diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,64951 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 10821,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846154e-08,
+      "loss": 1.1249,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2307692307692308e-07,
+      "loss": 1.1321,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8461538461538464e-07,
+      "loss": 1.1957,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4615384615384616e-07,
+      "loss": 1.1259,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 1.1193,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6923076923076927e-07,
+      "loss": 1.0855,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.307692307692308e-07,
+      "loss": 1.0972,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.923076923076923e-07,
+      "loss": 1.1719,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.53846153846154e-07,
+      "loss": 1.1438,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 1.1164,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.769230769230769e-07,
+      "loss": 1.0768,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.384615384615385e-07,
+      "loss": 1.097,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.0531,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.615384615384616e-07,
+      "loss": 1.111,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 1.0811,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.846153846153847e-07,
+      "loss": 1.0351,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0461538461538463e-06,
+      "loss": 1.0275,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.107692307692308e-06,
+      "loss": 0.955,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1692307692307693e-06,
+      "loss": 1.0268,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 0.9784,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2923076923076924e-06,
+      "loss": 0.9726,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3538461538461538e-06,
+      "loss": 0.9291,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4153846153846155e-06,
+      "loss": 0.8857,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.476923076923077e-06,
+      "loss": 0.916,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.846,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.8468,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6615384615384616e-06,
+      "loss": 0.9231,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7230769230769232e-06,
+      "loss": 0.8438,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7846153846153846e-06,
+      "loss": 0.8992,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 0.8684,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.907692307692308e-06,
+      "loss": 0.8325,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9692307692307693e-06,
+      "loss": 0.8077,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.030769230769231e-06,
+      "loss": 0.7927,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0923076923076926e-06,
+      "loss": 0.8453,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 0.8529,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.215384615384616e-06,
+      "loss": 0.8431,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.276923076923077e-06,
+      "loss": 0.8282,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3384615384615387e-06,
+      "loss": 0.8626,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.7802,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 0.7509,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.523076923076923e-06,
+      "loss": 0.827,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.584615384615385e-06,
+      "loss": 0.7209,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6461538461538464e-06,
+      "loss": 0.78,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7076923076923076e-06,
+      "loss": 0.8494,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 0.7931,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.830769230769231e-06,
+      "loss": 0.7174,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8923076923076925e-06,
+      "loss": 0.7911,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.953846153846154e-06,
+      "loss": 0.7308,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0153846153846154e-06,
+      "loss": 0.731,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.7054,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1384615384615386e-06,
+      "loss": 0.7264,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.7527,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2615384615384615e-06,
+      "loss": 0.7807,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.323076923076923e-06,
+      "loss": 0.7979,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 0.7586,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4461538461538464e-06,
+      "loss": 0.7142,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.507692307692308e-06,
+      "loss": 0.7729,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5692307692307692e-06,
+      "loss": 0.7967,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.630769230769231e-06,
+      "loss": 0.8413,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 0.7511,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.753846153846154e-06,
+      "loss": 0.6662,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.815384615384616e-06,
+      "loss": 0.6758,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876923076923077e-06,
+      "loss": 0.7459,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.938461538461539e-06,
+      "loss": 0.7279,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7465,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.061538461538462e-06,
+      "loss": 0.7743,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.123076923076924e-06,
+      "loss": 0.7716,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.184615384615385e-06,
+      "loss": 0.7071,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.246153846153846e-06,
+      "loss": 0.751,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 0.7311,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.36923076923077e-06,
+      "loss": 0.7448,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.430769230769232e-06,
+      "loss": 0.7286,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.492307692307693e-06,
+      "loss": 0.6689,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.553846153846154e-06,
+      "loss": 0.7365,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.757,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.676923076923077e-06,
+      "loss": 0.7288,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.738461538461539e-06,
+      "loss": 0.7794,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.7103,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.861538461538462e-06,
+      "loss": 0.6367,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.7253,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.984615384615385e-06,
+      "loss": 0.6488,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.046153846153846e-06,
+      "loss": 0.7549,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1076923076923075e-06,
+      "loss": 0.7151,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.16923076923077e-06,
+      "loss": 0.7133,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 0.7219,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.292307692307693e-06,
+      "loss": 0.7033,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.353846153846154e-06,
+      "loss": 0.7817,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.415384615384615e-06,
+      "loss": 0.7081,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.476923076923077e-06,
+      "loss": 0.6959,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.7105,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.6802,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.661538461538462e-06,
+      "loss": 0.6957,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.723076923076923e-06,
+      "loss": 0.6904,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.784615384615385e-06,
+      "loss": 0.6803,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 0.6963,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.907692307692308e-06,
+      "loss": 0.6843,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9692307692307695e-06,
+      "loss": 0.7316,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.030769230769231e-06,
+      "loss": 0.6549,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.092307692307693e-06,
+      "loss": 0.7497,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.621,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.215384615384615e-06,
+      "loss": 0.6512,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.276923076923077e-06,
+      "loss": 0.7101,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3384615384615385e-06,
+      "loss": 0.7158,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.6504,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 0.6839,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.523076923076923e-06,
+      "loss": 0.6776,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.584615384615385e-06,
+      "loss": 0.7361,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.646153846153846e-06,
+      "loss": 0.6444,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.707692307692308e-06,
+      "loss": 0.6777,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.6382,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.830769230769231e-06,
+      "loss": 0.705,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.892307692307693e-06,
+      "loss": 0.6523,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.953846153846154e-06,
+      "loss": 0.6977,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.015384615384616e-06,
+      "loss": 0.7095,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 0.667,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1384615384615385e-06,
+      "loss": 0.683,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.7205,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.261538461538462e-06,
+      "loss": 0.6655,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.323076923076924e-06,
+      "loss": 0.6877,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.7027,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.446153846153846e-06,
+      "loss": 0.7937,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.507692307692308e-06,
+      "loss": 0.6917,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5692307692307695e-06,
+      "loss": 0.6615,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.630769230769232e-06,
+      "loss": 0.6398,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.6389,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.753846153846155e-06,
+      "loss": 0.683,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.815384615384616e-06,
+      "loss": 0.6548,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.876923076923077e-06,
+      "loss": 0.6127,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.93846153846154e-06,
+      "loss": 0.6898,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6386,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.061538461538463e-06,
+      "loss": 0.7114,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.123076923076924e-06,
+      "loss": 0.6004,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.184615384615385e-06,
+      "loss": 0.6833,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.246153846153848e-06,
+      "loss": 0.6766,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.6989,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.36923076923077e-06,
+      "loss": 0.6651,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.430769230769231e-06,
+      "loss": 0.7119,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.492307692307693e-06,
+      "loss": 0.6692,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.553846153846156e-06,
+      "loss": 0.6857,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.7141,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.676923076923078e-06,
+      "loss": 0.6448,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.73846153846154e-06,
+      "loss": 0.6874,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 0.7084,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.861538461538463e-06,
+      "loss": 0.7026,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.7614,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.984615384615386e-06,
+      "loss": 0.7179,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.046153846153847e-06,
+      "loss": 0.7115,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.107692307692308e-06,
+      "loss": 0.6714,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.169230769230771e-06,
+      "loss": 0.6527,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.6779,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.292307692307694e-06,
+      "loss": 0.6252,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.353846153846155e-06,
+      "loss": 0.7213,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.415384615384616e-06,
+      "loss": 0.6947,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.476923076923079e-06,
+      "loss": 0.6811,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.6947,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.6705,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.661538461538462e-06,
+      "loss": 0.6276,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.723076923076924e-06,
+      "loss": 0.6765,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.784615384615387e-06,
+      "loss": 0.6593,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.6574,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.907692307692309e-06,
+      "loss": 0.6474,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96923076923077e-06,
+      "loss": 0.6151,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0030769230769231e-05,
+      "loss": 0.6488,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0092307692307693e-05,
+      "loss": 0.7123,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.6232,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0215384615384615e-05,
+      "loss": 0.6606,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0276923076923078e-05,
+      "loss": 0.7054,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.033846153846154e-05,
+      "loss": 0.6917,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.04e-05,
+      "loss": 0.6791,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.6627,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0523076923076924e-05,
+      "loss": 0.6059,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0584615384615386e-05,
+      "loss": 0.6593,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0646153846153845e-05,
+      "loss": 0.6921,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0707692307692308e-05,
+      "loss": 0.7043,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.7157,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.083076923076923e-05,
+      "loss": 0.6672,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0892307692307693e-05,
+      "loss": 0.6859,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0953846153846155e-05,
+      "loss": 0.6764,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1015384615384616e-05,
+      "loss": 0.6876,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.7226,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.113846153846154e-05,
+      "loss": 0.7655,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.6413,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.126153846153846e-05,
+      "loss": 0.6765,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1323076923076924e-05,
+      "loss": 0.6987,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.6556,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1446153846153846e-05,
+      "loss": 0.7041,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1507692307692309e-05,
+      "loss": 0.6976,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.156923076923077e-05,
+      "loss": 0.6902,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1630769230769231e-05,
+      "loss": 0.689,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.6352,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1753846153846155e-05,
+      "loss": 0.7049,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1815384615384617e-05,
+      "loss": 0.6464,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1876923076923076e-05,
+      "loss": 0.6687,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1938461538461539e-05,
+      "loss": 0.6481,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6277,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2061538461538462e-05,
+      "loss": 0.6359,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2123076923076924e-05,
+      "loss": 0.6779,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2184615384615386e-05,
+      "loss": 0.5707,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2246153846153847e-05,
+      "loss": 0.6992,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.6224,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2369230769230771e-05,
+      "loss": 0.6143,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.243076923076923e-05,
+      "loss": 0.6836,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2492307692307692e-05,
+      "loss": 0.6948,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2553846153846155e-05,
+      "loss": 0.6297,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.656,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2676923076923077e-05,
+      "loss": 0.6103,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.273846153846154e-05,
+      "loss": 0.6449,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.6541,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2861538461538462e-05,
+      "loss": 0.635,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.6986,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2984615384615386e-05,
+      "loss": 0.5922,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3046153846153846e-05,
+      "loss": 0.6252,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3107692307692307e-05,
+      "loss": 0.6359,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.316923076923077e-05,
