-07,
+      "loss": 3.5248,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.928313079667234e-07,
+      "loss": 3.4254,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.916908861438972e-07,
+      "loss": 3.4239,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.905511218148627e-07,
+      "loss": 3.3956,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.894120151620306e-07,
+      "loss": 3.3962,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.882735663677028e-07,
+      "loss": 3.3687,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.871357756140763e-07,
+      "loss": 3.3538,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.859986430832424e-07,
+      "loss": 3.4222,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.848621689571907e-07,
+      "loss": 3.3533,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.837263534177981e-07,
+      "loss": 3.3115,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.825911966468442e-07,
+      "loss": 3.2913,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.814566988259981e-07,
+      "loss": 3.4135,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.803228601368252e-07,
+      "loss": 3.3574,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.791896807607847e-07,
+      "loss": 3.4845,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.780571608792304e-07,
+      "loss": 3.287,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.769253006734101e-07,
+      "loss": 3.3331,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.757941003244691e-07,
+      "loss": 3.2994,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.746635600134429e-07,
+      "loss": 3.4147,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.73533679921264e-07,
+      "loss": 3.3495,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.72404460228758e-07,
+      "loss": 3.3645,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.71275901116645e-07,
+      "loss": 3.4184,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.701480027655396e-07,
+      "loss": 3.4488,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.690207653559496e-07,
+      "loss": 3.2894,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.678941890682807e-07,
+      "loss": 3.3642,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.667682740828292e-07,
+      "loss": 3.4767,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.656430205797856e-07,
+      "loss": 3.4301,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.645184287392367e-07,
+      "loss": 3.4051,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.633944987411607e-07,
+      "loss": 3.4582,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.622712307654313e-07,
+      "loss": 3.5422,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.611486249918193e-07,
+      "loss": 3.4492,
+      "step": 6864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.600266815999825e-07,
+      "loss": 3.3844,
+      "step": 6865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589054007694808e-07,
+      "loss": 3.4211,
+      "step": 6866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.577847826797592e-07,
+      "loss": 3.4124,
+      "step": 6867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.566648275101652e-07,
+      "loss": 3.3467,
+      "step": 6868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.555455354399351e-07,
+      "loss": 3.3651,
+      "step": 6869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.544269066482009e-07,
+      "loss": 3.4197,
+      "step": 6870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.533089413139861e-07,
+      "loss": 3.402,
+      "step": 6871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.521916396162144e-07,
+      "loss": 3.4442,
+      "step": 6872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510750017336927e-07,
+      "loss": 3.3407,
+      "step": 6873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499590278451331e-07,
+      "loss": 3.4292,
+      "step": 6874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.488437181291315e-07,
+      "loss": 3.3999,
+      "step": 6875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.477290727641857e-07,
+      "loss": 3.4375,
+      "step": 6876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.466150919286819e-07,
+      "loss": 3.3416,
+      "step": 6877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.455017758009021e-07,
+      "loss": 3.4211,
+      "step": 6878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443891245590197e-07,
+      "loss": 3.3745,
+      "step": 6879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.432771383811072e-07,
+      "loss": 3.4096,
+      "step": 6880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421658174451219e-07,
+      "loss": 3.362,
+      "step": 6881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.410551619289253e-07,
+      "loss": 3.473,
+      "step": 6882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.399451720102608e-07,
+      "loss": 3.3426,
+      "step": 6883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388358478667752e-07,
+      "loss": 3.5343,
+      "step": 6884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.377271896760036e-07,
+      "loss": 3.3811,
+      "step": 6885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.366191976153754e-07,
+      "loss": 3.451,
+      "step": 6886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.355118718622129e-07,
+      "loss": 3.3409,
+      "step": 6887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.344052125937358e-07,
+      "loss": 3.3105,
+      "step": 6888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.332992199870493e-07,
+      "loss": 3.3425,
+      "step": 6889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.321938942191593e-07,
+      "loss": 3.5605,
+      "step": 6890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31089235466962e-07,
+      "loss": 3.3838,
+      "step": 6891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.299852439072464e-07,
+      "loss": 3.3835,
+      "step": 6892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.288819197166958e-07,
+      "loss": 3.2762,
+      "step": 6893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.277792630718856e-07,
+      "loss": 3.3698,
+      "step": 6894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.266772741492834e-07,
+      "loss": 3.411,
+      "step": 6895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.25575953125255e-07,
+      "loss": 3.3701,
+      "step": 6896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.244753001760542e-07,
+      "loss": 3.4598,
+      "step": 6897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.233753154778296e-07,
+      "loss": 3.4268,
+      "step": 6898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.222759992066221e-07,
+      "loss": 3.401,
+      "step": 6899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.211773515383659e-07,
+      "loss": 3.3735,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.200793726488915e-07,
+      "loss": 3.2955,
+      "step": 6901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.18982062713915e-07,
+      "loss": 3.4159,
+      "step": 6902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.178854219090532e-07,
+      "loss": 3.4105,
+      "step": 6903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.167894504098122e-07,
+      "loss": 3.468,
+      "step": 6904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.156941483915904e-07,
+      "loss": 3.4519,
+      "step": 6905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.145995160296799e-07,
+      "loss": 3.3939,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.135055534992658e-07,
+      "loss": 3.4353,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.124122609754254e-07,
+      "loss": 3.2621,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.113196386331318e-07,
+      "loss": 3.3933,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.102276866472436e-07,
+      "loss": 3.4037,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.09136405192521e-07,
+      "loss": 3.4359,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.080457944436115e-07,
+      "loss": 3.4073,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.069558545750556e-07,
+      "loss": 3.2914,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.058665857612879e-07,
+      "loss": 3.3473,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.047779881766355e-07,
+      "loss": 3.4577,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.036900619953147e-07,
+      "loss": 3.475,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.026028073914427e-07,
+      "loss": 3.3942,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.015162245390179e-07,
+      "loss": 3.473,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.004303136119407e-07,
+      "loss": 3.4964,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.993450747839987e-07,
+      "loss": 3.4537,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.982605082288747e-07,
+      "loss": 3.404,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.97176614120142e-07,
+      "loss": 3.4152,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.960933926312664e-07,
+      "loss": 3.2962,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.950108439356069e-07,
+      "loss": 3.3381,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.939289682064171e-07,
+      "loss": 3.4285,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.92847765616836e-07,
+      "loss": 3.392,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.917672363399037e-07,
+      "loss": 3.4211,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.906873805485438e-07,
+      "loss": 3.4205,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.896081984155796e-07,
+      "loss": 3.3904,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.885296901137224e-07,
+      "loss": 3.4295,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.874518558155769e-07,
+      "loss": 3.3674,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.86374695693638e-07,
+      "loss": 3.4557,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.852982099202972e-07,
+      "loss": 3.4667,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.842223986678343e-07,
+      "loss": 3.4791,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.831472621084213e-07,
+      "loss": 3.3263,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.820728004141237e-07,
+      "loss": 3.4674,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.809990137568984e-07,
+      "loss": 3.4175,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.799259023085937e-07,
+      "loss": 3.2648,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.788534662409497e-07,
+      "loss": 3.4425,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.777817057256016e-07,
+      "loss": 3.2919,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.767106209340719e-07,
+      "loss": 3.2884,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.756402120377776e-07,
+      "loss": 3.4952,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.745704792080267e-07,
+      "loss": 3.3854,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.735014226160198e-07,
+      "loss": 3.3794,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.724330424328469e-07,
+      "loss": 3.5026,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.713653388294956e-07,
+      "loss": 3.339,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.70298311976836e-07,
+      "loss": 3.416,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.692319620456384e-07,
+      "loss": 3.4312,
+      "step": 6948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.681662892065611e-07,
+      "loss": 3.4452,
+      "step": 6949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.671012936301541e-07,
+      "loss": 3.4187,
+      "step": 6950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.660369754868591e-07,
+      "loss": 3.4663,
+      "step": 6951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.649733349470095e-07,
+      "loss": 3.2702,
+      "step": 6952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639103721808294e-07,
+      "loss": 3.3031,
+      "step": 6953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.628480873584387e-07,
+      "loss": 3.4353,
+      "step": 6954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.61786480649841e-07,
+      "loss": 3.3093,
+      "step": 6955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60725552224939e-07,
+      "loss": 3.4829,
+      "step": 6956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.596653022535227e-07,
+      "loss": 3.4287,
+      "step": 6957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.586057309052747e-07,
+      "loss": 3.5388,
+      "step": 6958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.575468383497686e-07,
+      "loss": 3.4104,
+      "step": 6959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.564886247564696e-07,
+      "loss": 3.4261,
+      "step": 6960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.554310902947332e-07,
+      "loss": 3.4591,
+      "step": 6961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.543742351338101e-07,
+      "loss": 3.3645,
+      "step": 6962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.533180594428357e-07,
+      "loss": 3.3475,
+      "step": 6963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.522625633908432e-07,
+      "loss": 3.3661,
+      "step": 6964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.512077471467527e-07,
+      "loss": 3.418,
+      "step": 6965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.501536108793777e-07,
+      "loss": 3.4597,
+      "step": 6966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.491001547574222e-07,
+      "loss": 3.3359,
+      "step": 6967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.480473789494807e-07,
+      "loss": 3.4584,
+      "step": 6968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.469952836240391e-07,
+      "loss": 3.4738,
+      "step": 6969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.459438689494775e-07,
+      "loss": 3.43,
+      "step": 6970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.448931350940597e-07,
+      "loss": 3.4716,
+      "step": 6971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.438430822259485e-07,
+      "loss": 3.4389,
+      "step": 6972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.427937105131943e-07,
+      "loss": 3.4488,
+      "step": 6973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.417450201237375e-07,
+      "loss": 3.4309,
+      "step": 6974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.406970112254108e-07,
+      "loss": 3.4139,
+      "step": 6975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.396496839859373e-07,
+      "loss": 3.377,
+      "step": 6976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.386030385729304e-07,
+      "loss": 3.43,
+      "step": 6977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.375570751538979e-07,
+      "loss": 3.3472,
+      "step": 6978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.365117938962341e-07,
+      "loss": 3.3269,
+      "step": 6979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.354671949672254e-07,
+      "loss": 3.4002,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.344232785340505e-07,
+      "loss": 3.4089,
+      "step": 6981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.333800447637768e-07,
+      "loss": 3.4602,
+      "step": 6982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.32337493823364e-07,
+      "loss": 3.4034,
+      "step": 6983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.312956258796611e-07,
+      "loss": 3.455,
+      "step": 6984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.3025444109941e-07,
+      "loss": 3.3497,
+      "step": 6985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.292139396492414e-07,
+      "loss": 3.3329,
+      "step": 6986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.281741216956767e-07,
+      "loss": 3.3502,
+      "step": 6987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.271349874051293e-07,
+      "loss": 3.3983,
+      "step": 6988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.260965369439011e-07,
+      "loss": 3.4668,
+      "step": 6989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.250587704781852e-07,
+      "loss": 3.4568,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.240216881740697e-07,
+      "loss": 3.2956,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.229852901975232e-07,
+      "loss": 3.4413,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.219495767144158e-07,
+      "loss": 3.43,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.209145478905011e-07,
+      "loss": 3.4059,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.198802038914248e-07,
+      "loss": 3.3622,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.188465448827242e-07,
+      "loss": 3.4595,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.178135710298262e-07,
+      "loss": 3.3569,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.167812824980452e-07,
+      "loss": 3.4454,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.157496794525943e-07,
+      "loss": 3.4238,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.147187620585649e-07,
+      "loss": 3.3769,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.136885304809487e-07,
+      "loss": 3.3776,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.126589848846238e-07,
+      "loss": 3.3523,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.116301254343583e-07,
+      "loss": 3.5041,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.106019522948105e-07,
+      "loss": 3.4882,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.095744656305286e-07,
+      "loss": 3.4401,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.08547665605952e-07,
+      "loss": 3.2977,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.075215523854123e-07,
+      "loss": 3.4542,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.064961261331243e-07,
+      "loss": 3.2667,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.054713870132007e-07,
+      "loss": 3.3072,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.044473351896392e-07,
+      "loss": 3.3838,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.034239708263291e-07,
+      "loss": 3.3953,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.024012940870501e-07,
+      "loss": 3.3352,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.013793051354711e-07,
+      "loss": 3.3878,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.003580041351498e-07,
+      "loss": 3.4351,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.993373912495393e-07,
+      "loss": 3.4264,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.98317466641974e-07,
+      "loss": 3.4859,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.972982304756848e-07,
+      "loss": 3.3646,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.962796829137903e-07,
+      "loss": 3.4343,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.952618241192988e-07,
+      "loss": 3.3749,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.942446542551073e-07,
+      "loss": 3.5541,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.932281734840052e-07,
+      "loss": 3.383,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.922123819686683e-07,
+      "loss": 3.4433,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.911972798716654e-07,
+      "loss": 3.4712,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.901828673554533e-07,
+      "loss": 3.4784,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.891691445823783e-07,
+      "loss": 3.4218,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.881561117146768e-07,
+      "loss": 3.3042,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.871437689144733e-07,
+      "loss": 3.4429,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.861321163437862e-07,
+      "loss": 3.4082,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.851211541645158e-07,
+      "loss": 3.3764,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.841108825384601e-07,
+      "loss": 3.3404,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.831013016273025e-07,
+      "loss": 3.4841,
+      "step": 7031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.820924115926153e-07,
+      "loss": 3.4209,
+      "step": 7032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.810842125958611e-07,
+      "loss": 3.4586,
+      "step": 7033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.800767047983941e-07,
+      "loss": 3.4861,
+      "step": 7034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.790698883614527e-07,
+      "loss": 3.4538,
+      "step": 7035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.780637634461717e-07,
+      "loss": 3.4091,
+      "step": 7036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.77058330213567e-07,
+      "loss": 3.4157,
+      "step": 7037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.760535888245512e-07,
+      "loss": 3.3484,
+      "step": 7038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.750495394399226e-07,
+      "loss": 3.4024,
+      "step": 7039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.740461822203687e-07,
+      "loss": 3.3942,
+      "step": 7040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.730435173264654e-07,
+      "loss": 3.4813,
+      "step": 7041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.720415449186835e-07,
+      "loss": 3.3802,
+      "step": 7042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.710402651573738e-07,
+      "loss": 3.4308,
+      "step": 7043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.700396782027852e-07,
+      "loss": 3.3616,
+      "step": 7044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.69039784215047e-07,
+      "loss": 3.4473,
+      "step": 7045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.68040583354187e-07,
+      "loss": 3.5026,
+      "step": 7046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.670420757801144e-07,
+      "loss": 3.4011,
+      "step": 7047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.660442616526314e-07,
+      "loss": 3.3601,
+      "step": 7048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.650471411314269e-07,
+      "loss": 3.3541,
+      "step": 7049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.640507143760832e-07,
+      "loss": 3.4296,
+      "step": 7050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.63054981546064e-07,
+      "loss": 3.3821,
+      "step": 7051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.620599428007313e-07,
+      "loss": 3.4212,
+      "step": 7052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.610655982993264e-07,
+      "loss": 3.4043,
+      "step": 7053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.600719482009871e-07,
+      "loss": 3.3838,
+      "step": 7054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.590789926647362e-07,
+      "loss": 3.4562,
+      "step": 7055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.580867318494865e-07,
+      "loss": 3.3472,
+      "step": 7056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.570951659140401e-07,
+      "loss": 3.4384,
+      "step": 7057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.561042950170859e-07,
+      "loss": 3.426,
+      "step": 7058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.551141193172028e-07,
+      "loss": 3.3465,
+      "step": 7059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.5412463897286e-07,
+      "loss": 3.4465,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.531358541424132e-07,
+      "loss": 3.4032,
+      "step": 7061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.521477649841074e-07,
+      "loss": 3.316,
+      "step": 7062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.511603716560766e-07,
+      "loss": 3.4124,
+      "step": 7063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.501736743163433e-07,
+      "loss": 3.3144,
+      "step": 7064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.49187673122818e-07,
+      "loss": 3.4467,
+      "step": 7065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.482023682333e-07,
+      "loss": 3.3329,
+      "step": 7066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.472177598054785e-07,
+      "loss": 3.3595,
+      "step": 7067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.462338479969305e-07,
+      "loss": 3.4456,
+      "step": 7068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.452506329651209e-07,
+      "loss": 3.3998,
+      "step": 7069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.442681148674029e-07,
+      "loss": 3.4429,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.432862938610185e-07,
+      "loss": 3.313,
+      "step": 7071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.423051701030986e-07,
+      "loss": 3.3743,
+      "step": 7072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.413247437506637e-07,
+      "loss": 3.3538,
+      "step": 7073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.403450149606183e-07,
+      "loss": 3.4583,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.393659838897614e-07,
+      "loss": 3.4435,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.383876506947751e-07,
+      "loss": 3.3932,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.374100155322327e-07,
+      "loss": 3.4333,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.364330785585931e-07,
+      "loss": 3.4012,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.354568399302092e-07,
+      "loss": 3.4375,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.344812998033134e-07,
+      "loss": 3.4439,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.335064583340361e-07,
+      "loss": 3.3706,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.325323156783864e-07,
+      "loss": 3.3499,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.315588719922688e-07,
+      "loss": 3.323,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.305861274314722e-07,
+      "loss": 3.4399,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.296140821516756e-07,
+      "loss": 3.4107,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.286427363084429e-07,
+      "loss": 3.3445,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.276720900572326e-07,
+      "loss": 3.4538,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.267021435533816e-07,
+      "loss": 3.4896,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.257328969521254e-07,
+      "loss": 3.3671,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.24764350408577e-07,
+      "loss": 3.4007,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.237965040777461e-07,
+      "loss": 3.393,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.228293581145261e-07,
+      "loss": 3.3497,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.218629126736992e-07,
+      "loss": 3.4799,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.208971679099335e-07,
+      "loss": 3.4525,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1993212397779e-07,
+      "loss": 3.3597,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.189677810317113e-07,
+      "loss": 3.4547,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.180041392260338e-07,
+      "loss": 3.3551,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.170411987149749e-07,
+      "loss": 3.4106,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160789596526468e-07,
+      "loss": 3.5085,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151174221930451e-07,
+      "loss": 3.3424,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.141565864900545e-07,
+      "loss": 3.35,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.13196452697446e-07,
+      "loss": 3.4183,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.122370209688827e-07,
+      "loss": 3.45,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.112782914579074e-07,
+      "loss": 3.3991,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.103202643179589e-07,
+      "loss": 3.4327,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.093629397023582e-07,
+      "loss": 3.4077,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.084063177643156e-07,
+      "loss": 3.438,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.074503986569297e-07,
+      "loss": 3.4457,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.064951825331849e-07,
+      "loss": 3.4019,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.055406695459532e-07,
+      "loss": 3.434,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.045868598479971e-07,
+      "loss": 3.3633,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036337535919634e-07,
+      "loss": 3.4415,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.026813509303869e-07,
+      "loss": 3.4236,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017296520156901e-07,
+      "loss": 3.3578,
+      "step": 7114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.007786570001828e-07,
+      "loss": 3.335,
+      "step": 7115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.998283660360627e-07,
+      "loss": 3.3704,
+      "step": 7116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.988787792754131e-07,
+      "loss": 3.3622,
+      "step": 7117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.979298968702079e-07,
+      "loss": 3.4556,
+      "step": 7118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.969817189723048e-07,
+      "loss": 3.4178,
+      "step": 7119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.960342457334511e-07,
+      "loss": 3.3783,
+      "step": 7120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9508747730528e-07,
+      "loss": 3.3687,
+      "step": 7121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.94141413839312e-07,
+      "loss": 3.4783,
+      "step": 7122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.931960554869542e-07,
+      "loss": 3.4226,
+      "step": 7123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922514023995051e-07,
+      "loss": 3.4595,
+      "step": 7124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913074547281428e-07,
+      "loss": 3.3603,
+      "step": 7125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903642126239412e-07,
+      "loss": 3.4407,
+      "step": 7126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.894216762378513e-07,
+      "loss": 3.4705,
+      "step": 7127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.884798457207214e-07,
+      "loss": 3.4513,
+      "step": 7128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.875387212232797e-07,
+      "loss": 3.372,
+      "step": 7129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.865983028961441e-07,
+      "loss": 3.4989,
+      "step": 7130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.856585908898188e-07,
+      "loss": 3.3538,
+      "step": 7131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.847195853546978e-07,
+      "loss": 3.2616,
+      "step": 7132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.837812864410553e-07,
+      "loss": 3.4364,
+      "step": 7133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.828436942990611e-07,
+      "loss": 3.4259,
+      "step": 7134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.819068090787633e-07,
+      "loss": 3.3849,
+      "step": 7135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.809706309301039e-07,
+      "loss": 3.4467,
+      "step": 7136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.800351600029081e-07,
+      "loss": 3.3155,
+      "step": 7137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.791003964468883e-07,
+      "loss": 3.3374,
+      "step": 7138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781663404116433e-07,
+      "loss": 3.3994,
+      "step": 7139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.772329920466629e-07,
+      "loss": 3.3415,
+      "step": 7140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.76300351501315e-07,
+      "loss": 3.4116,
+      "step": 7141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.753684189248648e-07,
+      "loss": 3.4196,
+      "step": 7142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.744371944664534e-07,
+      "loss": 3.3685,
+      "step": 7143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.735066782751181e-07,
+      "loss": 3.398,
+      "step": 7144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.72576870499777e-07,
+      "loss": 3.3106,
+      "step": 7145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.716477712892371e-07,
+      "loss": 3.3389,
+      "step": 7146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.707193807921896e-07,
+      "loss": 3.3424,
+      "step": 7147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.697916991572172e-07,
+      "loss": 3.4665,
+      "step": 7148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.688647265327841e-07,
+      "loss": 3.448,
+      "step": 7149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.679384630672436e-07,
+      "loss": 3.4051,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.670129089088344e-07,
+      "loss": 3.4276,
+      "step": 7151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.660880642056832e-07,
+      "loss": 3.3811,
+      "step": 7152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.651639291058009e-07,
+      "loss": 3.4314,
+      "step": 7153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.642405037570859e-07,
+      "loss": 3.4161,
+      "step": 7154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.633177883073243e-07,
+      "loss": 3.4479,
+      "step": 7155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.623957829041876e-07,
+      "loss": 3.4094,
+      "step": 7156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.614744876952332e-07,
+      "loss": 3.4423,
+      "step": 7157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.60553902827905e-07,
+      "loss": 3.4068,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.596340284495333e-07,
+      "loss": 3.498,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.587148647073337e-07,
+      "loss": 3.3939,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.577964117484125e-07,
+      "loss": 3.4182,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.568786697197547e-07,
+      "loss": 3.4148,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.559616387682393e-07,
+      "loss": 3.4096,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.550453190406263e-07,
+      "loss": 3.3685,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.541297106835635e-07,
+      "loss": 3.447,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.532148138435855e-07,
+      "loss": 3.405,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.523006286671118e-07,
+      "loss": 3.4065,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.513871553004474e-07,
+      "loss": 3.4031,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.504743938897883e-07,
+      "loss": 3.3752,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.495623445812082e-07,
+      "loss": 3.4203,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.486510075206747e-07,
+      "loss": 3.4519,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.477403828540373e-07,
+      "loss": 3.3827,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.468304707270322e-07,
+      "loss": 3.4787,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.459212712852819e-07,
+      "loss": 3.3936,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.450127846742947e-07,
+      "loss": 3.37,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.441050110394636e-07,
+      "loss": 3.5052,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.431979505260715e-07,
+      "loss": 3.4147,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.422916032792813e-07,
+      "loss": 3.3523,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.413859694441464e-07,
+      "loss": 3.5039,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.404810491656048e-07,
+      "loss": 3.473,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.395768425884796e-07,
+      "loss": 3.4209,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.386733498574792e-07,
+      "loss": 3.426,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.377705711171993e-07,
+      "loss": 3.4691,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.368685065121193e-07,
+      "loss": 3.4279,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.359671561866092e-07,
+      "loss": 3.4547,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.350665202849164e-07,
+      "loss": 3.5377,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341665989511819e-07,
+      "loss": 3.4979,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.332673923294285e-07,
+      "loss": 3.4417,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.323689005635646e-07,
+      "loss": 3.3605,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.314711237973851e-07,
+      "loss": 3.3162,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.305740621745703e-07,
+      "loss": 3.3709,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.29677715838685e-07,
+      "loss": 3.4239,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.287820849331821e-07,
+      "loss": 3.5406,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.278871696013976e-07,
+      "loss": 3.4745,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.269929699865539e-07,
+      "loss": 3.3877,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.260994862317587e-07,
+      "loss": 3.4932,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.252067184800047e-07,
+      "loss": 3.2452,
+      "step": 7197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.243146668741706e-07,
+      "loss": 3.3909,
+      "step": 7198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.234233315570197e-07,
+      "loss": 3.359,
+      "step": 7199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.225327126712033e-07,
+      "loss": 3.3949,
+      "step": 7200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.216428103592547e-07,
+      "loss": 3.3921,
+      "step": 7201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.207536247635941e-07,
+      "loss": 3.3756,
+      "step": 7202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198651560265267e-07,
+      "loss": 3.4122,
+      "step": 7203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.189774042902423e-07,
+      "loss": 3.3572,
+      "step": 7204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.18090369696817e-07,
+      "loss": 3.4201,
+      "step": 7205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.172040523882143e-07,
+      "loss": 3.3434,
+      "step": 7206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.163184525062764e-07,
+      "loss": 3.3437,
+      "step": 7207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.154335701927371e-07,
+      "loss": 3.3957,
+      "step": 7208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.14549405589213e-07,
+      "loss": 3.4933,
+      "step": 7209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.136659588372051e-07,
+      "loss": 3.3679,
+      "step": 7210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.127832300781005e-07,
+      "loss": 3.3414,
+      "step": 7211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.119012194531709e-07,
+      "loss": 3.4075,
+      "step": 7212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110199271035726e-07,
+      "loss": 3.4248,
+      "step": 7213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.101393531703502e-07,
+      "loss": 3.2786,
+      "step": 7214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.092594977944276e-07,
+      "loss": 3.4214,
+      "step": 7215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.083803611166189e-07,
+      "loss": 3.4161,
+      "step": 7216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075019432776201e-07,
+      "loss": 3.3013,
+      "step": 7217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.066242444180142e-07,
+      "loss": 3.3633,
+      "step": 7218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.057472646782668e-07,
+      "loss": 3.4542,
+      "step": 7219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.04871004198731e-07,
+      "loss": 3.4019,
+      "step": 7220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.039954631196419e-07,
+      "loss": 3.4898,
+      "step": 7221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.03120641581124e-07,
+      "loss": 3.4687,
+      "step": 7222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.022465397231802e-07,
+      "loss": 3.4953,
+      "step": 7223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.013731576857039e-07,
+      "loss": 3.3912,
+      "step": 7224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.005004956084713e-07,
+      "loss": 3.3697,