+      "loss": 0.6921,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.7077,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3292307692307692e-05,
+      "loss": 0.6839,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3353846153846155e-05,
+      "loss": 0.6378,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3415384615384617e-05,
+      "loss": 0.6783,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3476923076923078e-05,
+      "loss": 0.694,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.6663,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.6621,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3661538461538461e-05,
+      "loss": 0.6658,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3723076923076923e-05,
+      "loss": 0.7005,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3784615384615386e-05,
+      "loss": 0.6365,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.6724,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3907692307692308e-05,
+      "loss": 0.6604,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3969230769230771e-05,
+      "loss": 0.7287,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4030769230769232e-05,
+      "loss": 0.688,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4092307692307693e-05,
+      "loss": 0.6501,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.6676,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4215384615384617e-05,
+      "loss": 0.6333,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4276923076923077e-05,
+      "loss": 0.6642,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4338461538461538e-05,
+      "loss": 0.6577,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6351,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.6598,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4523076923076923e-05,
+      "loss": 0.6897,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4584615384615386e-05,
+      "loss": 0.6742,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4646153846153848e-05,
+      "loss": 0.6039,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4707692307692309e-05,
+      "loss": 0.6512,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.6605,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4830769230769233e-05,
+      "loss": 0.6477,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4892307692307692e-05,
+      "loss": 0.6723,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4953846153846154e-05,
+      "loss": 0.6665,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5015384615384617e-05,
+      "loss": 0.6833,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.6164,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5138461538461539e-05,
+      "loss": 0.7332,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.6577,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5261538461538465e-05,
+      "loss": 0.6977,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5323076923076926e-05,
+      "loss": 0.6556,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.6353,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.544615384615385e-05,
+      "loss": 0.598,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.550769230769231e-05,
+      "loss": 0.7091,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.556923076923077e-05,
+      "loss": 0.6819,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5630769230769232e-05,
+      "loss": 0.7,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.7283,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5753846153846154e-05,
+      "loss": 0.6576,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5815384615384616e-05,
+      "loss": 0.6729,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.587692307692308e-05,
+      "loss": 0.6542,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.593846153846154e-05,
+      "loss": 0.7363,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6778,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.606153846153846e-05,
+      "loss": 0.6622,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6123076923076925e-05,
+      "loss": 0.6867,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6184615384615386e-05,
+      "loss": 0.6447,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6246153846153848e-05,
+      "loss": 0.7,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.6244,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.636923076923077e-05,
+      "loss": 0.6542,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.643076923076923e-05,
+      "loss": 0.6039,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6492307692307696e-05,
+      "loss": 0.6518,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6553846153846157e-05,
+      "loss": 0.6103,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.6681,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6676923076923076e-05,
+      "loss": 0.6611,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.673846153846154e-05,
+      "loss": 0.6837,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.6229,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6861538461538463e-05,
+      "loss": 0.6506,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.6975,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6984615384615385e-05,
+      "loss": 0.6609,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7046153846153847e-05,
+      "loss": 0.6704,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.710769230769231e-05,
+      "loss": 0.6802,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7169230769230772e-05,
+      "loss": 0.633,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.6491,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.729230769230769e-05,
+      "loss": 0.6608,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7353846153846156e-05,
+      "loss": 0.7327,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7415384615384617e-05,
+      "loss": 0.6312,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.747692307692308e-05,
+      "loss": 0.6561,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.7053,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 0.6327,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7661538461538462e-05,
+      "loss": 0.6707,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7723076923076927e-05,
+      "loss": 0.7461,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7784615384615388e-05,
+      "loss": 0.7292,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.651,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7907692307692307e-05,
+      "loss": 0.644,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.796923076923077e-05,
+      "loss": 0.6565,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8030769230769233e-05,
+      "loss": 0.6759,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8092307692307694e-05,
+      "loss": 0.6791,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.6602,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8215384615384616e-05,
+      "loss": 0.6685,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8276923076923078e-05,
+      "loss": 0.6159,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8338461538461542e-05,
+      "loss": 0.6373,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.6431,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.6148,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8523076923076922e-05,
+      "loss": 0.7623,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8584615384615387e-05,
+      "loss": 0.642,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8646153846153848e-05,
+      "loss": 0.7105,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.870769230769231e-05,
+      "loss": 0.6276,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.6439,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8830769230769232e-05,
+      "loss": 0.5637,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8892307692307693e-05,
+      "loss": 0.671,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8953846153846158e-05,
+      "loss": 0.6597,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.901538461538462e-05,
+      "loss": 0.6704,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.6058,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9138461538461538e-05,
+      "loss": 0.6484,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6983,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9261538461538464e-05,
+      "loss": 0.6224,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9323076923076925e-05,
+      "loss": 0.5968,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 0.6082,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9446153846153847e-05,
+      "loss": 0.6413,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.950769230769231e-05,
+      "loss": 0.6162,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9569230769230773e-05,
+      "loss": 0.6304,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630769230769234e-05,
+      "loss": 0.6713,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.6836,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9753846153846153e-05,
+      "loss": 0.6946,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815384615384618e-05,
+      "loss": 0.6859,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.987692307692308e-05,
+      "loss": 0.6759,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.993846153846154e-05,
+      "loss": 0.6586,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.6918,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999552057673e-05,
+      "loss": 0.7868,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999820823073e-05,
+      "loss": 0.682,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995968519293e-05,
+      "loss": 0.6543,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992832923563e-05,
+      "loss": 0.659,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999988801443817e-05,
+      "loss": 0.6196,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998387408042e-05,
+      "loss": 0.7237,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997805083381e-05,
+      "loss": 0.6326,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999971331704515e-05,
+      "loss": 0.6541,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996371669313e-05,
+      "loss": 0.7029,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995520580034e-05,
+      "loss": 0.6279,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999945799026912e-05,
+      "loss": 0.677,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999993549637368e-05,
+      "loss": 0.6328,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999924297841576e-05,
+      "loss": 0.6437,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999912203431596e-05,
+      "loss": 0.7509,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999899213144825e-05,
+      "loss": 0.6504,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999885326982426e-05,
+      "loss": 0.6742,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999987054494565e-05,
+      "loss": 0.6067,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999854867035812e-05,
+      "loss": 0.628,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999838293254324e-05,
+      "loss": 0.618,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982082360267e-05,
+      "loss": 0.6616,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999802458082408e-05,
+      "loss": 0.6646,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999978319669519e-05,
+      "loss": 0.6787,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999976303944274e-05,
+      "loss": 0.5863,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999741986326865e-05,
+      "loss": 0.7035,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999720037349446e-05,
+      "loss": 0.6394,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999697192512455e-05,
+      "loss": 0.6314,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999967345181794e-05,
+      "loss": 0.6394,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999648815268018e-05,
+      "loss": 0.6921,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999623282864905e-05,
+      "loss": 0.582,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999959685461089e-05,
+      "loss": 0.6345,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999569530508335e-05,
+      "loss": 0.6793,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999541310559686e-05,
+      "loss": 0.6372,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999512194767478e-05,
+      "loss": 0.7065,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999482183134318e-05,
+      "loss": 0.6593,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945127566289e-05,
+      "loss": 0.6949,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999941947235597e-05,
+      "loss": 0.7116,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99993867732164e-05,
+      "loss": 0.5891,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999935317824711e-05,
+      "loss": 0.6238,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999318687451116e-05,
+      "loss": 0.6272,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999283300831504e-05,
+      "loss": 0.6657,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999924701839145e-05,
+      "loss": 0.6391,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999920984013419e-05,
+      "loss": 0.6847,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999917176606307e-05,
+      "loss": 0.7211,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999913279618149e-05,
+      "loss": 0.6401,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999092930492948e-05,
+      "loss": 0.6258,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999052169001015e-05,
+      "loss": 0.696,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999901051170934e-05,
+      "loss": 0.5999,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999896795862166e-05,
+      "loss": 0.6229,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998924509741783e-05,
+      "loss": 0.611,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998880165073597e-05,
+      "loss": 0.6091,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998834924621086e-05,
+      "loss": 0.6564,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998788788388292e-05,
+      "loss": 0.6013,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998741756379355e-05,
+      "loss": 0.6159,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998693828598488e-05,
+      "loss": 0.6367,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999864500504998e-05,
+      "loss": 0.6414,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999859528573821e-05,
+      "loss": 0.6369,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998544670667635e-05,
+      "loss": 0.6199,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998493159842784e-05,
+      "loss": 0.6586,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999844075326827e-05,
+      "loss": 0.6124,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998387450948794e-05,
+      "loss": 0.698,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998333252889125e-05,
+      "loss": 0.615,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998278159094124e-05,
+      "loss": 0.6348,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998222169568726e-05,
+      "loss": 0.6481,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998165284317944e-05,
+      "loss": 0.6393,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998107503346876e-05,
+      "loss": 0.5998,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998048826660702e-05,
+      "loss": 0.6015,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999798925426467e-05,
+      "loss": 0.6207,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997928786164127e-05,
+      "loss": 0.5939,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997867422364482e-05,
+      "loss": 0.6089,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997805162871236e-05,
+      "loss": 0.6772,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999774200768997e-05,
+      "loss": 0.6467,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997677956826334e-05,
+      "loss": 0.598,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997613010286075e-05,
+      "loss": 0.6503,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997547168075008e-05,
+      "loss": 0.5973,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999748043019903e-05,
+      "loss": 0.6436,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997412796664118e-05,
+      "loss": 0.6069,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997344267476338e-05,
+      "loss": 0.5778,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997274842641825e-05,
+      "loss": 0.6258,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99972045221668e-05,
+      "loss": 0.6304,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999713330605756e-05,
+      "loss": 0.6481,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997061194320487e-05,
+      "loss": 0.5956,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996988186962044e-05,
+      "loss": 0.5745,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996914283988768e-05,
+      "loss": 0.6464,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996839485407282e-05,
+      "loss": 0.6511,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996763791224288e-05,
+      "loss": 0.6355,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999668720144656e-05,
+      "loss": 0.6497,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999660971608097e-05,
+      "loss": 0.6328,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996531335134453e-05,
+      "loss": 0.6888,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996452058614032e-05,
+      "loss": 0.5905,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996371886526815e-05,
+      "loss": 0.5428,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996290818879974e-05,
+      "loss": 0.6427,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996208855680783e-05,
+      "loss": 0.7183,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996125996936574e-05,
+      "loss": 0.6617,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999604224265478e-05,
+      "loss": 0.7308,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995957592842902e-05,
+      "loss": 0.584,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995872047508516e-05,