+      "step": 7225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.996285536311423e-07,
+      "loss": 3.4746,
+      "step": 7226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.987573318932632e-07,
+      "loss": 3.33,
+      "step": 7227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.978868305342638e-07,
+      "loss": 3.4065,
+      "step": 7228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.970170496934584e-07,
+      "loss": 3.4208,
+      "step": 7229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.961479895100501e-07,
+      "loss": 3.3721,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.952796501231183e-07,
+      "loss": 3.4,
+      "step": 7231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.944120316716355e-07,
+      "loss": 3.3633,
+      "step": 7232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.935451342944548e-07,
+      "loss": 3.3969,
+      "step": 7233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.926789581303138e-07,
+      "loss": 3.3199,
+      "step": 7234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.918135033178362e-07,
+      "loss": 3.3981,
+      "step": 7235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.909487699955279e-07,
+      "loss": 3.4534,
+      "step": 7236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.900847583017815e-07,
+      "loss": 3.5247,
+      "step": 7237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.892214683748747e-07,
+      "loss": 3.3244,
+      "step": 7238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.88358900352967e-07,
+      "loss": 3.4347,
+      "step": 7239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.874970543741042e-07,
+      "loss": 3.3345,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.866359305762164e-07,
+      "loss": 3.3862,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.857755290971161e-07,
+      "loss": 3.4429,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.84915850074506e-07,
+      "loss": 3.3098,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.840568936459644e-07,
+      "loss": 3.3532,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.831986599489619e-07,
+      "loss": 3.3844,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.823411491208485e-07,
+      "loss": 3.4,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.814843612988618e-07,
+      "loss": 3.4145,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.806282966201198e-07,
+      "loss": 3.3906,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.797729552216302e-07,
+      "loss": 3.3717,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.789183372402789e-07,
+      "loss": 3.4123,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.780644428128417e-07,
+      "loss": 3.403,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.772112720759725e-07,
+      "loss": 3.4155,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7635882516621543e-07,
+      "loss": 3.3663,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.755071022199953e-07,
+      "loss": 3.5471,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7465610337362246e-07,
+      "loss": 3.4997,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7380582876328964e-07,
+      "loss": 3.3603,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729562785250752e-07,
+      "loss": 3.3289,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.721074527949404e-07,
+      "loss": 3.4237,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7125935170873436e-07,
+      "loss": 3.3348,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.704119754021824e-07,
+      "loss": 3.3507,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.695653240109027e-07,
+      "loss": 3.4002,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6871939767039186e-07,
+      "loss": 3.51,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6787419651603216e-07,
+      "loss": 3.4538,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.670297206830887e-07,
+      "loss": 3.3809,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.661859703067123e-07,
+      "loss": 3.3895,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6534294552193604e-07,
+      "loss": 3.3528,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.645006464636792e-07,
+      "loss": 3.3233,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.636590732667412e-07,
+      "loss": 3.3851,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.628182260658087e-07,
+      "loss": 3.374,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.619781049954508e-07,
+      "loss": 3.3188,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6113871019012045e-07,
+      "loss": 3.3897,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.603000417841541e-07,
+      "loss": 3.3583,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5946209991177216e-07,
+      "loss": 3.4421,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5862488470707843e-07,
+      "loss": 3.4195,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5778839630406246e-07,
+      "loss": 3.4343,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569526348365949e-07,
+      "loss": 3.402,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56117600438431e-07,
+      "loss": 3.3773,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5528329324320944e-07,
+      "loss": 3.3549,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5444971338445276e-07,
+      "loss": 3.458,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.536168609955677e-07,
+      "loss": 3.3565,
+      "step": 7280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5278473620984255e-07,
+      "loss": 3.3585,
+      "step": 7281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5195333916045245e-07,
+      "loss": 3.4539,
+      "step": 7282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5112266998045273e-07,
+      "loss": 3.388,
+      "step": 7283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5029272880278473e-07,
+      "loss": 3.326,
+      "step": 7284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.494635157602717e-07,
+      "loss": 3.4708,
+      "step": 7285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4863503098562135e-07,
+      "loss": 3.4164,
+      "step": 7286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.478072746114226e-07,
+      "loss": 3.4423,
+      "step": 7287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4698024677015284e-07,
+      "loss": 3.3725,
+      "step": 7288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4615394759416564e-07,
+      "loss": 3.4905,
+      "step": 7289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.453283772157052e-07,
+      "loss": 3.4967,
+      "step": 7290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.445035357668942e-07,
+      "loss": 3.4903,
+      "step": 7291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.436794233797415e-07,
+      "loss": 3.3911,
+      "step": 7292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4285604018613535e-07,
+      "loss": 3.4162,
+      "step": 7293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.420333863178544e-07,
+      "loss": 3.4821,
+      "step": 7294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.412114619065516e-07,
+      "loss": 3.398,
+      "step": 7295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.403902670837712e-07,
+      "loss": 3.4784,
+      "step": 7296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.395698019809347e-07,
+      "loss": 3.354,
+      "step": 7297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3875006672935096e-07,
+      "loss": 3.425,
+      "step": 7298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3793106146020945e-07,
+      "loss": 3.3671,
+      "step": 7299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3711278630458486e-07,
+      "loss": 3.3787,
+      "step": 7300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3629524139343236e-07,
+      "loss": 3.43,
+      "step": 7301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.35478426857594e-07,
+      "loss": 3.4038,
+      "step": 7302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.346623428277902e-07,
+      "loss": 3.4525,
+      "step": 7303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.338469894346303e-07,
+      "loss": 3.4527,
+      "step": 7304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3303236680859953e-07,
+      "loss": 3.3466,
+      "step": 7305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3221847508007285e-07,
+      "loss": 3.3815,
+      "step": 7306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3140531437930513e-07,
+      "loss": 3.4234,
+      "step": 7307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.305928848364338e-07,
+      "loss": 3.4585,
+      "step": 7308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2978118658147995e-07,
+      "loss": 3.3466,
+      "step": 7309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2897021974435017e-07,
+      "loss": 3.3814,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.281599844548279e-07,
+      "loss": 3.4284,
+      "step": 7311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2735048084258657e-07,
+      "loss": 3.4599,
+      "step": 7312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2654170903717597e-07,
+      "loss": 3.2868,
+      "step": 7313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.257336691680347e-07,
+      "loss": 3.3063,
+      "step": 7314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.249263613644805e-07,
+      "loss": 3.4203,
+      "step": 7315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2411978575571446e-07,
+      "loss": 3.4031,
+      "step": 7316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2331394247082104e-07,
+      "loss": 3.402,
+      "step": 7317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2250883163876823e-07,
+      "loss": 3.4146,
+      "step": 7318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.21704453388404e-07,
+      "loss": 3.3594,
+      "step": 7319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.209008078484633e-07,
+      "loss": 3.3999,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.200978951475609e-07,
+      "loss": 3.4142,
+      "step": 7321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.192957154141947e-07,
+      "loss": 3.2989,
+      "step": 7322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1849426877674516e-07,
+      "loss": 3.4292,
+      "step": 7323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1769355536347646e-07,
+      "loss": 3.3666,
+      "step": 7324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1689357530253494e-07,
+      "loss": 3.5317,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1609432872194767e-07,
+      "loss": 3.3984,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1529581574962784e-07,
+      "loss": 3.463,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1449803651337e-07,
+      "loss": 3.4807,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1370099114084903e-07,
+      "loss": 3.3488,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129046797596248e-07,
+      "loss": 3.3466,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1210910249713907e-07,
+      "loss": 3.4372,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.113142594807151e-07,
+      "loss": 3.371,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1052015083756223e-07,
+      "loss": 3.4268,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.09726776694766e-07,
+      "loss": 3.3923,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0893413717930096e-07,
+      "loss": 3.4625,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0814223241801954e-07,
+      "loss": 3.3517,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0735106253765866e-07,
+      "loss": 3.4112,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0656062766483596e-07,
+      "loss": 3.3699,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0577092792605634e-07,
+      "loss": 3.4306,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0498196344769814e-07,
+      "loss": 3.4579,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0419373435603205e-07,
+      "loss": 3.3797,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.034062407772021e-07,
+      "loss": 3.5164,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0261948283724206e-07,
+      "loss": 3.3723,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0183346066206276e-07,
+      "loss": 3.338,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.010481743774602e-07,
+      "loss": 3.4902,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002636241091101e-07,
+      "loss": 3.3371,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9947980998257476e-07,
+      "loss": 3.3788,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9869673212329263e-07,
+      "loss": 3.4086,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.979143906565902e-07,
+      "loss": 3.43,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.971327857076701e-07,
+      "loss": 3.3519,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9635191740162336e-07,
+      "loss": 3.3802,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.955717858634195e-07,
+      "loss": 3.3895,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9479239121791023e-07,
+      "loss": 3.463,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.940137335898292e-07,
+      "loss": 3.5004,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.93235813103795e-07,
+      "loss": 3.3977,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9245862988430304e-07,
+      "loss": 3.5199,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.916821840557372e-07,
+      "loss": 3.3544,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.909064757423564e-07,
+      "loss": 3.4814,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.901315050683069e-07,
+      "loss": 3.4299,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89357272157615e-07,
+      "loss": 3.3779,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.885837771341888e-07,
+      "loss": 3.3494,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.878110201218177e-07,
+      "loss": 3.3298,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.870390012441755e-07,
+      "loss": 3.3733,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8626772062481334e-07,
+      "loss": 3.4594,
+      "step": 7364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8549717838716916e-07,
+      "loss": 3.4026,
+      "step": 7365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8472737465456047e-07,
+      "loss": 3.3253,
+      "step": 7366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8395830955018587e-07,
+      "loss": 3.4603,
+      "step": 7367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.831899831971264e-07,
+      "loss": 3.4144,
+      "step": 7368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8242239571834483e-07,
+      "loss": 3.4077,
+      "step": 7369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8165554723668783e-07,
+      "loss": 3.3775,
+      "step": 7370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.808894378748795e-07,
+      "loss": 3.3743,
+      "step": 7371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.801240677555296e-07,
+      "loss": 3.4262,
+      "step": 7372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.793594370011272e-07,
+      "loss": 3.3532,
+      "step": 7373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7859554573404347e-07,
+      "loss": 3.4578,
+      "step": 7374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.778323940765316e-07,
+      "loss": 3.3696,
+      "step": 7375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.770699821507279e-07,
+      "loss": 3.4154,
+      "step": 7376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7630831007864634e-07,
+      "loss": 3.3594,
+      "step": 7377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7554737798218723e-07,
+      "loss": 3.3969,
+      "step": 7378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.747871859831287e-07,
+      "loss": 3.4023,
+      "step": 7379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.740277342031323e-07,
+      "loss": 3.4801,
+      "step": 7380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7326902276374087e-07,
+      "loss": 3.3745,
+      "step": 7381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7251105178637834e-07,
+      "loss": 3.3938,
+      "step": 7382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7175382139234997e-07,
+      "loss": 3.4223,
+      "step": 7383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.709973317028448e-07,
+      "loss": 3.3433,
+      "step": 7384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.702415828389283e-07,
+      "loss": 3.3607,
+      "step": 7385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6948657492155424e-07,
+      "loss": 3.3812,
+      "step": 7386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6873230807155046e-07,
+      "loss": 3.4172,
+      "step": 7387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6797878240963203e-07,
+      "loss": 3.4857,
+      "step": 7388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6722599805639246e-07,
+      "loss": 3.4081,
+      "step": 7389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.664739551323082e-07,
+      "loss": 3.2324,
+      "step": 7390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.657226537577341e-07,
+      "loss": 3.4545,
+      "step": 7391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6497209405291165e-07,
+      "loss": 3.3953,
+      "step": 7392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.642222761379571e-07,
+      "loss": 3.4347,
+      "step": 7393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6347320013287504e-07,
+      "loss": 3.4735,
+      "step": 7394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6272486615754334e-07,
+      "loss": 3.4068,
+      "step": 7395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6197727433172793e-07,
+      "loss": 3.4321,
+      "step": 7396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6123042477507307e-07,
+      "loss": 3.2822,
+      "step": 7397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.604843176071049e-07,
+      "loss": 3.488,
+      "step": 7398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.597389529472289e-07,
+      "loss": 3.5441,
+      "step": 7399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5899433091473534e-07,
+      "loss": 3.3048,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.582504516287916e-07,
+      "loss": 3.3329,
+      "step": 7401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.575073152084496e-07,
+      "loss": 3.4194,
+      "step": 7402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.567649217726399e-07,
+      "loss": 3.4644,
+      "step": 7403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.560232714401757e-07,
+      "loss": 3.411,
+      "step": 7404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.552823643297509e-07,
+      "loss": 3.4183,
+      "step": 7405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5454220055993916e-07,
+      "loss": 3.4196,
+      "step": 7406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.538027802491967e-07,
+      "loss": 3.3717,
+      "step": 7407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530641035158616e-07,
+      "loss": 3.3897,
+      "step": 7408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5232617047815053e-07,
+      "loss": 3.472,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5158898125416287e-07,
+      "loss": 3.333,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.508525359618775e-07,
+      "loss": 3.4747,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5011683471915635e-07,
+      "loss": 3.4115,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4938187764373964e-07,
+      "loss": 3.3665,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.486476648532505e-07,
+      "loss": 3.3809,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4791419646519274e-07,
+      "loss": 3.3797,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4718147259695077e-07,
+      "loss": 3.3756,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4644949336578916e-07,
+      "loss": 3.3235,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.457182588888541e-07,
+      "loss": 3.3912,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4498776928317214e-07,
+      "loss": 3.5055,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.442580246656507e-07,