+      "loss": 0.5849,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995785606659296e-05,
+      "loss": 0.61,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999569827030298e-05,
+      "loss": 0.7136,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999561003844739e-05,
+      "loss": 0.6,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999552091110044e-05,
+      "loss": 0.5997,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995430888270106e-05,
+      "loss": 0.6822,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995339969964457e-05,
+      "loss": 0.659,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999524815619164e-05,
+      "loss": 0.646,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995155446959877e-05,
+      "loss": 0.6272,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999506184227747e-05,
+      "loss": 0.6544,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994967342152817e-05,
+      "loss": 0.6892,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994871946594373e-05,
+      "loss": 0.6907,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999477565561069e-05,
+      "loss": 0.6398,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999467846921039e-05,
+      "loss": 0.6243,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994580387402186e-05,
+      "loss": 0.5767,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999448141019486e-05,
+      "loss": 0.6161,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994381537597277e-05,
+      "loss": 0.6177,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994280769618393e-05,
+      "loss": 0.633,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994179106267227e-05,
+      "loss": 0.6223,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994076547552895e-05,
+      "loss": 0.6031,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993973093484578e-05,
+      "loss": 0.6894,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993868744071547e-05,
+      "loss": 0.6066,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999376349932315e-05,
+      "loss": 0.6443,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993657359248818e-05,
+      "loss": 0.6345,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993550323858054e-05,
+      "loss": 0.6438,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993442393160455e-05,
+      "loss": 0.5882,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993333567165686e-05,
+      "loss": 0.5847,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993223845883496e-05,
+      "loss": 0.6732,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993113229323717e-05,
+      "loss": 0.5294,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993001717496255e-05,
+      "loss": 0.6427,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999288931041111e-05,
+      "loss": 0.6548,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999277600807834e-05,
+      "loss": 0.6027,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99926618105081e-05,
+      "loss": 0.6381,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992546717710625e-05,
+      "loss": 0.6103,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992430729696225e-05,
+      "loss": 0.6266,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999231384647528e-05,
+      "loss": 0.6199,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999219606805828e-05,
+      "loss": 0.628,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999207739445576e-05,
+      "loss": 0.6626,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991957825678362e-05,
+      "loss": 0.6615,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991837361736796e-05,
+      "loss": 0.585,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991716002641853e-05,
+      "loss": 0.6568,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991593748404405e-05,
+      "loss": 0.652,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99914705990354e-05,
+      "loss": 0.6452,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999134655454588e-05,
+      "loss": 0.6054,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991221614946953e-05,
+      "loss": 0.6545,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991095780249812e-05,
+      "loss": 0.5595,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990969050465732e-05,
+      "loss": 0.6611,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990841425606064e-05,
+      "loss": 0.5857,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990712905682245e-05,
+      "loss": 0.6001,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990583490705785e-05,
+      "loss": 0.6148,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999045318068828e-05,
+      "loss": 0.6262,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999032197564141e-05,
+      "loss": 0.5861,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990189875576918e-05,
+      "loss": 0.6012,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999005688050665e-05,
+      "loss": 0.6646,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989922990442513e-05,
+      "loss": 0.6189,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99897882053965e-05,
+      "loss": 0.6008,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989652525380695e-05,
+      "loss": 0.6787,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989515950407247e-05,
+      "loss": 0.6758,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989378480488396e-05,
+      "loss": 0.5822,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989240115636455e-05,
+      "loss": 0.6401,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998910085586382e-05,
+      "loss": 0.6334,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998896070118297e-05,
+      "loss": 0.6375,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988819651606454e-05,
+      "loss": 0.643,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988677707146914e-05,
+      "loss": 0.5968,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988534867817065e-05,
+      "loss": 0.6354,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988391133629705e-05,
+      "loss": 0.5885,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988246504597712e-05,
+      "loss": 0.6384,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998810098073404e-05,
+      "loss": 0.6054,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987954562051724e-05,
+      "loss": 0.6889,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998780724856389e-05,
+      "loss": 0.6238,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987659040283728e-05,
+      "loss": 0.5666,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998750993722452e-05,
+      "loss": 0.6389,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998735993939962e-05,
+      "loss": 0.6715,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987209046822472e-05,
+      "loss": 0.6619,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987057259506585e-05,
+      "loss": 0.6449,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986904577465567e-05,
+      "loss": 0.5701,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986751000713092e-05,
+      "loss": 0.6142,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986596529262918e-05,
+      "loss": 0.5883,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986441163128884e-05,
+      "loss": 0.637,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986284902324912e-05,
+      "loss": 0.6449,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986127746865e-05,
+      "loss": 0.6367,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985969696763224e-05,
+      "loss": 0.6463,
+      "step": 502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998581075203375e-05,
+      "loss": 0.5909,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985650912690812e-05,
+      "loss": 0.6119,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998549017874873e-05,
+      "loss": 0.67,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985328550221906e-05,
+      "loss": 0.6293,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985166027124817e-05,
+      "loss": 0.6242,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985002609472025e-05,
+      "loss": 0.5678,
+      "step": 508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984838297278176e-05,
+      "loss": 0.6774,
+      "step": 509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998467309055798e-05,
+      "loss": 0.5627,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998450698932625e-05,
+      "loss": 0.6184,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984339993597854e-05,
+      "loss": 0.5731,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998417210338776e-05,
+      "loss": 0.5943,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998400331871101e-05,
+      "loss": 0.5683,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983833639582717e-05,
+      "loss": 0.6956,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983663066018092e-05,
+      "loss": 0.5967,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998349159803241e-05,
+      "loss": 0.5946,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998331923564104e-05,
+      "loss": 0.6594,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983145978859414e-05,
+      "loss": 0.6344,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998297182770306e-05,
+      "loss": 0.5915,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998279678218758e-05,
+      "loss": 0.5883,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982620842328654e-05,
+      "loss": 0.6346,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982444008142047e-05,
+      "loss": 0.6924,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982266279643595e-05,
+      "loss": 0.6294,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982087656849227e-05,
+      "loss": 0.6206,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981908139774942e-05,
+      "loss": 0.6185,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998172772843682e-05,
+      "loss": 0.6543,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981546422851032e-05,
+      "loss": 0.6087,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981364223033814e-05,
+      "loss": 0.6848,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998118112900149e-05,
+      "loss": 0.5923,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998099714077047e-05,
+      "loss": 0.5987,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998081225835723e-05,
+      "loss": 0.6053,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980626481778335e-05,
+      "loss": 0.6619,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980439811050425e-05,
+      "loss": 0.6111,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998025224619023e-05,
+      "loss": 0.5965,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998006378721455e-05,
+      "loss": 0.6198,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979874434140266e-05,
+      "loss": 0.6139,
+      "step": 537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997968418698435e-05,
+      "loss": 0.5997,
+      "step": 538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997949304576384e-05,
+      "loss": 0.6268,
+      "step": 539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997930101049586e-05,
+      "loss": 0.6152,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979108081197617e-05,
+      "loss": 0.5837,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978914257886392e-05,
+      "loss": 0.6147,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997871954057955e-05,
+      "loss": 0.6557,
+      "step": 543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997852392929454e-05,
+      "loss": 0.6364,
+      "step": 544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978327424048877e-05,
+      "loss": 0.6216,
+      "step": 545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978130024860176e-05,
+      "loss": 0.5784,
+      "step": 546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977931731746117e-05,
+      "loss": 0.601,
+      "step": 547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977732544724464e-05,
+      "loss": 0.6298,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977532463813064e-05,
+      "loss": 0.5801,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997733148902984e-05,
+      "loss": 0.6823,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99771296203928e-05,
+      "loss": 0.6278,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976926857920024e-05,
+      "loss": 0.6138,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997672320162968e-05,
+      "loss": 0.614,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976518651540013e-05,
+      "loss": 0.6141,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997631320766935e-05,
+      "loss": 0.6667,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976106870036095e-05,
+      "loss": 0.6406,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975899638658733e-05,
+      "loss": 0.6316,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997569151355583e-05,
+      "loss": 0.5914,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975482494746032e-05,
+      "loss": 0.575,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975272582248064e-05,
+      "loss": 0.641,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975061776080732e-05,
+      "loss": 0.6447,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997485007626292e-05,
+      "loss": 0.6711,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99746374828136e-05,
+      "loss": 0.5449,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997442399575181e-05,
+      "loss": 0.5888,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974209615096682e-05,
+      "loss": 0.6288,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997399434086742e-05,
+      "loss": 0.572,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973778173083307e-05,
+      "loss": 0.5719,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973561111763715e-05,
+      "loss": 0.6074,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973343156928085e-05,
+      "loss": 0.6288,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973124308595945e-05,
+      "loss": 0.6338,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972904566786903e-05,
+      "loss": 0.6013,
+      "step": 571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972683931520645e-05,
+      "loss": 0.5926,
+      "step": 572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972462402816935e-05,
+      "loss": 0.6239,
+      "step": 573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997223998069562e-05,
+      "loss": 0.5965,
+      "step": 574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997201666517663e-05,
+      "loss": 0.6383,
+      "step": 575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971792456279964e-05,
+      "loss": 0.6264,
+      "step": 576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971567354025715e-05,
+      "loss": 0.5451,
+      "step": 577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997134135843405e-05,
+      "loss": 0.6217,
+      "step": 578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971114469525214e-05,
+      "loss": 0.6102,
+      "step": 579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997088668731953e-05,
+      "loss": 0.6001,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970658011837404e-05,
+      "loss": 0.6704,
+      "step": 581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997042844309933e-05,
+      "loss": 0.6446,
+      "step": 582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970197981125875e-05,
+      "loss": 0.6442,
+      "step": 583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969966625937677e-05,
+      "loss": 0.5724,
+      "step": 584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969734377555472e-05,
+      "loss": 0.5988,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996950123600006e-05,
+      "loss": 0.6362,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969267201292327e-05,
+      "loss": 0.6192,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969032273453246e-05,
+      "loss": 0.6509,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996879645250386e-05,
+      "loss": 0.5963,
+      "step": 589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968559738465297e-05,
+      "loss": 0.6643,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968322131358767e-05,
+      "loss": 0.6718,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968083631205547e-05,
+      "loss": 0.6637,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967844238027015e-05,
+      "loss": 0.6713,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996760395184461e-05,
+      "loss": 0.6083,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967362772679864e-05,
+      "loss": 0.6187,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996712070055438e-05,
+      "loss": 0.6655,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966877735489846e-05,
+      "loss": 0.6292,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966633877508034e-05,
+      "loss": 0.6075,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996638912663078e-05,
+      "loss": 0.5767,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996614348288002e-05,
+      "loss": 0.7054,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965896946277762e-05,
+      "loss": 0.581,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965649516846085e-05,
+      "loss": 0.5996,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965401194607158e-05,
+      "loss": 0.5839,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996515197958323e-05,
+      "loss": 0.6739,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964901871796632e-05,
+      "loss": 0.5798,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964650871269764e-05,
+      "loss": 0.6531,
+      "step": 606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996439897802511e-05,
+      "loss": 0.6347,
+      "step": 607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996414619208525e-05,
+      "loss": 0.6586,
+      "step": 608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963892513472818e-05,
+      "loss": 0.5672,
+      "step": 609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963637942210544e-05,
+      "loss": 0.641,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963382478321237e-05,
+      "loss": 0.5885,
+      "step": 611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996312612182778e-05,
+      "loss": 0.6215,
+      "step": 612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962868872753144e-05,
+      "loss": 0.5743,
+      "step": 613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962610731120376e-05,
+      "loss": 0.5885,
+      "step": 614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962351696952597e-05,
+      "loss": 0.6178,
+      "step": 615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962091770273014e-05,
+      "loss": 0.6107,
+      "step": 616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996183095110492e-05,
+      "loss": 0.5659,
+      "step": 617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961569239471675e-05,
+      "loss": 0.5571,
+      "step": 618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961306635396725e-05,
+      "loss": 0.6659,
+      "step": 619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961043138903603e-05,
+      "loss": 0.5865,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996077875001591e-05,
+      "loss": 0.5629,
+      "step": 621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960513468757332e-05,
+      "loss": 0.5246,
+      "step": 622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960247295151637e-05,
+      "loss": 0.5917,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995998022922267e-05,
+      "loss": 0.5923,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995971227099436e-05,
+      "loss": 0.5928,
+      "step": 625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995944342049071e-05,
+      "loss": 0.6643,
+      "step": 626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959173677735804e-05,
+      "loss": 0.6277,
+      "step": 627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958903042753813e-05,
+      "loss": 0.6682,
+      "step": 628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995863151556898e-05,
+      "loss": 0.628,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958359096205626e-05,
+      "loss": 0.64,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958085784688165e-05,
+      "loss": 0.5781,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995781158104108e-05,
+      "loss": 0.6376,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995753648528893e-05,
+      "loss": 0.5584,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995726049745637e-05,
+      "loss": 0.6436,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956983617568123e-05,
+      "loss": 0.5618,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995670584564899e-05,
+      "loss": 0.5944,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995642718172386e-05,
+      "loss": 0.5889,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995614762581769e-05,
+      "loss": 0.612,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995586717795554e-05,
+      "loss": 0.6515,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955585838162525e-05,
+      "loss": 0.5625,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995530360646385e-05,
+      "loss": 0.587,
+      "step": 641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99550204828848e-05,
+      "loss": 0.5575,
+      "step": 642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995473646745074e-05,
+      "loss": 0.5751,
+      "step": 643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954451560187116e-05,
+      "loss": 0.5953,
+      "step": 644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995416576111945e-05,
+      "loss": 0.6331,
+      "step": 645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995387907027335e-05,
+      "loss": 0.5682,
+      "step": 646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953591487674497e-05,
+      "loss": 0.5674,
+      "step": 647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953303013348656e-05,
+      "loss": 0.6703,
+      "step": 648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953013647321673e-05,
+      "loss": 0.6236,
+      "step": 649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952723389619468e-05,
+      "loss": 0.5652,
+      "step": 650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952432240268048e-05,
+      "loss": 0.5932,
+      "step": 651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952140199293494e-05,
+      "loss": 0.5637,
+      "step": 652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.6031,
+      "step": 653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995155344257972e-05,
+      "loss": 0.5865,
+      "step": 654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995125872689307e-05,
+      "loss": 0.6128,
+      "step": 655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950963119688417e-05,
+      "loss": 0.6403,
+      "step": 656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950666620992253e-05,
+      "loss": 0.5478,
+      "step": 657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995036923083113e-05,
+      "loss": 0.6132,
+      "step": 658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99500709492317e-05,
+      "loss": 0.5774,
+      "step": 659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994977177622068e-05,
+      "loss": 0.6448,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949471711824873e-05,
+      "loss": 0.6235,
+      "step": 661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949170756071163e-05,
+      "loss": 0.6314,
+      "step": 662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948868908986513e-05,
+      "loss": 0.6607,
+      "step": 663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948566170597963e-05,
+      "loss": 0.5914,
+      "step": 664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948262540932634e-05,
+      "loss": 0.6327,
+      "step": 665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994795802001773e-05,
+      "loss": 0.5777,
+      "step": 666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947652607880534e-05,
+      "loss": 0.5978,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947346304548403e-05,
+      "loss": 0.5879,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994703911004878e-05,
+      "loss": 0.6519,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946731024409187e-05,
+      "loss": 0.6379,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946422047657224e-05,
+      "loss": 0.6287,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946112179820576e-05,
+      "loss": 0.5126,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945801420926996e-05,
+      "loss": 0.5736,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994548977100433e-05,
+      "loss": 0.5816,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99451772300805e-05,
+      "loss": 0.6418,
+      "step": 675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944863798183495e-05,
+      "loss": 0.6195,
+      "step": 676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944549475341404e-05,
+      "loss": 0.6008,
+      "step": 677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944234261582386e-05,
+      "loss": 0.6049,
+      "step": 678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994391815693468e-05,
+      "loss": 0.6106,
+      "step": 679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943601161426603e-05,
+      "loss": 0.6235,
+      "step": 680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994328327508656e-05,
+      "loss": 0.5672,
+      "step": 681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942964497943026e-05,
+      "loss": 0.6075,
+      "step": 682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942644830024558e-05,
+      "loss": 0.602,
+      "step": 683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942324271359794e-05,
+      "loss": 0.6706,
+      "step": 684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942002821977457e-05,
+      "loss": 0.5621,
+      "step": 685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941680481906345e-05,
+      "loss": 0.6004,
+      "step": 686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994135725117533e-05,
+      "loss": 0.6416,
+      "step": 687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941033129813374e-05,
+      "loss": 0.5918,
+      "step": 688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940708117849516e-05,
+      "loss": 0.673,
+      "step": 689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994038221531287e-05,
+      "loss": 0.6316,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940055422232635e-05,
+      "loss": 0.5993,
+      "step": 691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993972773863809e-05,
+      "loss": 0.6792,
+      "step": 692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939399164558588e-05,
+      "loss": 0.6202,
+      "step": 693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939069700023564e-05,
+      "loss": 0.6312,
+      "step": 694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938739345062537e-05,
+      "loss": 0.6174,
+      "step": 695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938408099705103e-05,
+      "loss": 0.5793,
+      "step": 696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938075963980936e-05,
+      "loss": 0.6266,
+      "step": 697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937742937919795e-05,
+      "loss": 0.6371,
+      "step": 698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937409021551514e-05,
+      "loss": 0.6349,
+      "step": 699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993707421490601e-05,
+      "loss": 0.5613,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993673851801327e-05,
+      "loss": 0.5894,
+      "step": 701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936401930903375e-05,
+      "loss": 0.588,
+      "step": 702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936064453606477e-05,
+      "loss": 0.5715,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935726086152816e-05,
+      "loss": 0.6346,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935386828572695e-05,
+      "loss": 0.5122,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993504668089652e-05,
+      "loss": 0.622,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934705643154754e-05,
+      "loss": 0.6431,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934363715377954e-05,
+      "loss": 0.602,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934020897596752e-05,
+      "loss": 0.6417,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993367718984186e-05,
+      "loss": 0.6163,
+      "step": 710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933332592144076e-05,
+      "loss": 0.5995,
+      "step": 711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932987104534265e-05,
+      "loss": 0.5723,
+      "step": 712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993264072704338e-05,
+      "loss": 0.5626,
+      "step": 713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932293459702456e-05,
+      "loss": 0.6383,
+      "step": 714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931945302542602e-05,
+      "loss": 0.613,
+      "step": 715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931596255595007e-05,
+      "loss": 0.6854,
+      "step": 716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931246318890943e-05,
+      "loss": 0.6002,
+      "step": 717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993089549246176e-05,
+      "loss": 0.6397,
+      "step": 718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993054377633889e-05,
+      "loss": 0.5888,
+      "step": 719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930191170553838e-05,
+      "loss": 0.5536,
+      "step": 720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99298376751382e-05,
+      "loss": 0.6368,
+      "step": 721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929483290123638e-05,
+      "loss": 0.5788,
+      "step": 722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929128015541908e-05,
+      "loss": 0.6,
+      "step": 723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928771851424835e-05,
+      "loss": 0.5959,
+      "step": 724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928414797804326e-05,
+      "loss": 0.5881,
+      "step": 725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992805685471237e-05,
+      "loss": 0.644,
+      "step": 726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927698022181035e-05,
+      "loss": 0.5426,
+      "step": 727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992733830024247e-05,
+      "loss": 0.5916,
+      "step": 728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926977688928896e-05,
+      "loss": 0.5712,
+      "step": 729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926616188272626e-05,
+      "loss": 0.6037,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926253798306043e-05,
+      "loss": 0.6136,
+      "step": 731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992589051906161e-05,
+      "loss": 0.5916,
+      "step": 732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925526350571882e-05,
+      "loss": 0.5787,
+      "step": 733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992516129286948e-05,
+      "loss": 0.6128,
+      "step": 734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924795345987103e-05,
+      "loss": 0.6767,
+      "step": 735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992442850995754e-05,
+      "loss": 0.5854,
+      "step": 736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924060784813656e-05,
+      "loss": 0.5442,
+      "step": 737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923692170588395e-05,
+      "loss": 0.5658,
+      "step": 738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992332266731478e-05,
+      "loss": 0.5684,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922952275025914e-05,
+      "loss": 0.5887,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922580993754985e-05,
+      "loss": 0.6864,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922208823535243e-05,
+      "loss": 0.6037,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992183576440004e-05,
+      "loss": 0.5644,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921461816382798e-05,
+      "loss": 0.6002,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921086979517015e-05,
+      "loss": 0.5785,
+      "step": 745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920711253836273e-05,
+      "loss": 0.5674,
+      "step": 746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920334639374234e-05,
+      "loss": 0.6151,
+      "step": 747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919957136164637e-05,
+      "loss": 0.609,
+      "step": 748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99195787442413e-05,
+      "loss": 0.5933,
+      "step": 749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919199463638125e-05,
+      "loss": 0.6382,
+      "step": 750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918819294389088e-05,
+      "loss": 0.61,
+      "step": 751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918438236528255e-05,
+      "loss": 0.5698,
+      "step": 752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991805629008976e-05,
+      "loss": 0.5674,
+      "step": 753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917673455107818e-05,
+      "loss": 0.5509,
+      "step": 754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991728973161673e-05,
+      "loss": 0.5414,
+      "step": 755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916905119650874e-05,
+      "loss": 0.6035,
+      "step": 756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916519619244707e-05,
+      "loss": 0.5037,
+      "step": 757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991613323043276e-05,
+      "loss": 0.5731,
+      "step": 758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915745953249655e-05,
+      "loss": 0.5653,
+      "step": 759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991535778773008e-05,
+      "loss": 0.5848,
+      "step": 760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914968733908825e-05,
+      "loss": 0.5902,
+      "step": 761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991457879182073e-05,
+      "loss": 0.5944,
+      "step": 762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914187961500737e-05,
+      "loss": 0.6032,
+      "step": 763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913796242983857e-05,
+      "loss": 0.5793,
+      "step": 764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913403636305183e-05,
+      "loss": 0.6161,
+      "step": 765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913010141499888e-05,
+      "loss": 0.641,
+      "step": 766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912615758603227e-05,
+      "loss": 0.6368,
+      "step": 767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912220487650532e-05,
+      "loss": 0.5516,
+      "step": 768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991182432867721e-05,
+      "loss": 0.5686,
+      "step": 769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911427281718758e-05,
+      "loss": 0.5705,
+      "step": 770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911029346810744e-05,
+      "loss": 0.6128,
+      "step": 771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991063052398882e-05,
+      "loss": 0.6363,
+      "step": 772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910230813288713e-05,
+      "loss": 0.638,
+      "step": 773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909830214746236e-05,
+      "loss": 0.5632,
+      "step": 774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909428728397278e-05,
+      "loss": 0.6294,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99090263542778e-05,
+      "loss": 0.5876,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908623092423862e-05,
+      "loss": 0.6245,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990821894287158e-05,
+      "loss": 0.6227,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907813905657173e-05,
+      "loss": 0.5505,
+      "step": 779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907407980816917e-05,
+      "loss": 0.5862,
+      "step": 780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907001168387186e-05,
+      "loss": 0.6108,
+      "step": 781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990659346840442e-05,
+      "loss": 0.5734,