+      "loss": 3.4704,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.435290251530793e-07,
+      "loss": 3.3754,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.42800770862125e-07,
+      "loss": 3.3882,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.420732619093392e-07,
+      "loss": 3.4101,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.41346498411152e-07,
+      "loss": 3.3673,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4062048048387475e-07,
+      "loss": 3.3593,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.398952082436996e-07,
+      "loss": 3.5246,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.391706818066981e-07,
+      "loss": 3.3178,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.384469012888236e-07,
+      "loss": 3.4603,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3772386680591186e-07,
+      "loss": 3.3535,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.370015784736741e-07,
+      "loss": 3.369,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3628003640770847e-07,
+      "loss": 3.3692,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3555924072348866e-07,
+      "loss": 3.3891,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.348391915363719e-07,
+      "loss": 3.4647,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.341198889615943e-07,
+      "loss": 3.4304,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.334013331142727e-07,
+      "loss": 3.457,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3268352410940453e-07,
+      "loss": 3.3432,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.319664620618701e-07,
+      "loss": 3.3222,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.312501470864249e-07,
+      "loss": 3.4101,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.305345792977105e-07,
+      "loss": 3.3644,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.298197588102459e-07,
+      "loss": 3.4107,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.291056857384295e-07,
+      "loss": 3.3922,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.283923601965433e-07,
+      "loss": 3.3581,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.276797822987471e-07,
+      "loss": 3.4373,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2696795215908073e-07,
+      "loss": 3.5042,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2625686989146865e-07,
+      "loss": 3.3764,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.255465356097093e-07,
+      "loss": 3.2433,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2483694942748613e-07,
+      "loss": 3.4075,
+      "step": 7447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.241281114583611e-07,
+      "loss": 3.4093,
+      "step": 7448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2342002181577626e-07,
+      "loss": 3.421,
+      "step": 7449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2271268061305486e-07,
+      "loss": 3.5036,
+      "step": 7450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.220060879633996e-07,
+      "loss": 3.3389,
+      "step": 7451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.213002439798923e-07,
+      "loss": 3.373,
+      "step": 7452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2059514877549824e-07,
+      "loss": 3.4537,
+      "step": 7453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1989080246306035e-07,
+      "loss": 3.4468,
+      "step": 7454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1918720515530133e-07,
+      "loss": 3.4118,
+      "step": 7455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1848435696482606e-07,
+      "loss": 3.3903,
+      "step": 7456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.177822580041162e-07,
+      "loss": 3.4182,
+      "step": 7457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1708090838553916e-07,
+      "loss": 3.4494,
+      "step": 7458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.163803082213357e-07,
+      "loss": 3.4079,
+      "step": 7459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1568045762363107e-07,
+      "loss": 3.4237,
+      "step": 7460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.149813567044302e-07,
+      "loss": 3.4012,
+      "step": 7461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.142830055756163e-07,
+      "loss": 3.4682,
+      "step": 7462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.135854043489533e-07,
+      "loss": 3.4707,
+      "step": 7463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.128885531360859e-07,
+      "loss": 3.4254,
+      "step": 7464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.12192452048537e-07,
+      "loss": 3.3442,
+      "step": 7465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.11497101197713e-07,
+      "loss": 3.4216,
+      "step": 7466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1080250069489505e-07,
+      "loss": 3.487,
+      "step": 7467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1010865065124905e-07,
+      "loss": 3.4234,
+      "step": 7468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.094155511778174e-07,
+      "loss": 3.3634,
+      "step": 7469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0872320238552524e-07,
+      "loss": 3.4687,
+      "step": 7470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0803160438517445e-07,
+      "loss": 3.2946,
+      "step": 7471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0734075728744983e-07,
+      "loss": 3.3262,
+      "step": 7472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0665066120291243e-07,
+      "loss": 3.392,
+      "step": 7473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.059613162420089e-07,
+      "loss": 3.4259,
+      "step": 7474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0527272251505755e-07,
+      "loss": 3.3406,
+      "step": 7475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0458488013226426e-07,
+      "loss": 3.4601,
+      "step": 7476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.038977892037098e-07,
+      "loss": 3.4764,
+      "step": 7477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0321144983935734e-07,
+      "loss": 3.4764,
+      "step": 7478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.025258621490473e-07,
+      "loss": 3.3156,
+      "step": 7479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0184102624250143e-07,
+      "loss": 3.4243,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.011569422293209e-07,
+      "loss": 3.4171,
+      "step": 7481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0047361021898815e-07,
+      "loss": 3.4342,
+      "step": 7482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.997910303208612e-07,
+      "loss": 3.451,
+      "step": 7483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.991092026441811e-07,
+      "loss": 3.3967,
+      "step": 7484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9842812729806824e-07,
+      "loss": 3.366,
+      "step": 7485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.977478043915211e-07,
+      "loss": 3.2511,
+      "step": 7486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.970682340334191e-07,
+      "loss": 3.3657,
+      "step": 7487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9638941633251985e-07,
+      "loss": 3.4214,
+      "step": 7488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9571135139746134e-07,
+      "loss": 3.4433,
+      "step": 7489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9503403933676236e-07,
+      "loss": 3.4117,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.943574802588195e-07,
+      "loss": 3.4782,
+      "step": 7491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9368167427190843e-07,
+      "loss": 3.3421,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.930066214841859e-07,
+      "loss": 3.3413,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.923323220036867e-07,
+      "loss": 3.4432,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.916587759383266e-07,
+      "loss": 3.4399,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9098598339589836e-07,
+      "loss": 3.4041,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9031394448407803e-07,
+      "loss": 3.3867,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8964265931041735e-07,
+      "loss": 3.4767,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8897212798234875e-07,
+      "loss": 3.5358,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8830235060718414e-07,
+      "loss": 3.4077,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8763332729211444e-07,
+      "loss": 3.438,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8696505814421017e-07,
+      "loss": 3.4654,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8629754327042294e-07,
+      "loss": 3.512,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.856307827775789e-07,
+      "loss": 3.3385,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849647767723884e-07,
+      "loss": 3.4146,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.842995253614389e-07,
+      "loss": 3.4255,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.836350286511963e-07,
+      "loss": 3.4281,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8297128674800665e-07,
+      "loss": 3.4677,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8230829975809726e-07,
+      "loss": 3.3355,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.816460677875693e-07,
+      "loss": 3.3438,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.809845909424097e-07,
+      "loss": 3.3074,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8032386932847823e-07,
+      "loss": 3.4459,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.796639030515191e-07,
+      "loss": 3.2886,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7900469221715185e-07,
+      "loss": 3.324,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7834623693087757e-07,
+      "loss": 3.3999,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.776885372980748e-07,
+      "loss": 3.3825,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.77031593424002e-07,
+      "loss": 3.4288,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.763754054137957e-07,
+      "loss": 3.3162,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7571997337247415e-07,
+      "loss": 3.4387,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.750652974049306e-07,
+      "loss": 3.3469,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.744113776159407e-07,
+      "loss": 3.4825,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7375821411015745e-07,
+      "loss": 3.4089,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7310580699211277e-07,
+      "loss": 3.3598,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7245415636621863e-07,
+      "loss": 3.4773,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.718032623367645e-07,
+      "loss": 3.4292,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711531250079191e-07,
+      "loss": 3.2654,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7050374448373276e-07,
+      "loss": 3.4231,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.698551208681288e-07,
+      "loss": 3.3627,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6920725426491544e-07,
+      "loss": 3.4228,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.685601447777769e-07,
+      "loss": 3.2965,
+      "step": 7530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6791379251027595e-07,
+      "loss": 3.3686,
+      "step": 7531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6726819756585534e-07,
+      "loss": 3.3312,
+      "step": 7532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666233600478357e-07,
+      "loss": 3.5234,
+      "step": 7533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6597928005941633e-07,
+      "loss": 3.3206,
+      "step": 7534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.653359577036768e-07,
+      "loss": 3.4097,
+      "step": 7535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.646933930835749e-07,
+      "loss": 3.4207,
+      "step": 7536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.640515863019449e-07,
+      "loss": 3.3547,
+      "step": 7537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6341053746150314e-07,
+      "loss": 3.4449,
+      "step": 7538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.627702466648424e-07,
+      "loss": 3.4106,
+      "step": 7539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621307140144341e-07,
+      "loss": 3.4407,
+      "step": 7540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6149193961262975e-07,
+      "loss": 3.443,
+      "step": 7541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.608539235616597e-07,
+      "loss": 3.4975,
+      "step": 7542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.602166659636307e-07,
+      "loss": 3.4561,
+      "step": 7543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5958016692053054e-07,
+      "loss": 3.3548,
+      "step": 7544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5894442653422334e-07,
+      "loss": 3.4322,
+      "step": 7545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5830944490645383e-07,
+      "loss": 3.414,
+      "step": 7546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5767522213884346e-07,
+      "loss": 3.4693,
+      "step": 7547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5704175833289544e-07,
+      "loss": 3.4281,
+      "step": 7548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5640905358998657e-07,
+      "loss": 3.3587,
+      "step": 7549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557771080113769e-07,
+      "loss": 3.3599,
+      "step": 7550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551459216982022e-07,
+      "loss": 3.45,
+      "step": 7551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.545154947514777e-07,
+      "loss": 3.4406,
+      "step": 7552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538858272720962e-07,
+      "loss": 3.4566,
+      "step": 7553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.53256919360832e-07,
+      "loss": 3.4032,
+      "step": 7554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526287711183323e-07,
+      "loss": 3.4759,
+      "step": 7555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5200138264512895e-07,
+      "loss": 3.4332,
+      "step": 7556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5137475404162616e-07,
+      "loss": 3.3305,
+      "step": 7557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5074888540811246e-07,
+      "loss": 3.4041,
+      "step": 7558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501237768447501e-07,
+      "loss": 3.4432,
+      "step": 7559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4949942845158227e-07,
+      "loss": 3.4739,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.488758403285285e-07,
+      "loss": 3.4189,
+      "step": 7561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4825301257539015e-07,
+      "loss": 3.4795,
+      "step": 7562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4763094529184185e-07,
+      "loss": 3.4602,
+      "step": 7563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4700963857744173e-07,
+      "loss": 3.3096,
+      "step": 7564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4638909253162145e-07,
+      "loss": 3.4068,
+      "step": 7565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457693072536949e-07,
+      "loss": 3.4107,
+      "step": 7566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45150282842852e-07,
+      "loss": 3.362,
+      "step": 7567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4453201939816096e-07,
+      "loss": 3.4164,
+      "step": 7568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439145170185686e-07,
+      "loss": 3.4489,
+      "step": 7569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4329777580290146e-07,
+      "loss": 3.4531,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.426817958498606e-07,
+      "loss": 3.4572,
+      "step": 7571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4206657725802997e-07,
+      "loss": 3.4101,
+      "step": 7572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.414521201258663e-07,
+      "loss": 3.3128,
+      "step": 7573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408384245517098e-07,
+      "loss": 3.4071,
+      "step": 7574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4022549063377533e-07,
+      "loss": 3.4345,
+      "step": 7575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396133184701566e-07,
+      "loss": 3.2989,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.390019081588257e-07,
+      "loss": 3.4989,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.383912597976329e-07,
+      "loss": 3.4129,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3778137348430597e-07,
+      "loss": 3.3868,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3717224931645244e-07,
+      "loss": 3.3598,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.365638873915549e-07,
+      "loss": 3.3764,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3595628780697655e-07,
+      "loss": 3.4767,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3534945065995796e-07,
+      "loss": 3.3469,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3474337604761534e-07,
+      "loss": 3.4358,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.341380640669483e-07,
+      "loss": 3.4487,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.335335148148271e-07,
+      "loss": 3.4229,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3292972838800666e-07,
+      "loss": 3.5282,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.323267048831157e-07,
+      "loss": 3.3964,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3172444439666263e-07,
+      "loss": 3.4061,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.31122947025032e-07,
+      "loss": 3.4056,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3052221286449017e-07,
+      "loss": 3.3615,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.299222420111752e-07,
+      "loss": 3.5069,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.293230345611097e-07,
+      "loss": 3.4495,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2872459061018758e-07,
+      "loss": 3.3751,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.281269102541861e-07,
+      "loss": 3.3925,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275299935887576e-07,
+      "loss": 3.3963,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2693384070943235e-07,
+      "loss": 3.396,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.263384517116185e-07,
+      "loss": 3.4128,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2574382669060313e-07,
+      "loss": 3.3395,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2514996574154792e-07,
+      "loss": 3.4689,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2455686895949736e-07,
+      "loss": 3.3764,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2396453643936723e-07,
+      "loss": 3.3803,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2337296827595723e-07,
+      "loss": 3.3929,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2278216456394054e-07,
+      "loss": 3.4134,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.221921253978704e-07,
+      "loss": 3.4987,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.216028508721746e-07,
+      "loss": 3.2794,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2101434108116382e-07,
+      "loss": 3.485,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.204265961190194e-07,
+      "loss": 3.3704,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1983961607980832e-07,
+      "loss": 3.3786,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.192534010574665e-07,
+      "loss": 3.4384,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1866795114581452e-07,
+      "loss": 3.4063,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1808326643854738e-07,
+      "loss": 3.5076,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1749934702923804e-07,
+      "loss": 3.4027,
+      "step": 7613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.169161930113356e-07,
+      "loss": 3.4442,
+      "step": 7614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1633380447817097e-07,
+      "loss": 3.339,
+      "step": 7615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1575218152294576e-07,
+      "loss": 3.3179,