+      "step": 782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990618488090515e-05,
+      "loss": 0.6015,
+      "step": 783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990577540592597e-05,
+      "loss": 0.6104,
+      "step": 784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905365043503577e-05,
+      "loss": 0.5748,
+      "step": 785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904953793674728e-05,
+      "loss": 0.5997,
+      "step": 786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904541656476265e-05,
+      "loss": 0.6204,
+      "step": 787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904128631945117e-05,
+      "loss": 0.6149,
+      "step": 788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903714720118283e-05,
+      "loss": 0.6552,
+      "step": 789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990329992103284e-05,
+      "loss": 0.5777,
+      "step": 790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902884234725954e-05,
+      "loss": 0.5571,
+      "step": 791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902467661234865e-05,
+      "loss": 0.6344,
+      "step": 792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902050200596898e-05,
+      "loss": 0.6299,
+      "step": 793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901631852849442e-05,
+      "loss": 0.6277,
+      "step": 794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901212618029985e-05,
+      "loss": 0.5991,
+      "step": 795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900792496176084e-05,
+      "loss": 0.5946,
+      "step": 796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990037148732537e-05,
+      "loss": 0.5638,
+      "step": 797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899949591515572e-05,
+      "loss": 0.608,
+      "step": 798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989952680878448e-05,
+      "loss": 0.5837,
+      "step": 799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899103139169972e-05,
+      "loss": 0.5861,
+      "step": 800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989867858271e-05,
+      "loss": 0.5819,
+      "step": 801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898253139442606e-05,
+      "loss": 0.5242,
+      "step": 802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897826809405903e-05,
+      "loss": 0.5889,
+      "step": 803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989739959263808e-05,
+      "loss": 0.5763,
+      "step": 804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896971489177417e-05,
+      "loss": 0.5735,
+      "step": 805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896542499062265e-05,
+      "loss": 0.5652,
+      "step": 806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896112622331058e-05,
+      "loss": 0.5937,
+      "step": 807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895681859022304e-05,
+      "loss": 0.5727,
+      "step": 808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98952502091746e-05,
+      "loss": 0.6185,
+      "step": 809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894817672826606e-05,
+      "loss": 0.6028,
+      "step": 810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989438425001709e-05,
+      "loss": 0.5764,
+      "step": 811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893949940784865e-05,
+      "loss": 0.5805,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893514745168846e-05,
+      "loss": 0.5807,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893078663208028e-05,
+      "loss": 0.6088,
+      "step": 814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989264169494147e-05,
+      "loss": 0.5507,
+      "step": 815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989220384040832e-05,
+      "loss": 0.5635,
+      "step": 816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989176509964781e-05,
+      "loss": 0.6216,
+      "step": 817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891325472699244e-05,
+      "loss": 0.5965,
+      "step": 818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890884959602006e-05,
+      "loss": 0.5459,
+      "step": 819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989044356039556e-05,
+      "loss": 0.5824,
+      "step": 820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890001275119456e-05,
+      "loss": 0.5804,
+      "step": 821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988955810381331e-05,
+      "loss": 0.5515,
+      "step": 822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889114046516828e-05,
+      "loss": 0.6481,
+      "step": 823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888669103269796e-05,
+      "loss": 0.5955,
+      "step": 824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988822327411207e-05,
+      "loss": 0.5852,
+      "step": 825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98877765590836e-05,
+      "loss": 0.5622,
+      "step": 826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887328958224394e-05,
+      "loss": 0.6008,
+      "step": 827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988688047157456e-05,
+      "loss": 0.5985,
+      "step": 828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886431099174278e-05,
+      "loss": 0.5957,
+      "step": 829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885980841063802e-05,
+      "loss": 0.6304,
+      "step": 830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885529697283472e-05,
+      "loss": 0.5987,
+      "step": 831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885077667873706e-05,
+      "loss": 0.6088,
+      "step": 832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884624752874998e-05,
+      "loss": 0.538,
+      "step": 833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988417095232793e-05,
+      "loss": 0.6096,
+      "step": 834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883716266273153e-05,
+      "loss": 0.5915,
+      "step": 835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98832606947514e-05,
+      "loss": 0.6385,
+      "step": 836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882804237803487e-05,
+      "loss": 0.6216,
+      "step": 837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882346895470306e-05,
+      "loss": 0.5464,
+      "step": 838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988188866779283e-05,
+      "loss": 0.6072,
+      "step": 839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988142955481212e-05,
+      "loss": 0.6085,
+      "step": 840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988096955656929e-05,
+      "loss": 0.6063,
+      "step": 841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880508673105562e-05,
+      "loss": 0.5639,
+      "step": 842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880046904462223e-05,
+      "loss": 0.6534,
+      "step": 843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879584250680642e-05,
+      "loss": 0.6307,
+      "step": 844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879120711802273e-05,
+      "loss": 0.577,
+      "step": 845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878656287868633e-05,
+      "loss": 0.6263,
+      "step": 846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878190978921335e-05,
+      "loss": 0.6114,
+      "step": 847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877724785002065e-05,
+      "loss": 0.5535,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987725770615259e-05,
+      "loss": 0.6082,
+      "step": 849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876789742414755e-05,
+      "loss": 0.5553,
+      "step": 850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876320893830477e-05,
+      "loss": 0.5635,
+      "step": 851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875851160441772e-05,
+      "loss": 0.5902,
+      "step": 852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987538054229071e-05,
+      "loss": 0.5784,
+      "step": 853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987490903941946e-05,
+      "loss": 0.5778,
+      "step": 854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874436651870264e-05,
+      "loss": 0.5681,
+      "step": 855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873963379685438e-05,
+      "loss": 0.6615,
+      "step": 856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873489222907388e-05,
+      "loss": 0.6057,
+      "step": 857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873014181578588e-05,
+      "loss": 0.5735,
+      "step": 858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872538255741592e-05,
+      "loss": 0.5319,
+      "step": 859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872061445439047e-05,
+      "loss": 0.5391,
+      "step": 860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871583750713664e-05,
+      "loss": 0.5817,
+      "step": 861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871105171608242e-05,
+      "loss": 0.5807,
+      "step": 862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870625708165654e-05,
+      "loss": 0.5796,
+      "step": 863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870145360428854e-05,
+      "loss": 0.5957,
+      "step": 864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869664128440876e-05,
+      "loss": 0.6584,
+      "step": 865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869182012244837e-05,
+      "loss": 0.6001,
+      "step": 866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868699011883922e-05,
+      "loss": 0.5906,
+      "step": 867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868215127401407e-05,
+      "loss": 0.6288,
+      "step": 868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986773035884064e-05,
+      "loss": 0.5296,
+      "step": 869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867244706245055e-05,
+      "loss": 0.5792,
+      "step": 870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866758169658154e-05,
+      "loss": 0.5499,
+      "step": 871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986627074912353e-05,
+      "loss": 0.5604,
+      "step": 872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986578244468485e-05,
+      "loss": 0.5693,
+      "step": 873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986529325638586e-05,
+      "loss": 0.5738,
+      "step": 874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864803184270383e-05,
+      "loss": 0.6692,
+      "step": 875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864312228382324e-05,
+      "loss": 0.5524,
+      "step": 876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863820388765672e-05,
+      "loss": 0.5983,
+      "step": 877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986332766546449e-05,
+      "loss": 0.5671,
+      "step": 878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862834058522914e-05,
+      "loss": 0.5299,
+      "step": 879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986233956798517e-05,
+      "loss": 0.6073,
+      "step": 880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861844193895553e-05,
+      "loss": 0.5668,
+      "step": 881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861347936298454e-05,
+      "loss": 0.6789,
+      "step": 882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860850795238324e-05,
+      "loss": 0.6019,
+      "step": 883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860352770759696e-05,
+      "loss": 0.559,
+      "step": 884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98598538629072e-05,
+      "loss": 0.5913,
+      "step": 885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859354071725526e-05,
+      "loss": 0.5763,
+      "step": 886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858853397259447e-05,
+      "loss": 0.5774,
+      "step": 887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985835183955382e-05,
+      "loss": 0.5712,
+      "step": 888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857849398653582e-05,
+      "loss": 0.5322,
+      "step": 889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985734607460374e-05,
+      "loss": 0.5658,
+      "step": 890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985684186744939e-05,
+      "loss": 0.5496,
+      "step": 891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856336777235705e-05,
+      "loss": 0.613,
+      "step": 892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855830804007928e-05,
+      "loss": 0.6251,
+      "step": 893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985532394781139e-05,
+      "loss": 0.5623,
+      "step": 894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985481620869151e-05,
+      "loss": 0.6483,
+      "step": 895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854307586693762e-05,
+      "loss": 0.5962,
+      "step": 896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853798081863723e-05,
+      "loss": 0.6444,
+      "step": 897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985328769424703e-05,
+      "loss": 0.5027,
+      "step": 898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852776423889414e-05,
+      "loss": 0.5309,
+      "step": 899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852264270836675e-05,
+      "loss": 0.5604,
+      "step": 900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98517512351347e-05,
+      "loss": 0.5509,
+      "step": 901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985123731682945e-05,
+      "loss": 0.5853,
+      "step": 902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850722515966962e-05,
+      "loss": 0.5336,
+      "step": 903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850206832593365e-05,
+      "loss": 0.5453,
+      "step": 904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849690266754846e-05,
+      "loss": 0.5709,
+      "step": 905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849172818497694e-05,
+      "loss": 0.5858,
+      "step": 906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848654487868263e-05,
+      "loss": 0.5995,
+      "step": 907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848135274912986e-05,
+      "loss": 0.606,
+      "step": 908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847615179678382e-05,
+      "loss": 0.6057,
+      "step": 909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847094202211043e-05,
+      "loss": 0.6589,
+      "step": 910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846572342557648e-05,
+      "loss": 0.622,
+      "step": 911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846049600764945e-05,
+      "loss": 0.5577,
+      "step": 912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845525976879768e-05,
+      "loss": 0.5326,
+      "step": 913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845001470949024e-05,
+      "loss": 0.5969,
+      "step": 914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844476083019705e-05,
+      "loss": 0.5904,
+      "step": 915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843949813138882e-05,
+      "loss": 0.5836,
+      "step": 916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843422661353697e-05,
+      "loss": 0.5487,
+      "step": 917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842894627711383e-05,
+      "loss": 0.5841,
+      "step": 918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984236571225924e-05,
+      "loss": 0.6331,
+      "step": 919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984183591504466e-05,
+      "loss": 0.6162,
+      "step": 920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98413052361151e-05,
+      "loss": 0.5947,
+      "step": 921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840773675518107e-05,
+      "loss": 0.6327,
+      "step": 922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98402412333013e-05,
+      "loss": 0.5953,
+      "step": 923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983970790951238e-05,
+      "loss": 0.5818,
+      "step": 924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839173704199124e-05,
+      "loss": 0.5414,
+      "step": 925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838638617409397e-05,
+      "loss": 0.6082,
+      "step": 926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983810264919113e-05,
+      "loss": 0.6575,
+      "step": 927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837565799592345e-05,
+      "loss": 0.5999,
+      "step": 928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837028068661138e-05,
+      "loss": 0.5589,
+      "step": 929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836489456445678e-05,
+      "loss": 0.5836,
+      "step": 930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835949962994223e-05,
+      "loss": 0.61,
+      "step": 931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835409588355104e-05,
+      "loss": 0.5774,
+      "step": 932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834868332576727e-05,
+      "loss": 0.5726,
+      "step": 933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834326195707595e-05,
+      "loss": 0.5535,
+      "step": 934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833783177796264e-05,
+      "loss": 0.5167,
+      "step": 935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983323927889139e-05,
+      "loss": 0.56,
+      "step": 936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832694499041696e-05,
+      "loss": 0.5956,
+      "step": 937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832148838295993e-05,
+      "loss": 0.5795,
+      "step": 938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983160229670316e-05,
+      "loss": 0.5781,
+      "step": 939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831054874312167e-05,
+      "loss": 0.5799,
+      "step": 940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983050657117205e-05,
+      "loss": 0.6029,
+      "step": 941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982995738733193e-05,
+      "loss": 0.5442,
+      "step": 942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982940732284102e-05,
+      "loss": 0.5718,
+      "step": 943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828856377748585e-05,
+      "loss": 0.5846,
+      "step": 944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828304552103992e-05,
+      "loss": 0.6208,
+      "step": 945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827751845956673e-05,
+      "loss": 0.5954,