+      "step": 7616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1517132423874598e-07,
+      "loss": 3.3596,
+      "step": 7617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.145912327185312e-07,
+      "loss": 3.3839,
+      "step": 7618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1401190705513886e-07,
+      "loss": 3.3606,
+      "step": 7619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1343334734128418e-07,
+      "loss": 3.3567,
+      "step": 7620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1285555366956033e-07,
+      "loss": 3.2822,
+      "step": 7621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1227852613243616e-07,
+      "loss": 3.575,
+      "step": 7622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1170226482226108e-07,
+      "loss": 3.3464,
+      "step": 7623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1112676983125912e-07,
+      "loss": 3.4626,
+      "step": 7624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1055204125153216e-07,
+      "loss": 3.5865,
+      "step": 7625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0997807917506053e-07,
+      "loss": 3.3964,
+      "step": 7626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0940488369370015e-07,
+      "loss": 3.4456,
+      "step": 7627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0883245489918602e-07,
+      "loss": 3.4417,
+      "step": 7628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0826079288312817e-07,
+      "loss": 3.3583,
+      "step": 7629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0768989773701732e-07,
+      "loss": 3.3466,
+      "step": 7630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0711976955221924e-07,
+      "loss": 3.2955,
+      "step": 7631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0655040841997654e-07,
+      "loss": 3.3397,
+      "step": 7632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0598181443140964e-07,
+      "loss": 3.4609,
+      "step": 7633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0541398767751685e-07,
+      "loss": 3.3216,
+      "step": 7634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.048469282491722e-07,
+      "loss": 3.3476,
+      "step": 7635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0428063623713024e-07,
+      "loss": 3.3555,
+      "step": 7636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0371511173201741e-07,
+      "loss": 3.4403,
+      "step": 7637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0315035482434186e-07,
+      "loss": 3.5011,
+      "step": 7638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.025863656044874e-07,
+      "loss": 3.4002,
+      "step": 7639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0202314416271463e-07,
+      "loss": 3.3563,
+      "step": 7640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0146069058916085e-07,
+      "loss": 3.4213,
+      "step": 7641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.008990049738424e-07,
+      "loss": 3.3509,
+      "step": 7642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0033808740664961e-07,
+      "loss": 3.4364,
+      "step": 7643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.99777937977354e-07,
+      "loss": 3.3114,
+      "step": 7644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.992185567755994e-07,
+      "loss": 3.4066,
+      "step": 7645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9865994389091204e-07,
+      "loss": 3.4801,
+      "step": 7646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9810209941268978e-07,
+      "loss": 3.259,
+      "step": 7647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.975450234302112e-07,
+      "loss": 3.3645,
+      "step": 7648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.969887160326306e-07,
+      "loss": 3.3998,
+      "step": 7649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.964331773089795e-07,
+      "loss": 3.3943,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9587840734816566e-07,
+      "loss": 3.4044,
+      "step": 7651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9532440623897587e-07,
+      "loss": 3.3528,
+      "step": 7652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9477117407007084e-07,
+      "loss": 3.4099,
+      "step": 7653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9421871092999144e-07,
+      "loss": 3.3902,
+      "step": 7654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.936670169071525e-07,
+      "loss": 3.3914,
+      "step": 7655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9311609208984784e-07,
+      "loss": 3.3365,
+      "step": 7656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9256593656624744e-07,
+      "loss": 3.3858,
+      "step": 7657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.920165504243976e-07,
+      "loss": 3.3343,
+      "step": 7658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9146793375222238e-07,
+      "loss": 3.5093,
+      "step": 7659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.909200866375238e-07,
+      "loss": 3.5304,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9037300916797673e-07,
+      "loss": 3.4439,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8982670143113778e-07,
+      "loss": 3.4698,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8928116351443648e-07,
+      "loss": 3.2344,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.887363955051813e-07,
+      "loss": 3.3075,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8819239749055752e-07,
+      "loss": 3.2873,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8764916955762547e-07,
+      "loss": 3.2888,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.871067117933234e-07,
+      "loss": 3.3221,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8656502428446742e-07,
+      "loss": 3.4053,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.860241071177482e-07,
+      "loss": 3.5023,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8548396037973425e-07,
+      "loss": 3.3364,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8494458415687145e-07,
+      "loss": 3.5225,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8440597853548014e-07,
+      "loss": 3.3758,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.838681436017603e-07,
+      "loss": 3.4648,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.833310794417853e-07,
+      "loss": 3.432,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8279478614150803e-07,
+      "loss": 3.3838,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8225926378675762e-07,
+      "loss": 3.5301,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8172451246323718e-07,
+      "loss": 3.3216,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8119053225652993e-07,
+      "loss": 3.3265,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.806573232520936e-07,
+      "loss": 3.3681,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8012488553526165e-07,
+      "loss": 3.4576,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7959321919124806e-07,
+      "loss": 3.374,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7906232430513814e-07,
+      "loss": 3.3914,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7853220096189727e-07,
+      "loss": 3.5028,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7800284924636701e-07,
+      "loss": 3.4349,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.774742692432646e-07,
+      "loss": 3.3923,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7694646103718293e-07,
+      "loss": 3.3234,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7641942471259388e-07,
+      "loss": 3.4347,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.758931603538422e-07,
+      "loss": 3.3855,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7536766804515437e-07,
+      "loss": 3.3261,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.748429478706276e-07,
+      "loss": 3.4677,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7431899991423917e-07,
+      "loss": 3.3574,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7379582425984144e-07,
+      "loss": 3.3218,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.732734209911635e-07,
+      "loss": 3.4282,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7275179019181132e-07,
+      "loss": 3.3381,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7223093194526586e-07,
+      "loss": 3.4024,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.717108463348849e-07,
+      "loss": 3.4543,
+      "step": 7696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.711915334439046e-07,
+      "loss": 3.3344,
+      "step": 7697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.706729933554341e-07,
+      "loss": 3.3505,
+      "step": 7698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7015522615246138e-07,
+      "loss": 3.4936,
+      "step": 7699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6963823191785024e-07,
+      "loss": 3.3164,
+      "step": 7700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6912201073433943e-07,
+      "loss": 3.4153,
+      "step": 7701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6860656268454621e-07,
+      "loss": 3.3511,
+      "step": 7702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6809188785096186e-07,
+      "loss": 3.3468,
+      "step": 7703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.675779863159549e-07,
+      "loss": 3.4186,
+      "step": 7704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.670648581617712e-07,
+      "loss": 3.4138,
+      "step": 7705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6655250347053063e-07,
+      "loss": 3.543,
+      "step": 7706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.660409223242315e-07,
+      "loss": 3.3816,
+      "step": 7707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6553011480474667e-07,
+      "loss": 3.4286,
+      "step": 7708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6502008099382572e-07,
+      "loss": 3.4913,
+      "step": 7709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6451082097309444e-07,
+      "loss": 3.3786,
+      "step": 7710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6400233482405436e-07,
+      "loss": 3.4048,
+      "step": 7711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6349462262808425e-07,
+      "loss": 3.3526,
+      "step": 7712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629876844664391e-07,
+      "loss": 3.4204,
+      "step": 7713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624815204202479e-07,
+      "loss": 3.4516,
+      "step": 7714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6197613057051808e-07,
+      "loss": 3.4832,
+      "step": 7715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6147151499813162e-07,
+      "loss": 3.3204,
+      "step": 7716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.609676737838467e-07,
+      "loss": 3.3746,
+      "step": 7717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6046460700830048e-07,
+      "loss": 3.4094,
+      "step": 7718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5996231475200074e-07,
+      "loss": 3.3617,
+      "step": 7719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5946079709533603e-07,
+      "loss": 3.412,
+      "step": 7720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5896005411856873e-07,
+      "loss": 3.3073,
+      "step": 7721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.584600859018387e-07,
+      "loss": 3.3875,
+      "step": 7722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.579608925251591e-07,
+      "loss": 3.3666,
+      "step": 7723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5746247406842275e-07,
+      "loss": 3.3486,
+      "step": 7724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5696483061139412e-07,
+      "loss": 3.415,
+      "step": 7725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5646796223371951e-07,
+      "loss": 3.3935,
+      "step": 7726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5597186901491423e-07,
+      "loss": 3.3823,
+      "step": 7727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5547655103437476e-07,
+      "loss": 3.3871,
+      "step": 7728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5498200837137213e-07,
+      "loss": 3.5024,
+      "step": 7729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5448824110505244e-07,
+      "loss": 3.3188,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.539952493144381e-07,
+      "loss": 3.466,
+      "step": 7731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5350303307842705e-07,
+      "loss": 3.4748,
+      "step": 7732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5301159247579413e-07,
+      "loss": 3.3431,
+      "step": 7733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5252092758519076e-07,
+      "loss": 3.3971,
+      "step": 7734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5203103848514034e-07,
+      "loss": 3.2659,
+      "step": 7735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5154192525404677e-07,
+      "loss": 3.3767,
+      "step": 7736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.510535879701869e-07,
+      "loss": 3.3615,
+      "step": 7737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.505660267117143e-07,
+      "loss": 3.4189,
+      "step": 7738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.500792415566582e-07,
+      "loss": 3.418,
+      "step": 7739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4959323258292413e-07,
+      "loss": 3.3848,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4910799986829206e-07,
+      "loss": 3.3068,
+      "step": 7741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.486235434904204e-07,
+      "loss": 3.3854,
+      "step": 7742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4813986352683886e-07,
+      "loss": 3.3884,
+      "step": 7743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4765696005495822e-07,
+      "loss": 3.3514,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.471748331520606e-07,
+      "loss": 3.3803,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.466934828953065e-07,
+      "loss": 3.304,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4621290936173148e-07,
+      "loss": 3.4024,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.457331126282452e-07,
+      "loss": 3.3929,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4525409277163505e-07,
+      "loss": 3.2942,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.447758498685642e-07,
+      "loss": 3.4113,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4429838399556917e-07,
+      "loss": 3.431,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4382169522906498e-07,
+      "loss": 3.3771,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4334578364534003e-07,
+      "loss": 3.3624,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.428706493205595e-07,
+      "loss": 3.3349,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.423962923307637e-07,
+      "loss": 3.3416,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4192271275186798e-07,
+      "loss": 3.3943,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4144991065966564e-07,
+      "loss": 3.3603,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4097788612982333e-07,
+      "loss": 3.4544,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4050663923788342e-07,
+      "loss": 3.3254,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4003617005926452e-07,
+      "loss": 3.4788,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3956647866926075e-07,
+      "loss": 3.2882,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3909756514304094e-07,
+      "loss": 3.4761,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.386294295556506e-07,
+      "loss": 3.307,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3816207198200983e-07,
+      "loss": 3.4339,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3769549249691426e-07,
+      "loss": 3.4176,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3722969117503648e-07,
+      "loss": 3.4148,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3676466809092182e-07,
+      "loss": 3.3306,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3630042331899295e-07,
+      "loss": 3.3403,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.358369569335488e-07,
+      "loss": 3.2633,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3537426900876115e-07,
+      "loss": 3.3297,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3491235961867966e-07,
+      "loss": 3.4689,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.344512288372263e-07,
+      "loss": 3.3657,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3399087673820255e-07,
+      "loss": 3.3599,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3353130339528287e-07,
+      "loss": 3.394,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3307250888201674e-07,
+      "loss": 3.1924,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3261449327182985e-07,
+      "loss": 3.4349,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3215725663802303e-07,
+      "loss": 3.4531,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3170079905377164e-07,
+      "loss": 3.4061,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.312451205921289e-07,
+      "loss": 3.3824,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.307902213260198e-07,
+      "loss": 3.3758,
+      "step": 7780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303361013282478e-07,
+      "loss": 3.4051,
+      "step": 7781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2988276067148974e-07,
+      "loss": 3.4584,
+      "step": 7782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2943019942829816e-07,
+      "loss": 3.3956,
+      "step": 7783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.289784176711012e-07,
+      "loss": 3.4487,
+      "step": 7784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.285274154722016e-07,
+      "loss": 3.4735,
+      "step": 7785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2807719290377775e-07,
+      "loss": 3.3817,
+      "step": 7786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2762775003788419e-07,
+      "loss": 3.4053,
+      "step": 7787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2717908694644843e-07,
+      "loss": 3.2453,
+      "step": 7788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2673120370127522e-07,
+      "loss": 3.3668,
+      "step": 7789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2628410037404394e-07,
+      "loss": 3.412,
+      "step": 7790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2583777703630894e-07,
+      "loss": 3.4311,
+      "step": 7791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2539223375949983e-07,
+      "loss": 3.5071,
+      "step": 7792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2494747061492064e-07,
+      "loss": 3.3953,
+      "step": 7793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2450348767375164e-07,
+      "loss": 3.4152,
+      "step": 7794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2406028500704826e-07,
+      "loss": 3.5039,
+      "step": 7795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2361786268574095e-07,
+      "loss": 3.3521,
+      "step": 7796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.231762207806342e-07,
+      "loss": 3.4279,
+      "step": 7797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2273535936240866e-07,
+      "loss": 3.4096,
+      "step": 7798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2229527850161905e-07,
+      "loss": 3.3866,
+      "step": 7799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2185597826869788e-07,
+      "loss": 3.4313,
+      "step": 7800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2141745873394839e-07,
+      "loss": 3.3507,
+      "step": 7801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2097971996755277e-07,
+      "loss": 3.3404,
+      "step": 7802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.205427620395666e-07,
+      "loss": 3.324,
+      "step": 7803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2010658501992013e-07,
+      "loss": 3.2783,
+      "step": 7804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1967118897841912e-07,
+      "loss": 3.3547,
+      "step": 7805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1923657398474564e-07,
+      "loss": 3.3545,
+      "step": 7806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1880274010845294e-07,
+      "loss": 3.4271,
+      "step": 7807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1836968741897492e-07,
+      "loss": 3.4212,
+      "step": 7808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1793741598561447e-07,
+      "loss": 3.395,
+      "step": 7809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1750592587755405e-07,
+      "loss": 3.4056,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.170752171638484e-07,
+      "loss": 3.471,
+      "step": 7811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1664528991342905e-07,
+      "loss": 3.4567,
+      "step": 7812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1621614419510041e-07,
+      "loss": 3.4109,
+      "step": 7813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1578778007754477e-07,
+      "loss": 3.3939,
+      "step": 7814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1536019762931505e-07,
+      "loss": 3.3022,
+      "step": 7815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1493339691884376e-07,
+      "loss": 3.3896,
+      "step": 7816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1450737801443457e-07,
+      "loss": 3.4262,
+      "step": 7817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.140821409842685e-07,
+      "loss": 3.4144,
+      "step": 7818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1365768589640059e-07,
+      "loss": 3.3454,
+      "step": 7819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1323401281875979e-07,
+      "loss": 3.4108,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1281112181915077e-07,
+      "loss": 3.3745,
+      "step": 7821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.123890129652544e-07,
+      "loss": 3.3705,
+      "step": 7822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1196768632462274e-07,
+      "loss": 3.4959,
+      "step": 7823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1154714196468741e-07,
+      "loss": 3.4397,
+      "step": 7824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1112737995275013e-07,
+      "loss": 3.274,
+      "step": 7825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1070840035599051e-07,
+      "loss": 3.4382,
+      "step": 7826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1029020324146266e-07,
+      "loss": 3.4241,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0987278867609419e-07,
+      "loss": 3.3499,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0945615672668719e-07,
+      "loss": 3.388,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0904030745992222e-07,
+      "loss": 3.4261,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0862524094234827e-07,
+      "loss": 3.3307,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0821095724039498e-07,
+      "loss": 3.3536,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0779745642036432e-07,
+      "loss": 3.3372,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0738473854843167e-07,
+      "loss": 3.4152,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0697280369064921e-07,
+      "loss": 3.4243,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0656165191294254e-07,
+      "loss": 3.4773,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0615128328111346e-07,
+      "loss": 3.3565,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0574169786083554e-07,
+      "loss": 3.4175,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0533289571766026e-07,
+      "loss": 3.3129,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0492487691701192e-07,
+      "loss": 3.4009,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0451764152418997e-07,
+      "loss": 3.3723,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0411118960436839e-07,
+      "loss": 3.3787,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0370552122259569e-07,
+      "loss": 3.4077,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.033006364437944e-07,
+      "loss": 3.4513,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0289653533276378e-07,
+      "loss": 3.3825,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0249321795417377e-07,
+      "loss": 3.4313,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0209068437257385e-07,
+      "loss": 3.4905,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0168893465238416e-07,
+      "loss": 3.4736,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0128796885790104e-07,
+      "loss": 3.322,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0088778705329427e-07,
+      "loss": 3.3781,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0048838930261095e-07,
+      "loss": 3.3844,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.000897756697683e-07,
+      "loss": 3.4058,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.969194621856249e-08,
+      "loss": 3.4262,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.929490101265982e-08,
+      "loss": 3.4007,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.889864011560557e-08,
+      "loss": 3.4706,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.85031635908168e-08,
+      "loss": 3.4723,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.810847150158509e-08,
+      "loss": 3.3762,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.77145639110766e-08,
+      "loss": 3.4173,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.732144088233419e-08,
+      "loss": 3.4615,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.692910247827037e-08,
+      "loss": 3.4063,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.653754876167821e-08,
+      "loss": 3.4348,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.614677979521925e-08,
+      "loss": 3.4597,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.575679564143347e-08,
+      "loss": 3.3836,
+      "step": 7863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.536759636273373e-08,
+      "loss": 3.4246,
+      "step": 7864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.497918202140744e-08,
+      "loss": 3.4964,
+      "step": 7865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.459155267961595e-08,
+      "loss": 3.386,
+      "step": 7866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.420470839939688e-08,
+      "loss": 3.3681,
+      "step": 7867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.381864924265904e-08,
+      "loss": 3.3508,
+      "step": 7868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.343337527118912e-08,
+      "loss": 3.3512,
+      "step": 7869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.304888654664446e-08,
+      "loss": 3.3625,
+      "step": 7870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.266518313056027e-08,
+      "loss": 3.4441,
+      "step": 7871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.228226508434357e-08,
+      "loss": 3.4436,
+      "step": 7872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.190013246927642e-08,
+      "loss": 3.4574,
+      "step": 7873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.151878534651603e-08,
+      "loss": 3.3809,
+      "step": 7874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.113822377709303e-08,
+      "loss": 3.3962,
+      "step": 7875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.075844782191145e-08,
+      "loss": 3.3931,
+      "step": 7876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.037945754175214e-08,
+      "loss": 3.395,
+      "step": 7877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.00012529972677e-08,
+      "loss": 3.3525,
+      "step": 7878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.962383424898691e-08,
+      "loss": 3.4418,
+      "step": 7879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.92472013573109e-08,
+      "loss": 3.3857,
+      "step": 7880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.887135438251648e-08,
+      "loss": 3.4437,
+      "step": 7881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.84962933847533e-08,
+      "loss": 3.3415,
+      "step": 7882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.812201842404777e-08,
+      "loss": 3.3543,
+      "step": 7883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.774852956029756e-08,
+      "loss": 3.4447,
+      "step": 7884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.73758268532765e-08,
+      "loss": 3.4206,
+      "step": 7885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.70039103626319e-08,
+      "loss": 3.4265,
+      "step": 7886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6632780147885e-08,
+      "loss": 3.4209,
+      "step": 7887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.626243626843166e-08,
+      "loss": 3.3934,
+      "step": 7888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.589287878354114e-08,
+      "loss": 3.4212,
+      "step": 7889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.552410775235831e-08,
+      "loss": 3.4989,
+      "step": 7890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.515612323390044e-08,
+      "loss": 3.4735,
+      "step": 7891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.47889252870604e-08,
+      "loss": 3.434,
+      "step": 7892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.442251397060453e-08,
+      "loss": 3.4023,
+      "step": 7893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.405688934317313e-08,
+      "loss": 3.3727,
+      "step": 7894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.369205146328052e-08,
+      "loss": 3.3133,
+      "step": 7895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.33280003893161e-08,
+      "loss": 3.4379,
+      "step": 7896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.296473617954159e-08,
+      "loss": 3.3901,
+      "step": 7897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.26022588920944e-08,
+      "loss": 3.3525,
+      "step": 7898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.22405685849853e-08,
+      "loss": 3.342,
+      "step": 7899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.187966531609915e-08,
+      "loss": 3.4013,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.151954914319527e-08,
+      "loss": 3.3298,
+      "step": 7901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.11602201239059e-08,
+      "loss": 3.3183,
+      "step": 7902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.080167831573837e-08,
+      "loss": 3.4756,
+      "step": 7903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.044392377607458e-08,
+      "loss": 3.4082,
+      "step": 7904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.008695656216814e-08,
+      "loss": 3.4924,
+      "step": 7905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.973077673114948e-08,
+      "loss": 3.3197,
+      "step": 7906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.937538434002024e-08,
+      "loss": 3.4131,
+      "step": 7907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.902077944565823e-08,
+      "loss": 3.3885,
+      "step": 7908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.866696210481472e-08,
+      "loss": 3.3502,
+      "step": 7909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.831393237411389e-08,
+      "loss": 3.3685,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.79616903100544e-08,
+      "loss": 3.4061,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.761023596901063e-08,
+      "loss": 3.4079,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.725956940722701e-08,
+      "loss": 3.3647,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.690969068082698e-08,
+      "loss": 3.5029,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.656059984580244e-08,
+      "loss": 3.4188,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.621229695802368e-08,
+      "loss": 3.3128,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.586478207323222e-08,
+      "loss": 3.4407,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.551805524704525e-08,
+      "loss": 3.4053,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.517211653495115e-08,
+      "loss": 3.4979,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.482696599231676e-08,
+      "loss": 3.3881,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.448260367437676e-08,
+      "loss": 3.3993,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.413902963624542e-08,
+      "loss": 3.3758,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.379624393290707e-08,
+      "loss": 3.3908,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.345424661922118e-08,
+      "loss": 3.3257,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.311303774992173e-08,
+      "loss": 3.4598,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.277261737961505e-08,
+      "loss": 3.4603,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.243298556278256e-08,
+      "loss": 3.4659,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.209414235377854e-08,
+      "loss": 3.4319,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.17560878068313e-08,
+      "loss": 3.5176,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.141882197604366e-08,
+      "loss": 3.4271,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.108234491539134e-08,
+      "loss": 3.3756,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.074665667872405e-08,
+      "loss": 3.5107,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.041175731976546e-08,
+      "loss": 3.5054,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.007764689211272e-08,
+      "loss": 3.3111,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.974432544923748e-08,
+      "loss": 3.2756,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.94117930444832e-08,
+      "loss": 3.3611,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.908004973107007e-08,
+      "loss": 3.4442,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.874909556208898e-08,
+      "loss": 3.3835,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.841893059050587e-08,
+      "loss": 3.3973,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.808955486916068e-08,
+      "loss": 3.3693,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.776096845076796e-08,
+      "loss": 3.353,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.743317138791228e-08,
+      "loss": 3.4935,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.710616373305556e-08,
+      "loss": 3.4115,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.677994553853207e-08,
+      "loss": 3.2901,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.645451685655002e-08,
+      "loss": 3.4465,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.612987773919055e-08,
+      "loss": 3.3862,
+      "step": 7946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.580602823840932e-08,
+      "loss": 3.4301,
+      "step": 7947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.54829684060343e-08,
+      "loss": 3.4353,
+      "step": 7948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.516069829376914e-08,
+      "loss": 3.3739,
+      "step": 7949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.483921795318927e-08,
+      "loss": 3.3228,
+      "step": 7950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.45185274357446e-08,
+      "loss": 3.5302,
+      "step": 7951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.419862679275801e-08,
+      "loss": 3.3581,
+      "step": 7952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.387951607542742e-08,
+      "loss": 3.4351,
+      "step": 7953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.356119533482252e-08,
+      "loss": 3.4443,
+      "step": 7954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.324366462188759e-08,
+      "loss": 3.3469,
+      "step": 7955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.292692398743971e-08,
+      "loss": 3.4218,
+      "step": 7956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.261097348217116e-08,
+      "loss": 3.406,
+      "step": 7957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.229581315664535e-08,
+      "loss": 3.3641,
+      "step": 7958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.198144306130139e-08,
+      "loss": 3.4549,
+      "step": 7959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.166786324645124e-08,
+      "loss": 3.4549,
+      "step": 7960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.135507376227978e-08,
+      "loss": 3.4145,
+      "step": 7961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.104307465884584e-08,
+      "loss": 3.3595,
+      "step": 7962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.073186598608172e-08,
+      "loss": 3.3157,
+      "step": 7963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.042144779379366e-08,
+      "loss": 3.4435,
+      "step": 7964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.011182013166028e-08,
+      "loss": 3.4734,
+      "step": 7965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.980298304923526e-08,
+      "loss": 3.414,
+      "step": 7966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9494936595944076e-08,
+      "loss": 3.4518,
+      "step": 7967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.918768082108728e-08,
+      "loss": 3.3308,
+      "step": 7968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.888121577383776e-08,
+      "loss": 3.4195,
+      "step": 7969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.857554150324185e-08,
+      "loss": 3.329,
+      "step": 7970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.82706580582193e-08,
+      "loss": 3.3898,
+      "step": 7971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7966565487564965e-08,
+      "loss": 3.394,
+      "step": 7972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7663263839944915e-08,
+      "loss": 3.384,
+      "step": 7973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7360753163899774e-08,
+      "loss": 3.3058,
+      "step": 7974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.705903350784303e-08,
+      "loss": 3.3093,
+      "step": 7975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.675810492006273e-08,
+      "loss": 3.476,
+      "step": 7976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.645796744871812e-08,
+      "loss": 3.5225,
+      "step": 7977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.615862114184467e-08,
+      "loss": 3.4699,
+      "step": 7978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.586006604734906e-08,
+      "loss": 3.3494,
+      "step": 7979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.556230221301195e-08,
+      "loss": 3.3992,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5265329686487455e-08,
+      "loss": 3.393,
+      "step": 7981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.496914851530366e-08,
+      "loss": 3.4805,
+      "step": 7982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.467375874686043e-08,
+      "loss": 3.3458,
+      "step": 7983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.437916042843272e-08,
+      "loss": 3.4603,
+      "step": 7984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.408535360716782e-08,
+      "loss": 3.354,
+      "step": 7985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3792338330086455e-08,
+      "loss": 3.4039,
+      "step": 7986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.350011464408278e-08,
+      "loss": 3.2983,
+      "step": 7987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.320868259592493e-08,
+      "loss": 3.4304,
+      "step": 7988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.291804223225283e-08,
+      "loss": 3.3501,
+      "step": 7989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2628193599580377e-08,
+      "loss": 3.4263,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2339136744296023e-08,
+      "loss": 3.4717,
+      "step": 7991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.205087171265999e-08,
+      "loss": 3.4185,
+      "step": 7992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.176339855080537e-08,
+      "loss": 3.4593,
+      "step": 7993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1476717304740905e-08,
+      "loss": 3.4384,
+      "step": 7994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1190828020345476e-08,
+      "loss": 3.3296,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.090573074337357e-08,
+      "loss": 3.4303,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.06214255194526e-08,
+      "loss": 3.4142,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.033791239408225e-08,
+      "loss": 3.4209,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0055191412635705e-08,