+      "step": 946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982719825935615e-05,
+      "loss": 0.5603,
+      "step": 947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982664379235201e-05,
+      "loss": 0.5667,
+      "step": 948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826088444993933e-05,
+      "loss": 0.5583,
+      "step": 949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825532217331672e-05,
+      "loss": 0.5752,
+      "step": 950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824975109415053e-05,
+      "loss": 0.6043,
+      "step": 951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824417121293992e-05,
+      "loss": 0.5785,
+      "step": 952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982385825301848e-05,
+      "loss": 0.5699,
+      "step": 953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823298504638576e-05,
+      "loss": 0.637,
+      "step": 954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822737876204434e-05,
+      "loss": 0.5871,
+      "step": 955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982217636776628e-05,
+      "loss": 0.6491,
+      "step": 956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821613979374414e-05,
+      "loss": 0.5842,
+      "step": 957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821050711079225e-05,
+      "loss": 0.5892,
+      "step": 958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982048656293117e-05,
+      "loss": 0.5821,
+      "step": 959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819921534980796e-05,
+      "loss": 0.5978,
+      "step": 960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819355627278722e-05,
+      "loss": 0.559,
+      "step": 961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981878883987564e-05,
+      "loss": 0.5509,
+      "step": 962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818221172822335e-05,
+      "loss": 0.5592,
+      "step": 963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981765262616966e-05,
+      "loss": 0.5724,
+      "step": 964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817083199968552e-05,
+      "loss": 0.5441,
+      "step": 965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816512894270025e-05,
+      "loss": 0.5996,
+      "step": 966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815941709125165e-05,
+      "loss": 0.5982,
+      "step": 967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815369644585156e-05,
+      "loss": 0.5946,
+      "step": 968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981479670070124e-05,
+      "loss": 0.5416,
+      "step": 969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981422287752475e-05,
+      "loss": 0.5997,
+      "step": 970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813648175107088e-05,
+      "loss": 0.612,
+      "step": 971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813072593499745e-05,
+      "loss": 0.5907,
+      "step": 972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812496132754287e-05,
+      "loss": 0.6232,
+      "step": 973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981191879292236e-05,
+      "loss": 0.5674,
+      "step": 974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811340574055683e-05,
+      "loss": 0.613,
+      "step": 975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810761476206057e-05,
+      "loss": 0.6012,
+      "step": 976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810181499425368e-05,
+      "loss": 0.5647,
+      "step": 977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980960064376557e-05,
+      "loss": 0.61,
+      "step": 978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98090189092787e-05,
+      "loss": 0.529,
+      "step": 979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808436296016883e-05,
+      "loss": 0.6266,
+      "step": 980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.6033,
+      "step": 981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807268433377247e-05,
+      "loss": 0.5713,
+      "step": 982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980668318410406e-05,
+      "loss": 0.6076,
+      "step": 983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806097056265168e-05,
+      "loss": 0.5828,
+      "step": 984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805510049913096e-05,
+      "loss": 0.4987,
+      "step": 985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980492216510042e-05,
+      "loss": 0.6793,
+      "step": 986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804333401879813e-05,
+      "loss": 0.5427,
+      "step": 987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980374376030402e-05,
+      "loss": 0.5638,
+      "step": 988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980315324042587e-05,
+      "loss": 0.5429,
+      "step": 989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980256184229826e-05,
+      "loss": 0.5758,
+      "step": 990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801969565974178e-05,
+      "loss": 0.5938,
+      "step": 991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801376411506682e-05,
+      "loss": 0.5654,
+      "step": 992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800782378948916e-05,
+      "loss": 0.5708,
+      "step": 993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800187468354097e-05,
+      "loss": 0.5752,
+      "step": 994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799591679775516e-05,
+      "loss": 0.5841,
+      "step": 995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798995013266556e-05,
+      "loss": 0.5904,
+      "step": 996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979839746888067e-05,
+      "loss": 0.5634,
+      "step": 997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979779904667139e-05,
+      "loss": 0.5318,
+      "step": 998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797199746692325e-05,
+      "loss": 0.5908,
+      "step": 999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979659956899717e-05,
+      "loss": 0.6556,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979599851363969e-05,
+      "loss": 0.5351,
+      "step": 1001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979539658067374e-05,
+      "loss": 0.5448,
+      "step": 1002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794793770153238e-05,
+      "loss": 0.5727,
+      "step": 1003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979419008213219e-05,
+      "loss": 0.5551,
+      "step": 1004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793585516664685e-05,
+      "loss": 0.5567,
+      "step": 1005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792980073804878e-05,
+      "loss": 0.5497,
+      "step": 1006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792373753607016e-05,
+      "loss": 0.5917,
+      "step": 1007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791766556125417e-05,
+      "loss": 0.5253,
+      "step": 1008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791158481414472e-05,
+      "loss": 0.5309,
+      "step": 1009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790549529528668e-05,
+      "loss": 0.5837,
+      "step": 1010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978993970052255e-05,
+      "loss": 0.5699,
+      "step": 1011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789328994450758e-05,
+      "loss": 0.5934,
+      "step": 1012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788717411368004e-05,
+      "loss": 0.5621,
+      "step": 1013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978810495132908e-05,
+      "loss": 0.565,
+      "step": 1014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978749161438885e-05,
+      "loss": 0.5691,
+      "step": 1015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786877400602266e-05,
+      "loss": 0.6215,
+      "step": 1016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786262310024352e-05,
+      "loss": 0.53,
+      "step": 1017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785646342710213e-05,
+      "loss": 0.5991,
+      "step": 1018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785029498715036e-05,
+      "loss": 0.6042,
+      "step": 1019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978441177809408e-05,
+      "loss": 0.6412,
+      "step": 1020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978379318090269e-05,
+      "loss": 0.5316,
+      "step": 1021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783173707196278e-05,
+      "loss": 0.5564,
+      "step": 1022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782553357030348e-05,
+      "loss": 0.5573,
+      "step": 1023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781932130460474e-05,
+      "loss": 0.6067,
+      "step": 1024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781310027542307e-05,
+      "loss": 0.5848,
+      "step": 1025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780687048331584e-05,
+      "loss": 0.5567,
+      "step": 1026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978006319288412e-05,
+      "loss": 0.528,
+      "step": 1027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97794384612558e-05,
+      "loss": 0.5334,
+      "step": 1028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778812853502592e-05,
+      "loss": 0.5566,
+      "step": 1029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977818636968055e-05,
+      "loss": 0.5572,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777559009845792e-05,
+      "loss": 0.5569,
+      "step": 1031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977693077405453e-05,
+      "loss": 0.6168,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977630166236304e-05,
+      "loss": 0.6068,
+      "step": 1033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775671674827685e-05,
+      "loss": 0.6147,
+      "step": 1034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97750408115049e-05,
+      "loss": 0.6363,
+      "step": 1035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774409072451213e-05,
+      "loss": 0.5839,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773776457723216e-05,
+      "loss": 0.5582,
+      "step": 1037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773142967377586e-05,
+      "loss": 0.6003,
+      "step": 1038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977250860147107e-05,
+      "loss": 0.589,
+      "step": 1039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771873360060508e-05,
+      "loss": 0.5736,
+      "step": 1040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771237243202806e-05,
+      "loss": 0.6152,
+      "step": 1041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977060025095495e-05,
+      "loss": 0.6044,
+      "step": 1042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769962383374008e-05,
+      "loss": 0.6353,
+      "step": 1043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769323640517133e-05,
+      "loss": 0.5868,
+      "step": 1044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976868402244154e-05,
+      "loss": 0.5743,
+      "step": 1045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976804352920454e-05,
+      "loss": 0.5675,
+      "step": 1046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767402160863505e-05,
+      "loss": 0.5957,
+      "step": 1047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766759917475897e-05,
+      "loss": 0.5747,
+      "step": 1048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976611679909926e-05,
+      "loss": 0.5406,
+      "step": 1049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97654728057912e-05,
+      "loss": 0.5887,
+      "step": 1050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764827937609417e-05,
+      "loss": 0.5497,
+      "step": 1051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976418219461168e-05,
+      "loss": 0.6039,
+      "step": 1052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976353557685585e-05,
+      "loss": 0.5578,
+      "step": 1053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762888084399844e-05,
+      "loss": 0.5737,
+      "step": 1054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762239717301678e-05,
+      "loss": 0.5598,
+      "step": 1055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761590475619435e-05,
+      "loss": 0.554,
+      "step": 1056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976094035941128e-05,
+      "loss": 0.5658,
+      "step": 1057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760289368735456e-05,
+      "loss": 0.5108,
+      "step": 1058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759637503650284e-05,
+      "loss": 0.5917,
+      "step": 1059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758984764214163e-05,
+      "loss": 0.5432,
+      "step": 1060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758331150485576e-05,
+      "loss": 0.577,
+      "step": 1061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757676662523072e-05,
+      "loss": 0.6206,
+      "step": 1062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757021300385288e-05,
+      "loss": 0.5708,
+      "step": 1063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756365064130935e-05,
+      "loss": 0.5375,
+      "step": 1064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975570795381881e-05,
+      "loss": 0.6109,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975504996950778e-05,
+      "loss": 0.5725,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754391111256792e-05,
+      "loss": 0.579,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975373137912487e-05,
+      "loss": 0.578,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753070773171122e-05,
+      "loss": 0.5543,
+      "step": 1069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975240929345473e-05,
+      "loss": 0.5547,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751746940034954e-05,
+      "loss": 0.5365,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751083712971133e-05,
+      "loss": 0.5868,
+      "step": 1072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975041961232268e-05,
+      "loss": 0.5621,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749754638149103e-05,
+      "loss": 0.5776,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749088790509965e-05,
+      "loss": 0.5217,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748422069464924e-05,
+      "loss": 0.6574,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747754475073707e-05,
+      "loss": 0.5939,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974708600739612e-05,
+      "loss": 0.5755,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974641666649206e-05,
+      "loss": 0.5314,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745746452421485e-05,
+      "loss": 0.609,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745075365244442e-05,
+      "loss": 0.5486,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744403405021047e-05,
+      "loss": 0.5723,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743730571811503e-05,
+      "loss": 0.6253,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974305686567609e-05,
+      "loss": 0.5964,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742382286675164e-05,
+      "loss": 0.5749,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741706834869155e-05,
+      "loss": 0.5558,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741030510318582e-05,
+      "loss": 0.6127,
+      "step": 1087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740353313084034e-05,
+      "loss": 0.5383,
+      "step": 1088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739675243226177e-05,
+      "loss": 0.5643,
+      "step": 1089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738996300805758e-05,
+      "loss": 0.5524,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738316485883605e-05,
+      "loss": 0.5549,
+      "step": 1091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737635798520624e-05,
+      "loss": 0.5243,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736954238777793e-05,
+      "loss": 0.5341,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973627180671617e-05,
+      "loss": 0.6009,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97355885023969e-05,
+      "loss": 0.5576,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734904325881192e-05,
+      "loss": 0.5585,
+      "step": 1096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734219277230347e-05,
+      "loss": 0.6111,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973353335650573e-05,
+      "loss": 0.6235,
+      "step": 1098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732846563768797e-05,
+      "loss": 0.5474,
+      "step": 1099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732158899081074e-05,
+      "loss": 0.5038,
+      "step": 1100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973147036250417e-05,
+      "loss": 0.5796,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730780954099768e-05,
+      "loss": 0.5328,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973009067392963e-05,
+      "loss": 0.5265,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729399522055603e-05,
+      "loss": 0.6057,
+      "step": 1104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97287074985396e-05,
+      "loss": 0.5939,
+      "step": 1105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728014603443622e-05,
+      "loss": 0.5356,
+      "step": 1106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727320836829742e-05,
+      "loss": 0.5187,
+      "step": 1107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726626198760116e-05,
+      "loss": 0.5306,
+      "step": 1108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972593068929697e-05,
+      "loss": 0.6342,
+      "step": 1109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725234308502622e-05,
+      "loss": 0.5765,
+      "step": 1110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724537056439454e-05,
+      "loss": 0.5606,
+      "step": 1111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723838933169933e-05,
+      "loss": 0.5477,
+      "step": 1112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97231399387566e-05,
+      "loss": 0.5266,
+      "step": 1113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722440073262084e-05,
+      "loss": 0.5449,
+      "step": 1114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972173933674908e-05,