+      "loss": 3.2888,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9773262620360084e-08,
+      "loss": 3.3983,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9492126062374856e-08,
+      "loss": 3.3906,
+      "step": 8001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9211781783674026e-08,
+      "loss": 3.385,
+      "step": 8002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.893222982912282e-08,
+      "loss": 3.3948,
+      "step": 8003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865347024346101e-08,
+      "loss": 3.4405,
+      "step": 8004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.837550307130179e-08,
+      "loss": 3.3702,
+      "step": 8005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.809832835713069e-08,
+      "loss": 3.4415,
+      "step": 8006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.782194614530666e-08,
+      "loss": 3.4884,
+      "step": 8007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.754635648006212e-08,
+      "loss": 3.3457,
+      "step": 8008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.727155940550177e-08,
+      "loss": 3.282,
+      "step": 8009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.699755496560543e-08,
+      "loss": 3.4198,
+      "step": 8010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.672434320422359e-08,
+      "loss": 3.3862,
+      "step": 8011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.645192416508182e-08,
+      "loss": 3.4169,
+      "step": 8012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.618029789177747e-08,
+      "loss": 3.3276,
+      "step": 8013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5909464427781324e-08,
+      "loss": 3.2976,
+      "step": 8014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5639423816439264e-08,
+      "loss": 3.4201,
+      "step": 8015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.537017610096672e-08,
+      "loss": 3.4942,
+      "step": 8016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.510172132445478e-08,
+      "loss": 3.3441,
+      "step": 8017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4834059529867414e-08,
+      "loss": 3.3546,
+      "step": 8018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4567190760040346e-08,
+      "loss": 3.3771,
+      "step": 8019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.430111505768331e-08,
+      "loss": 3.4061,
+      "step": 8020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4035832465380565e-08,
+      "loss": 3.3528,
+      "step": 8021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.377134302558594e-08,
+      "loss": 3.3849,
+      "step": 8022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.350764678063002e-08,
+      "loss": 3.4174,
+      "step": 8023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.324474377271293e-08,
+      "loss": 3.3914,
+      "step": 8024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.298263404391156e-08,
+      "loss": 3.3796,
+      "step": 8025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2721317636172374e-08,
+      "loss": 3.3797,
+      "step": 8026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2460794591317465e-08,
+      "loss": 3.3383,
+      "step": 8027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.220106495104071e-08,
+      "loss": 3.3843,
+      "step": 8028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.194212875690884e-08,
+      "loss": 3.3924,
+      "step": 8029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1683986050362614e-08,
+      "loss": 3.3508,
+      "step": 8030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1426636872715086e-08,
+      "loss": 3.3198,
+      "step": 8031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1170081265151654e-08,
+      "loss": 3.4319,
+      "step": 8032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.091431926873224e-08,
+      "loss": 3.3997,
+      "step": 8033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0659350924388555e-08,
+      "loss": 3.3074,
+      "step": 8034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0405176272926284e-08,
+      "loss": 3.3522,
+      "step": 8035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.015179535502345e-08,
+      "loss": 3.3992,
+      "step": 8036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9899208211230945e-08,
+      "loss": 3.3167,
+      "step": 8037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.964741488197199e-08,
+      "loss": 3.4404,
+      "step": 8038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.939641540754546e-08,
+      "loss": 3.4335,
+      "step": 8039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9146209828119233e-08,
+      "loss": 3.5174,
+      "step": 8040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8896798183737926e-08,
+      "loss": 3.3399,
+      "step": 8041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.864818051431629e-08,
+      "loss": 3.3036,
+      "step": 8042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840035685964361e-08,
+      "loss": 3.4727,
+      "step": 8043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.815332725938148e-08,
+      "loss": 3.4586,
+      "step": 8044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.79070917530644e-08,
+      "loss": 3.424,
+      "step": 8045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7661650380099726e-08,
+      "loss": 3.3945,
+      "step": 8046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7417003179768265e-08,
+      "loss": 3.4424,
+      "step": 8047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7173150191223696e-08,
+      "loss": 3.3937,
+      "step": 8048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.693009145349147e-08,
+      "loss": 3.3947,
+      "step": 8049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6687827005471024e-08,
+      "loss": 3.4283,
+      "step": 8050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.644635688593468e-08,
+      "loss": 3.453,
+      "step": 8051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.620568113352651e-08,
+      "loss": 3.2976,
+      "step": 8052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5965799786765175e-08,
+      "loss": 3.4354,
+      "step": 8053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5726712884040505e-08,
+      "loss": 3.328,
+      "step": 8054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.548842046361689e-08,
+      "loss": 3.3531,
+      "step": 8055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5250922563629386e-08,
+      "loss": 3.3316,
+      "step": 8056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5014219222088696e-08,
+      "loss": 3.4638,
+      "step": 8057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.477831047687508e-08,
+      "loss": 3.2898,
+      "step": 8058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.454319636574499e-08,
+      "loss": 3.2586,
+      "step": 8059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4308876926325005e-08,
+      "loss": 3.389,
+      "step": 8060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4075352196116244e-08,
+      "loss": 3.3652,
+      "step": 8061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3842622212491575e-08,
+      "loss": 3.3858,
+      "step": 8062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.361068701269732e-08,
+      "loss": 3.4727,
+      "step": 8063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3379546633852125e-08,
+      "loss": 3.351,
+      "step": 8064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.314920111294806e-08,
+      "loss": 3.3718,
+      "step": 8065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.291965048685009e-08,
+      "loss": 3.5028,
+      "step": 8066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.269089479229382e-08,
+      "loss": 3.3474,
+      "step": 8067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.246293406589163e-08,
+      "loss": 3.3722,
+      "step": 8068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.223576834412379e-08,
+      "loss": 3.3867,
+      "step": 8069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2009397663347874e-08,
+      "loss": 3.4953,
+      "step": 8070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.178382205979158e-08,
+      "loss": 3.3797,
+      "step": 8071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1559041569556026e-08,
+      "loss": 3.4203,
+      "step": 8072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.133505622861521e-08,
+      "loss": 3.3826,
+      "step": 8073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.111186607281546e-08,
+      "loss": 3.3531,
+      "step": 8074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0889471137875994e-08,
+      "loss": 3.435,
+      "step": 8075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.066787145938999e-08,
+      "loss": 3.3625,
+      "step": 8076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0447067072820745e-08,
+      "loss": 3.4208,
+      "step": 8077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.022705801350667e-08,
+      "loss": 3.3122,
+      "step": 8078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0007844316658486e-08,
+      "loss": 3.3971,
+      "step": 8079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.978942601735757e-08,
+      "loss": 3.367,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9571803150560962e-08,
+      "loss": 3.3767,
+      "step": 8081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9354975751096916e-08,
+      "loss": 3.4297,
+      "step": 8082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9138943853666e-08,
+      "loss": 3.4031,
+      "step": 8083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8923707492842233e-08,
+      "loss": 3.409,
+      "step": 8084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.870926670307139e-08,
+      "loss": 3.426,
+      "step": 8085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.84956215186738e-08,
+      "loss": 3.4199,
+      "step": 8086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.828277197384044e-08,
+      "loss": 3.4888,
+      "step": 8087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.807071810263573e-08,
+      "loss": 3.2736,
+      "step": 8088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7859459938996415e-08,
+      "loss": 3.3415,
+      "step": 8089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7648997516733777e-08,
+      "loss": 3.3956,
+      "step": 8090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7439330869528658e-08,
+      "loss": 3.4592,
+      "step": 8091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.723046003093699e-08,
+      "loss": 3.4941,
+      "step": 8092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7022385034385924e-08,
+      "loss": 3.4449,
+      "step": 8093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6815105913176042e-08,
+      "loss": 3.378,
+      "step": 8094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.660862270048026e-08,
+      "loss": 3.4072,
+      "step": 8095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.64029354293438e-08,
+      "loss": 3.4652,
+      "step": 8096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.619804413268534e-08,
+      "loss": 3.3398,
+      "step": 8097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599394884329587e-08,
+      "loss": 3.3662,
+      "step": 8098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5790649593838146e-08,
+      "loss": 3.377,
+      "step": 8099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.558814641684837e-08,
+      "loss": 3.3652,
+      "step": 8100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5386439344735614e-08,
+      "loss": 3.3934,
+      "step": 8101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5185528409780722e-08,
+      "loss": 3.4345,
+      "step": 8102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4985413644137426e-08,
+      "loss": 3.4239,
+      "step": 8103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.478609507983176e-08,
+      "loss": 3.483,
+      "step": 8104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.458757274876378e-08,
+      "loss": 3.3055,
+      "step": 8105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4389846682704167e-08,
+      "loss": 3.4577,
+      "step": 8106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.419291691329706e-08,
+      "loss": 3.4408,
+      "step": 8107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3996783472058916e-08,
+      "loss": 3.406,
+      "step": 8108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3801446390379068e-08,
+      "loss": 3.4162,
+      "step": 8109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3606905699519735e-08,
+      "loss": 3.3739,
+      "step": 8110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3413161430614895e-08,
+      "loss": 3.3962,
+      "step": 8111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3220213614670863e-08,
+      "loss": 3.4323,
+      "step": 8112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3028062282568487e-08,
+      "loss": 3.4417,
+      "step": 8113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2836707465057617e-08,
+      "loss": 3.3068,
+      "step": 8114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.264614919276431e-08,
+      "loss": 3.3016,
+      "step": 8115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2456387496184728e-08,
+      "loss": 3.4137,
+      "step": 8116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2267422405688467e-08,
+      "loss": 3.427,
+      "step": 8117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2079253951517442e-08,
+      "loss": 3.468,
+      "step": 8118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.189188216378646e-08,
+      "loss": 3.4501,
+      "step": 8119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1705307072482085e-08,
+      "loss": 3.4137,
+      "step": 8120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1519528707464875e-08,
+      "loss": 3.2783,
+      "step": 8121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1334547098464943e-08,
+      "loss": 3.4301,
+      "step": 8122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1150362275088042e-08,
+      "loss": 3.3858,
+      "step": 8123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0966974266810604e-08,
+      "loss": 3.423,
+      "step": 8124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0784383102982477e-08,
+      "loss": 3.3557,
+      "step": 8125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0602588812825287e-08,
+      "loss": 3.519,
+      "step": 8126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042159142543354e-08,
+      "loss": 3.4067,
+      "step": 8127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.024139096977351e-08,
+      "loss": 3.3837,
+      "step": 8128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0061987474685463e-08,
+      "loss": 3.5269,
+      "step": 8129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9883380968880316e-08,
+      "loss": 3.454,
+      "step": 8130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9705571480942988e-08,
+      "loss": 3.4408,
+      "step": 8131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9528559039329044e-08,
+      "loss": 3.4318,
+      "step": 8132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9352343672368047e-08,
+      "loss": 3.4057,
+      "step": 8133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9176925408261883e-08,
+      "loss": 3.3113,
+      "step": 8134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.900230427508476e-08,
+      "loss": 3.4218,
+      "step": 8135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8828480300781548e-08,
+      "loss": 3.3533,
+      "step": 8136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8655453513172217e-08,
+      "loss": 3.4452,
+      "step": 8137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8483223939947948e-08,
+      "loss": 3.449,
+      "step": 8138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8311791608672246e-08,
+      "loss": 3.5629,
+      "step": 8139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8141156546780947e-08,
+      "loss": 3.3567,
+      "step": 8140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.797131878158276e-08,
+      "loss": 3.332,
+      "step": 8141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.780227834025816e-08,
+      "loss": 3.4686,
+      "step": 8142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7634035249860514e-08,
+      "loss": 3.3912,
+      "step": 8143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.746658953731606e-08,
+      "loss": 3.3821,
+      "step": 8144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.729994122942169e-08,
+      "loss": 3.3637,
+      "step": 8145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.71340903528483e-08,
+      "loss": 3.3354,
+      "step": 8146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6969036934139095e-08,
+      "loss": 3.3117,
+      "step": 8147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6804780999709058e-08,
+      "loss": 3.3946,
+      "step": 8148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6641322575844943e-08,
+      "loss": 3.4616,
+      "step": 8149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6478661688707487e-08,
+      "loss": 3.3881,
+      "step": 8150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6316798364328645e-08,
+      "loss": 3.4404,
+      "step": 8151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6155732628612697e-08,
+      "loss": 3.3873,
+      "step": 8152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5995464507337356e-08,
+      "loss": 3.396,
+      "step": 8153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5835994026151545e-08,
+      "loss": 3.3958,
+      "step": 8154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5677321210576525e-08,
+      "loss": 3.304,
+      "step": 8155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5519446086006972e-08,
+      "loss": 3.438,
+      "step": 8156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5362368677708795e-08,
+      "loss": 3.3941,
+      "step": 8157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.520608901082077e-08,
+      "loss": 3.4059,
+      "step": 8158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5050607110353997e-08,
+      "loss": 3.4639,
+      "step": 8159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4895923001191914e-08,
+      "loss": 3.3504,
+      "step": 8160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4742036708089713e-08,
+      "loss": 3.3976,
+      "step": 8161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4588948255675473e-08,
+      "loss": 3.4466,
+      "step": 8162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4436657668449593e-08,
+      "loss": 3.3923,
+      "step": 8163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4285164970785359e-08,
+      "loss": 3.4011,
+      "step": 8164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4134470186926153e-08,
+      "loss": 3.4161,
+      "step": 8165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.398457334099046e-08,
+      "loss": 3.4057,
+      "step": 8166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3835474456966868e-08,
+      "loss": 3.312,
+      "step": 8167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.368717355871796e-08,
+      "loss": 3.3458,
+      "step": 8168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3539670669976967e-08,
+      "loss": 3.4406,
+      "step": 8169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3392965814351122e-08,
+      "loss": 3.4417,
+      "step": 8170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3247059015318864e-08,
+      "loss": 3.456,
+      "step": 8171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3101950296230403e-08,
+      "loss": 3.3334,
+      "step": 8172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2957639680309942e-08,
+      "loss": 3.3245,
+      "step": 8173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2814127190652337e-08,
+      "loss": 3.4168,
+      "step": 8174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2671412850225329e-08,
+      "loss": 3.4026,
+      "step": 8175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.252949668186898e-08,
+      "loss": 3.434,
+      "step": 8176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2388378708295678e-08,
+      "loss": 3.4625,
+      "step": 8177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2248058952090136e-08,