+      "loss": 0.6058,
+      "step": 1115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721037729280366e-05,
+      "loss": 0.5331,
+      "step": 1116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720335250918797e-05,
+      "loss": 0.5615,
+      "step": 1117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719631901727306e-05,
+      "loss": 0.5554,
+      "step": 1118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971892768176891e-05,
+      "loss": 0.5189,
+      "step": 1119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97182225911067e-05,
+      "loss": 0.5656,
+      "step": 1120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717516629803836e-05,
+      "loss": 0.5615,
+      "step": 1121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971680979792357e-05,
+      "loss": 0.5716,
+      "step": 1122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716102095529224e-05,
+      "loss": 0.4443,
+      "step": 1123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97153935226842e-05,
+      "loss": 0.6097,
+      "step": 1124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971468407945198e-05,
+      "loss": 0.5853,
+      "step": 1125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713973765896116e-05,
+      "loss": 0.4943,
+      "step": 1126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971326258208025e-05,
+      "loss": 0.5211,
+      "step": 1127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712550528068096e-05,
+      "loss": 0.5948,
+      "step": 1128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711837603923444e-05,
+      "loss": 0.4981,
+      "step": 1129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711123809710163e-05,
+      "loss": 0.5153,
+      "step": 1130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97104091454922e-05,
+      "loss": 0.4935,
+      "step": 1131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970969361133358e-05,
+      "loss": 0.5572,
+      "step": 1132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708977207298413e-05,
+      "loss": 0.5754,
+      "step": 1133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708259933450874e-05,
+      "loss": 0.5405,
+      "step": 1134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707541789855224e-05,
+      "loss": 0.573,
+      "step": 1135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706822776575804e-05,
+      "loss": 0.5392,
+      "step": 1136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706102893677022e-05,
+      "loss": 0.6109,
+      "step": 1137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705382141223378e-05,
+      "loss": 0.5531,
+      "step": 1138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970466051927944e-05,
+      "loss": 0.5535,
+      "step": 1139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703938027909854e-05,
+      "loss": 0.5499,
+      "step": 1140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703214667179355e-05,
+      "loss": 0.5236,
+      "step": 1141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970249043715274e-05,
+      "loss": 0.5583,
+      "step": 1142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97017653378949e-05,
+      "loss": 0.5862,
+      "step": 1143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701039369470788e-05,
+      "loss": 0.6118,
+      "step": 1144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700312531945444e-05,
+      "loss": 0.5563,
+      "step": 1145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699584825383984e-05,
+      "loss": 0.6487,
+      "step": 1146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698856249851604e-05,
+      "loss": 0.5793,
+      "step": 1147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698126805413575e-05,
+      "loss": 0.6165,
+      "step": 1148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697396492135244e-05,
+      "loss": 0.6204,
+      "step": 1149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696665310082047e-05,
+      "loss": 0.5645,
+      "step": 1150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695933259319485e-05,
+      "loss": 0.5826,
+      "step": 1151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695200339913133e-05,
+      "loss": 0.5976,
+      "step": 1152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694466551928667e-05,
+      "loss": 0.5711,
+      "step": 1153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693731895431813e-05,
+      "loss": 0.6003,
+      "step": 1154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692996370488394e-05,
+      "loss": 0.5642,
+      "step": 1155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969225997716431e-05,
+      "loss": 0.5352,
+      "step": 1156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969152271552552e-05,
+      "loss": 0.5263,
+      "step": 1157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690784585638086e-05,
+      "loss": 0.6111,
+      "step": 1158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690045587568123e-05,
+      "loss": 0.5322,
+      "step": 1159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689305721381852e-05,
+      "loss": 0.5947,
+      "step": 1160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688564987145547e-05,
+      "loss": 0.5475,
+      "step": 1161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687823384925576e-05,
+      "loss": 0.5918,
+      "step": 1162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968708091478837e-05,
+      "loss": 0.583,
+      "step": 1163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686337576800448e-05,
+      "loss": 0.5707,
+      "step": 1164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685593371028408e-05,
+      "loss": 0.5627,
+      "step": 1165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968484829753892e-05,
+      "loss": 0.6088,
+      "step": 1166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684102356398734e-05,
+      "loss": 0.5095,
+      "step": 1167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683355547674677e-05,
+      "loss": 0.5398,
+      "step": 1168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682607871433655e-05,
+      "loss": 0.615,
+      "step": 1169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968185932774265e-05,
+      "loss": 0.5437,
+      "step": 1170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681109916668727e-05,
+      "loss": 0.5895,
+      "step": 1171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968035963827902e-05,
+      "loss": 0.5577,
+      "step": 1172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679608492640748e-05,
+      "loss": 0.5675,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967885647982121e-05,
+      "loss": 0.5659,
+      "step": 1174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678103599887763e-05,
+      "loss": 0.5955,
+      "step": 1175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967734985290787e-05,
+      "loss": 0.5287,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676595238949052e-05,
+      "loss": 0.5459,
+      "step": 1177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675839758078915e-05,
+      "loss": 0.5536,
+      "step": 1178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967508341036514e-05,
+      "loss": 0.5564,
+      "step": 1179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967432619587549e-05,
+      "loss": 0.5965,
+      "step": 1180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96735681146778e-05,
+      "loss": 0.5304,
+      "step": 1181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967280916683999e-05,
+      "loss": 0.5326,
+      "step": 1182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967204935243004e-05,
+      "loss": 0.5719,
+      "step": 1183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967128867151604e-05,
+      "loss": 0.5243,
+      "step": 1184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670527124166124e-05,
+      "loss": 0.6,
+      "step": 1185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669764710448523e-05,
+      "loss": 0.5039,
+      "step": 1186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966900143043154e-05,
+      "loss": 0.6084,
+      "step": 1187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668237284183552e-05,
+      "loss": 0.5763,
+      "step": 1188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667472271773026e-05,
+      "loss": 0.5513,
+      "step": 1189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666706393268492e-05,
+      "loss": 0.5393,
+      "step": 1190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665939648738564e-05,
+      "loss": 0.5466,
+      "step": 1191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665172038251933e-05,
+      "loss": 0.5768,
+      "step": 1192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664403561877377e-05,
+      "loss": 0.5933,
+      "step": 1193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966363421968373e-05,
+      "loss": 0.5774,
+      "step": 1194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662864011739923e-05,
+      "loss": 0.6024,
+      "step": 1195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662092938114957e-05,
+      "loss": 0.5128,
+      "step": 1196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966132099887791e-05,
+      "loss": 0.5721,
+      "step": 1197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660548194097943e-05,
+      "loss": 0.5748,
+      "step": 1198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965977452384428e-05,
+      "loss": 0.5989,
+      "step": 1199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658999988186248e-05,
+      "loss": 0.5662,
+      "step": 1200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965822458719322e-05,
+      "loss": 0.549,
+      "step": 1201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965744832093468e-05,
+      "loss": 0.5933,
+      "step": 1202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656671189480157e-05,
+      "loss": 0.5484,
+      "step": 1203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655893192899286e-05,
+      "loss": 0.5424,
+      "step": 1204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655114331261757e-05,
+      "loss": 0.5721,
+      "step": 1205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965433460463735e-05,
+      "loss": 0.5506,
+      "step": 1206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965355401309592e-05,
+      "loss": 0.5335,
+      "step": 1207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9652772556707403e-05,
+      "loss": 0.5443,
+      "step": 1208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651990235541804e-05,
+      "loss": 0.5579,
+      "step": 1209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651207049669204e-05,
+      "loss": 0.5839,
+      "step": 1210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965042299915978e-05,
+      "loss": 0.5121,
+      "step": 1211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649638084083764e-05,
+      "loss": 0.5679,
+      "step": 1212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964885230451148e-05,
+      "loss": 0.566,
+      "step": 1213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648065660513328e-05,
+      "loss": 0.5069,
+      "step": 1214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647278152159773e-05,
+      "loss": 0.6044,
+      "step": 1215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646489779521374e-05,
+      "loss": 0.5987,
+      "step": 1216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645700542668758e-05,
+      "loss": 0.5617,
+      "step": 1217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964491044167263e-05,
+      "loss": 0.5716,
+      "step": 1218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644119476603775e-05,
+      "loss": 0.5652,
+      "step": 1219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964332764753306e-05,
+      "loss": 0.5895,
+      "step": 1220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964253495453141e-05,
+      "loss": 0.6027,
+      "step": 1221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641741397669857e-05,
+      "loss": 0.5701,
+      "step": 1222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964094697701948e-05,
+      "loss": 0.5096,
+      "step": 1223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640151692651463e-05,
+      "loss": 0.5759,
+      "step": 1224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639355544637046e-05,
+      "loss": 0.5176,
+      "step": 1225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638558533047558e-05,
+      "loss": 0.5277,
+      "step": 1226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96377606579544e-05,
+      "loss": 0.5232,
+      "step": 1227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636961919429055e-05,
+      "loss": 0.5641,
+      "step": 1228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636162317543077e-05,
+      "loss": 0.5606,
+      "step": 1229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635361852368105e-05,
+      "loss": 0.5209,
+      "step": 1230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963456052397585e-05,
+      "loss": 0.548,
+      "step": 1231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96337583324381e-05,
+      "loss": 0.5539,
+      "step": 1232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632955277826726e-05,
+      "loss": 0.5367,
+      "step": 1233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632151360213667e-05,
+      "loss": 0.5648,
+      "step": 1234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963134657967095e-05,
+      "loss": 0.5448,
+      "step": 1235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963054093627067e-05,
+      "loss": 0.5433,
+      "step": 1236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629734430085007e-05,
+      "loss": 0.5743,
+      "step": 1237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628927061186213e-05,
+      "loss": 0.544,
+      "step": 1238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962811882964662e-05,
+      "loss": 0.5623,
+      "step": 1239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962730973553863e-05,
+      "loss": 0.5785,
+      "step": 1240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626499778934737e-05,
+      "loss": 0.5534,
+      "step": 1241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625688959907502e-05,
+      "loss": 0.5443,
+      "step": 1242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624877278529562e-05,
+      "loss": 0.5576,
+      "step": 1243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624064734873634e-05,
+      "loss": 0.6045,
+      "step": 1244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623251329012518e-05,
+      "loss": 0.576,
+      "step": 1245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622437061019076e-05,
+      "loss": 0.5568,
+      "step": 1246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962162193096627e-05,
+      "loss": 0.5845,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620805938927115e-05,
+      "loss": 0.5111,
+      "step": 1248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961998908497472e-05,
+      "loss": 0.5704,
+      "step": 1249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619171369182265e-05,
+      "loss": 0.5475,
+      "step": 1250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618352791623003e-05,
+      "loss": 0.5734,
+      "step": 1251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961753335237028e-05,
+      "loss": 0.5554,
+      "step": 1252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616713051497496e-05,
+      "loss": 0.5671,
+      "step": 1253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961589188907815e-05,
+      "loss": 0.5204,
+      "step": 1254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96150698651858e-05,
+      "loss": 0.5399,
+      "step": 1255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96142469798941e-05,
+      "loss": 0.5495,
+      "step": 1256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613423233276764e-05,
+      "loss": 0.5392,
+      "step": 1257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612598625407594e-05,
+      "loss": 0.5455,
+      "step": 1258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961177315636046e-05,
+      "loss": 0.517,
+      "step": 1259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961094682620932e-05,
+      "loss": 0.5264,
+      "step": 1260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610119635028203e-05,
+      "loss": 0.5049,
+      "step": 1261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609291582891212e-05,
+      "loss": 0.5447,
+      "step": 1262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608462669872533e-05,
+      "loss": 0.5155,
+      "step": 1263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607632896046427e-05,
+      "loss": 0.6842,
+      "step": 1264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606802261487235e-05,
+      "loss": 0.539,
+      "step": 1265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605970766269365e-05,
+      "loss": 0.6072,
+      "step": 1266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605138410467318e-05,
+      "loss": 0.5749,
+      "step": 1267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960430519415566e-05,
+      "loss": 0.5626,
+      "step": 1268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603471117409033e-05,
+      "loss": 0.5648,
+      "step": 1269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960263618030217e-05,
+      "loss": 0.5675,
+      "step": 1270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601800382909862e-05,
+      "loss": 0.5197,
+      "step": 1271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600963725306994e-05,
+      "loss": 0.5804,
+      "step": 1272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960012620756852e-05,
+      "loss": 0.5883,
+      "step": 1273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599287829769465e-05,
+      "loss": 0.5425,
+      "step": 1274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598448591984947e-05,
+      "loss": 0.5566,
+      "step": 1275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597608494290148e-05,
+      "loss": 0.575,
+      "step": 1276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596767536760328e-05,
+      "loss": 0.568,