+      "loss": 3.404,
+      "step": 8178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2108537435708834e-08,
+      "loss": 3.4212,
+      "step": 8179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1969814181480577e-08,
+      "loss": 3.3196,
+      "step": 8180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.183188921160705e-08,
+      "loss": 3.398,
+      "step": 8181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.169476254816171e-08,
+      "loss": 3.329,
+      "step": 8182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1558434213090331e-08,
+      "loss": 3.4433,
+      "step": 8183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1422904228210464e-08,
+      "loss": 3.4412,
+      "step": 8184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.128817261521309e-08,
+      "loss": 3.3371,
+      "step": 8185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.115423939565985e-08,
+      "loss": 3.4669,
+      "step": 8186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1021104590985821e-08,
+      "loss": 3.3795,
+      "step": 8187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0888768222497848e-08,
+      "loss": 3.3865,
+      "step": 8188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.07572303113751e-08,
+      "loss": 3.4131,
+      "step": 8189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0626490878669071e-08,
+      "loss": 3.5311,
+      "step": 8190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.049654994530247e-08,
+      "loss": 3.3203,
+      "step": 8191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0367407532071993e-08,
+      "loss": 3.3917,
+      "step": 8192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0239063659645554e-08,
+      "loss": 3.4422,
+      "step": 8193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0111518348562833e-08,
+      "loss": 3.429,
+      "step": 8194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.98477161923639e-09,
+      "loss": 3.3027,
+      "step": 8195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.858823491950552e-09,
+      "loss": 3.3738,
+      "step": 8196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.733673986862536e-09,
+      "loss": 3.5238,
+      "step": 8197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.609323124001313e-09,
+      "loss": 3.4413,
+      "step": 8198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.48577092326708e-09,
+      "loss": 3.3402,
+      "step": 8199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.363017404433461e-09,
+      "loss": 3.4539,
+      "step": 8200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.241062587146965e-09,
+      "loss": 3.4017,
+      "step": 8201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.119906490924202e-09,
+      "loss": 3.3856,
+      "step": 8202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.999549135154661e-09,
+      "loss": 3.3247,
+      "step": 8203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.879990539101825e-09,
+      "loss": 3.3646,
+      "step": 8204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.761230721898161e-09,
+      "loss": 3.3124,
+      "step": 8205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.643269702551248e-09,
+      "loss": 3.4078,
+      "step": 8206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.526107499939317e-09,
+      "loss": 3.4398,
+      "step": 8207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.409744132812369e-09,
+      "loss": 3.4374,
+      "step": 8208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.29417961979384e-09,
+      "loss": 3.439,
+      "step": 8209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.17941397937838e-09,
+      "loss": 3.3577,
+      "step": 8210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.065447229932965e-09,
+      "loss": 3.3421,
+      "step": 8211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.952279389696893e-09,
+      "loss": 3.3439,
+      "step": 8212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.839910476782341e-09,
+      "loss": 3.3835,
+      "step": 8213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.728340509171594e-09,
+      "loss": 3.5001,
+      "step": 8214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.617569504721478e-09,
+      "loss": 3.3997,
+      "step": 8215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.507597481158368e-09,
+      "loss": 3.3959,
+      "step": 8216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.398424456083741e-09,
+      "loss": 3.2891,
+      "step": 8217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.290050446968622e-09,
+      "loss": 3.4263,
+      "step": 8218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.182475471157468e-09,
+      "loss": 3.3157,
+      "step": 8219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.07569954586651e-09,
+      "loss": 3.3928,
+      "step": 8220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.9697226881843e-09,
+      "loss": 3.3446,
+      "step": 8221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.864544915071158e-09,
+      "loss": 3.3627,
+      "step": 8222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.760166243360289e-09,
+      "loss": 3.3897,
+      "step": 8223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.656586689755551e-09,
+      "loss": 3.3575,
+      "step": 8224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.553806270834795e-09,
+      "loss": 3.3616,
+      "step": 8225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.45182500304653e-09,
+      "loss": 3.5377,
+      "step": 8226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.35064290271159e-09,
+      "loss": 3.3368,
+      "step": 8227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.250259986023688e-09,
+      "loss": 3.3946,
+      "step": 8228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.15067626904775e-09,
+      "loss": 3.5265,
+      "step": 8229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.05189176772103e-09,
+      "loss": 3.4405,
+      "step": 8230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.953906497853101e-09,
+      "loss": 3.3525,
+      "step": 8231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.856720475125865e-09,
+      "loss": 3.3958,
+      "step": 8232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.760333715092992e-09,
+      "loss": 3.3535,
+      "step": 8233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.664746233179918e-09,
+      "loss": 3.3323,
+      "step": 8234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.569958044683854e-09,
+      "loss": 3.3425,
+      "step": 8235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.475969164775996e-09,
+      "loss": 3.4466,
+      "step": 8236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.382779608497646e-09,
+      "loss": 3.4486,
+      "step": 8237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.29038939076243e-09,
+      "loss": 3.3546,
+      "step": 8238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.198798526357407e-09,
+      "loss": 3.5818,
+      "step": 8239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1080070299402985e-09,
+      "loss": 3.3204,
+      "step": 8240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.018014916041148e-09,
+      "loss": 3.4484,
+      "step": 8241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.928822199062877e-09,
+      "loss": 3.4169,
+      "step": 8242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.84042889327907e-09,
+      "loss": 3.3727,
+      "step": 8243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.752835012837298e-09,
+      "loss": 3.41,
+      "step": 8244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.666040571754682e-09,
+      "loss": 3.3726,
+      "step": 8245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.580045583923443e-09,
+      "loss": 3.3669,
+      "step": 8246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.494850063104794e-09,
+      "loss": 3.3438,
+      "step": 8247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.410454022933941e-09,
+      "loss": 3.3272,
+      "step": 8248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.326857476917856e-09,
+      "loss": 3.3025,
+      "step": 8249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.244060438435282e-09,
+      "loss": 3.3608,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.162062920736731e-09,
+      "loss": 3.3546,
+      "step": 8251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.080864936945595e-09,
+      "loss": 3.4633,
+      "step": 8252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.000466500055922e-09,
+      "loss": 3.3551,
+      "step": 8253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.920867622935753e-09,
+      "loss": 3.412,
+      "step": 8254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8420683183232335e-09,
+      "loss": 3.3978,
+      "step": 8255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.764068598830495e-09,
+      "loss": 3.4137,
+      "step": 8256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6868684769392203e-09,
+      "loss": 3.3674,
+      "step": 8257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6104679650050824e-09,
+      "loss": 3.419,
+      "step": 8258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5348670752555216e-09,
+      "loss": 3.4099,
+      "step": 8259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4600658197897485e-09,
+      "loss": 3.5098,
+      "step": 8260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3860642105781882e-09,
+      "loss": 3.3474,
+      "step": 8261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3128622594647e-09,
+      "loss": 3.467,
+      "step": 8262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2404599781643565e-09,
+      "loss": 3.3859,
+      "step": 8263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.168857378264556e-09,
+      "loss": 3.388,
+      "step": 8264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0980544712244655e-09,
+      "loss": 3.3697,
+      "step": 8265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.028051268375576e-09,
+      "loss": 3.393,
+      "step": 8266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.958847780921148e-09,
+      "loss": 3.3846,
+      "step": 8267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8904440199356564e-09,
+      "loss": 3.4313,
+      "step": 8268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8228399963675655e-09,
+      "loss": 3.3865,
+      "step": 8269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.756035721035999e-09,
+      "loss": 3.3999,
+      "step": 8270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6900312046318487e-09,
+      "loss": 3.3989,
+      "step": 8271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6248264577188876e-09,
+      "loss": 3.4854,
+      "step": 8272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.560421490732656e-09,
+      "loss": 3.5021,
+      "step": 8273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.496816313979911e-09,
+      "loss": 3.5175,
+      "step": 8274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4340109376402853e-09,
+      "loss": 3.3725,
+      "step": 8275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3720053717651846e-09,
+      "loss": 3.3613,
+      "step": 8276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.310799626277782e-09,
+      "loss": 3.4202,
+      "step": 8277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2503937109741304e-09,
+      "loss": 3.3084,
+      "step": 8278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1907876355209413e-09,
+      "loss": 3.3909,
+      "step": 8279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.131981409457806e-09,
+      "loss": 3.3279,
+      "step": 8280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0739750421960857e-09,
+      "loss": 3.403,
+      "step": 8281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0167685430189098e-09,
+      "loss": 3.3178,
+      "step": 8282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9603619210817325e-09,
+      "loss": 3.3514,
+      "step": 8283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9047551854117774e-09,
+      "loss": 3.5304,
+      "step": 8284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.849948344908592e-09,
+      "loss": 3.4003,
+      "step": 8285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7959414083434935e-09,
+      "loss": 3.3713,
+      "step": 8286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7427343843595679e-09,
+      "loss": 3.3301,
+      "step": 8287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.690327281471671e-09,
+      "loss": 3.3596,
+      "step": 8288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6387201080680926e-09,
+      "loss": 3.4088,
+      "step": 8289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5879128724066718e-09,
+      "loss": 3.4343,
+      "step": 8290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.537905582619792e-09,
+      "loss": 3.3748,
+      "step": 8291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.488698246710496e-09,
+      "loss": 3.3045,
+      "step": 8292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4402908725530406e-09,
+      "loss": 3.4812,
+      "step": 8293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3926834678951173e-09,
+      "loss": 3.3752,
+      "step": 8294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3458760403561865e-09,
+      "loss": 3.384,
+      "step": 8295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.299868597426368e-09,
+      "loss": 3.4033,
+      "step": 8296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.254661146469771e-09,
+      "loss": 3.3737,
+      "step": 8297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2102536947206088e-09,
+      "loss": 3.4217,
+      "step": 8298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1666462492859743e-09,
+      "loss": 3.3521,
+      "step": 8299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.123838817144729e-09,
+      "loss": 3.4237,
+      "step": 8300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0818314051486145e-09,
+      "loss": 3.4223,
+      "step": 8301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0406240200194762e-09,
+      "loss": 3.3971,
+      "step": 8302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0002166683525939e-09,
+      "loss": 3.3196,
+      "step": 8303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.606093566144613e-10,
+      "loss": 3.3696,
+      "step": 8304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.218020911444525e-10,
+      "loss": 3.3831,
+      "step": 8305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.837948781525996e-10,
+      "loss": 3.368,
+      "step": 8306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.465877237218145e-10,
+      "loss": 3.408,
+      "step": 8307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.101806338073337e-10,
+      "loss": 3.4649,
+      "step": 8308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.745736142344973e-10,
+      "loss": 3.3819,
+      "step": 8309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.397666707031903e-10,
+      "loss": 3.3977,
+      "step": 8310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.057598087828466e-10,
+      "loss": 3.4683,
+      "step": 8311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.725530339163345e-10,
+      "loss": 3.3839,
+      "step": 8312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.401463514182915e-10,
+      "loss": 3.3537,
+      "step": 8313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.085397664745696e-10,
+      "loss": 3.477,
+      "step": 8314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.777332841444549e-10,
+      "loss": 3.3781,
+      "step": 8315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.477269093567827e-10,
+      "loss": 3.4587,
+      "step": 8316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.185206469149329e-10,
+      "loss": 3.3078,
+      "step": 8317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.901145014929442e-10,
+      "loss": 3.4327,
+      "step": 8318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.625084776366251e-10,
+      "loss": 3.45,
+      "step": 8319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3570257976355277e-10,
+      "loss": 3.5292,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.096968121647393e-10,
+      "loss": 3.3593,
+      "step": 8321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8449117900185575e-10,
+      "loss": 3.4134,
+      "step": 8322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6008568430834223e-10,
+      "loss": 3.2419,
+      "step": 8323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3648033199051857e-10,
+      "loss": 3.3705,
+      "step": 8324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1367512582591853e-10,
+      "loss": 3.5076,
+      "step": 8325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9167006946440033e-10,
+      "loss": 3.3783,
+      "step": 8326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.704651664275915e-10,
+      "loss": 3.409,
+      "step": 8327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5006042010944367e-10,
+      "loss": 3.4945,
+      "step": 8328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3045583377512282e-10,
+      "loss": 3.3542,
+      "step": 8329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1165141056267435e-10,
+      "loss": 3.3666,
+      "step": 8330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9364715348080266e-10,
+      "loss": 3.4228,
+      "step": 8331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7644306541109157e-10,
+      "loss": 3.4501,
+      "step": 8332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6003914910744933e-10,
+      "loss": 3.4515,
+      "step": 8333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4443540719499828e-10,
+      "loss": 3.4505,
+      "step": 8334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2963184217007486e-10,
+      "loss": 3.4066,
+      "step": 8335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1562845640300524e-10,
+      "loss": 3.4289,
+      "step": 8336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.024252521342195e-10,
+      "loss": 3.3431,
+      "step": 8337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.002223147702715e-11,
+      "loss": 3.4867,
+      "step": 8338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.841939641650697e-11,
+      "loss": 3.3651,
+      "step": 8339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.761674880950697e-11,
+      "loss": 3.4946,
+      "step": 8340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.761429038464439e-11,
+      "loss": 3.291,
+      "step": 8341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.841202274230572e-11,
+      "loss": 3.4418,
+      "step": 8342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.000994735631203e-11,
+      "loss": 3.3565,
+      "step": 8343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.240806557058829e-11,
+      "loss": 3.3524,
+      "step": 8344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5606378602494043e-11,
+      "loss": 3.3708,
+      "step": 8345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9604887539492746e-11,
+      "loss": 3.4234,
+      "step": 8346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4403593342482424e-11,
+      "loss": 3.392,
+      "step": 8347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0002496844130349e-11,
+      "loss": 3.3468,
+      "step": 8348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.4015987488730284e-12,
+      "loss": 3.3301,
+      "step": 8349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6008996323610988e-12,
+      "loss": 3.3747,
+      "step": 8350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6003999431246643e-12,
+      "loss": 3.397,
+      "step": 8351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.001000020181778e-13,
+      "loss": 3.4441,
+      "step": 8352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.4411,
+      "step": 8353
+    },
+    {
+      "epoch": 1.0,
+      "step": 8353,
+      "total_flos": 1.0769385714635244e+18,
+      "train_loss": 3.426367837997678,
+      "train_runtime": 9211.7603,
+      "train_samples_per_second": 58.029,
+      "train_steps_per_second": 0.907
+    }
+  ],
+  "max_steps": 8353,
+  "num_train_epochs": 1,
+  "total_flos": 1.0769385714635244e+18,
+  "trial_name": null,
+  "trial_params": null
+}