+      "step": 1277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595925719470834e-05,
+      "loss": 0.5646,
+      "step": 1278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595083042497078e-05,
+      "loss": 0.5483,
+      "step": 1279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594239505914558e-05,
+      "loss": 0.6556,
+      "step": 1280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959339510979884e-05,
+      "loss": 0.5417,
+      "step": 1281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592549854225575e-05,
+      "loss": 0.559,
+      "step": 1282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591703739270484e-05,
+      "loss": 0.6074,
+      "step": 1283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959085676500938e-05,
+      "loss": 0.5883,
+      "step": 1284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590008931518133e-05,
+      "loss": 0.517,
+      "step": 1285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95891602388727e-05,
+      "loss": 0.5257,
+      "step": 1286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588310687149116e-05,
+      "loss": 0.5588,
+      "step": 1287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587460276423492e-05,
+      "loss": 0.5413,
+      "step": 1288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958660900677201e-05,
+      "loss": 0.5548,
+      "step": 1289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585756878270937e-05,
+      "loss": 0.5943,
+      "step": 1290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584903890996618e-05,
+      "loss": 0.5774,
+      "step": 1291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958405004502547e-05,
+      "loss": 0.5961,
+      "step": 1292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958319534043398e-05,
+      "loss": 0.548,
+      "step": 1293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582339777298724e-05,
+      "loss": 0.5761,
+      "step": 1294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581483355696354e-05,
+      "loss": 0.6254,
+      "step": 1295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958062607570359e-05,
+      "loss": 0.582,
+      "step": 1296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957976793739724e-05,
+      "loss": 0.6073,
+      "step": 1297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578908940854177e-05,
+      "loss": 0.5914,
+      "step": 1298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957804908615136e-05,
+      "loss": 0.6238,
+      "step": 1299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577188373365828e-05,
+      "loss": 0.6156,
+      "step": 1300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957632680257468e-05,
+      "loss": 0.6173,
+      "step": 1301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575464373855112e-05,
+      "loss": 0.5536,
+      "step": 1302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957460108728438e-05,
+      "loss": 0.5994,
+      "step": 1303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573736942939832e-05,
+      "loss": 0.5979,
+      "step": 1304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957287194089888e-05,
+      "loss": 0.5582,
+      "step": 1305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957200608123902e-05,
+      "loss": 0.515,
+      "step": 1306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571139364037823e-05,
+      "loss": 0.4689,
+      "step": 1307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957027178937294e-05,
+      "loss": 0.5769,
+      "step": 1308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.6044,
+      "step": 1309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568534067963078e-05,
+      "loss": 0.5843,
+      "step": 1310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956766392137378e-05,
+      "loss": 0.5595,
+      "step": 1311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566792917632153e-05,
+      "loss": 0.5611,
+      "step": 1312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565921056816232e-05,
+      "loss": 0.5687,
+      "step": 1313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565048339004118e-05,
+      "loss": 0.5267,
+      "step": 1314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564174764274003e-05,
+      "loss": 0.5946,
+      "step": 1315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563300332704146e-05,
+      "loss": 0.5925,
+      "step": 1316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562425044372884e-05,
+      "loss": 0.5907,
+      "step": 1317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561548899358638e-05,
+      "loss": 0.4989,
+      "step": 1318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560671897739898e-05,
+      "loss": 0.63,
+      "step": 1319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955979403959523e-05,
+      "loss": 0.5776,
+      "step": 1320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558915325003287e-05,
+      "loss": 0.4974,
+      "step": 1321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955803575404279e-05,
+      "loss": 0.5239,
+      "step": 1322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955715532679253e-05,
+      "loss": 0.5244,
+      "step": 1323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556274043331393e-05,
+      "loss": 0.5077,
+      "step": 1324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555391903738328e-05,
+      "loss": 0.5315,
+      "step": 1325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554508908092367e-05,
+      "loss": 0.5911,
+      "step": 1326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553625056472612e-05,
+      "loss": 0.5097,
+      "step": 1327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552740348958244e-05,
+      "loss": 0.5874,
+      "step": 1328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551854785628533e-05,
+      "loss": 0.5297,
+      "step": 1329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550968366562807e-05,
+      "loss": 0.5538,
+      "step": 1330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550081091840475e-05,
+      "loss": 0.5451,
+      "step": 1331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549192961541038e-05,
+      "loss": 0.5479,
+      "step": 1332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548303975744058e-05,
+      "loss": 0.5541,
+      "step": 1333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954741413452917e-05,
+      "loss": 0.5527,
+      "step": 1334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546523437976105e-05,
+      "loss": 0.5623,
+      "step": 1335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545631886164653e-05,
+      "loss": 0.5482,
+      "step": 1336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954473947917469e-05,
+      "loss": 0.5557,
+      "step": 1337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954384621708616e-05,
+      "loss": 0.5705,
+      "step": 1338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954295209997909e-05,
+      "loss": 0.5956,
+      "step": 1339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542057127933586e-05,
+      "loss": 0.599,
+      "step": 1340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541161301029826e-05,
+      "loss": 0.6086,
+      "step": 1341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540264619348067e-05,
+      "loss": 0.5535,
+      "step": 1342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953936708296864e-05,
+      "loss": 0.5696,
+      "step": 1343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538468691971957e-05,
+      "loss": 0.5664,
+      "step": 1344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537569446438496e-05,
+      "loss": 0.5338,
+      "step": 1345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536669346448822e-05,
+      "loss": 0.5779,
+      "step": 1346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953576839208358e-05,
+      "loss": 0.5545,
+      "step": 1347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953486658342348e-05,
+      "loss": 0.5481,
+      "step": 1348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533963920549307e-05,
+      "loss": 0.5615,
+      "step": 1349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533060403541937e-05,
+      "loss": 0.4884,
+      "step": 1350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953215603248232e-05,
+      "loss": 0.5442,
+      "step": 1351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531250807451467e-05,
+      "loss": 0.5971,
+      "step": 1352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953034472853048e-05,
+      "loss": 0.5425,
+      "step": 1353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529437795800533e-05,
+      "loss": 0.5296,
+      "step": 1354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528530009342875e-05,
+      "loss": 0.5926,
+      "step": 1355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527621369238836e-05,
+      "loss": 0.548,
+      "step": 1356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526711875569817e-05,
+      "loss": 0.5239,
+      "step": 1357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525801528417304e-05,
+      "loss": 0.585,
+      "step": 1358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524890327862845e-05,
+      "loss": 0.6152,
+      "step": 1359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952397827398808e-05,
+      "loss": 0.5636,
+      "step": 1360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952306536687471e-05,
+      "loss": 0.5963,
+      "step": 1361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952215160660453e-05,
+      "loss": 0.5609,
+      "step": 1362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95212369932594e-05,
+      "loss": 0.6029,
+      "step": 1363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520321526921257e-05,
+      "loss": 0.5344,
+      "step": 1364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519405207672118e-05,
+      "loss": 0.5767,
+      "step": 1365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518488035594073e-05,
+      "loss": 0.4891,
+      "step": 1366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517570010769288e-05,
+      "loss": 0.5311,
+      "step": 1367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516651133280013e-05,
+      "loss": 0.5206,
+      "step": 1368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515731403208566e-05,
+      "loss": 0.5074,
+      "step": 1369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514810820637343e-05,
+      "loss": 0.546,
+      "step": 1370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951388938564882e-05,
+      "loss": 0.5135,
+      "step": 1371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512967098325544e-05,
+      "loss": 0.5011,
+      "step": 1372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512043958750145e-05,
+      "loss": 0.5323,
+      "step": 1373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951111996700532e-05,
+      "loss": 0.5166,
+      "step": 1374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510195123173855e-05,
+      "loss": 0.5842,
+      "step": 1375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95092694273386e-05,
+      "loss": 0.5632,
+      "step": 1376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950834287958249e-05,
+      "loss": 0.5664,
+      "step": 1377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507415479988533e-05,
+      "loss": 0.5862,
+      "step": 1378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950648722863981e-05,
+      "loss": 0.604,
+      "step": 1379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505558125619482e-05,
+      "loss": 0.5857,
+      "step": 1380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950462817101079e-05,
+      "loss": 0.5526,
+      "step": 1381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503697364897046e-05,
+      "loss": 0.6062,
+      "step": 1382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502765707361636e-05,
+      "loss": 0.585,
+      "step": 1383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501833198488026e-05,
+      "loss": 0.5148,
+      "step": 1384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500899838359766e-05,
+      "loss": 0.5217,
+      "step": 1385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499965627060463e-05,
+      "loss": 0.5493,
+      "step": 1386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949903056467382e-05,
+      "loss": 0.5487,
+      "step": 1387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498094651283602e-05,
+      "loss": 0.6158,
+      "step": 1388
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949715788697366e-05,
+      "loss": 0.5107,
+      "step": 1389
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496220271827916e-05,
+      "loss": 0.5366,
+      "step": 1390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949528180593037e-05,
+      "loss": 0.5903,
+      "step": 1391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494342489365095e-05,
+      "loss": 0.5548,
+      "step": 1392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493402322216247e-05,
+      "loss": 0.5692,
+      "step": 1393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949246130456805e-05,
+      "loss": 0.5139,
+      "step": 1394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491519436504812e-05,
+      "loss": 0.5551,
+      "step": 1395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490576718110915e-05,
+      "loss": 0.5316,
+      "step": 1396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948963314947081e-05,
+      "loss": 0.5136,
+      "step": 1397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948868873066903e-05,
+      "loss": 0.5603,
+      "step": 1398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948774346179019e-05,
+      "loss": 0.5007,
+      "step": 1399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486797342918974e-05,
+      "loss": 0.5505,
+      "step": 1400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485850374140137e-05,
+      "loss": 0.5183,
+      "step": 1401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484902555538525e-05,
+      "loss": 0.5264,
+      "step": 1402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483953887199043e-05,
+      "loss": 0.5163,
+      "step": 1403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483004369206687e-05,
+      "loss": 0.5242,
+      "step": 1404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948205400164652e-05,
+      "loss": 0.5379,
+      "step": 1405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481102784603686e-05,
+      "loss": 0.5169,
+      "step": 1406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480150718163403e-05,
+      "loss": 0.5231,
+      "step": 1407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947919780241096e-05,
+      "loss": 0.5841,
+      "step": 1408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478244037431734e-05,
+      "loss": 0.5156,
+      "step": 1409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947728942331117e-05,
+      "loss": 0.5613,
+      "step": 1410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947633396013479e-05,
+      "loss": 0.5231,
+      "step": 1411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475377647988187e-05,
+      "loss": 0.5527,
+      "step": 1412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474420486957045e-05,
+      "loss": 0.5908,
+      "step": 1413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473462477127105e-05,
+      "loss": 0.5366,
+      "step": 1414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472503618584203e-05,
+      "loss": 0.5242,
+      "step": 1415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471543911414233e-05,
+      "loss": 0.5387,
+      "step": 1416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947058335570318e-05,
+      "loss": 0.5372,
+      "step": 1417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94696219515371e-05,
+      "loss": 0.5179,
+      "step": 1418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468659699002112e-05,
+      "loss": 0.603,
+      "step": 1419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467696598184438e-05,
+      "loss": 0.5334,
+      "step": 1420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946673264917035e-05,
+      "loss": 0.5373,
+      "step": 1421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465767852046212e-05,
+      "loss": 0.5681,
+      "step": 1422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464802206898456e-05,
+      "loss": 0.5905,
+      "step": 1423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946383571381359e-05,
+      "loss": 0.554,
+      "step": 1424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462868372878214e-05,
+      "loss": 0.5859,
+      "step": 1425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461900184178974e-05,
+      "loss": 0.5604,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460931147802614e-05,
+      "loss": 0.5496,
+      "step": 1427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459961263835954e-05,
+      "loss": 0.6143,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945899053236588e-05,
+      "loss": 0.5546,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458018953479356e-05,
+      "loss": 0.5739,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945704652726343e-05,
+      "loss": 0.5934,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456073253805214e-05,
+      "loss": 0.5549,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945509913319191e-05,
+      "loss": 0.5071,
+      "step": 1433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945412416551078e-05,
+      "loss": 0.5174,
+      "step": 1434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945314835084917e-05,
+      "loss": 0.5759,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945217168929451e-05,
+      "loss": 0.5545,
+      "step": 1436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945119418093429e-05,
+      "loss": 0.5683,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450215825856083e-05,
+      "loss": 0.5315,
+      "step": 1438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944923662414755e-05,
+      "loss": 0.5833,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94482565758964e-05,
+      "loss": 0.5692,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447275681190443e-05,
+      "loss": 0.5706,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446293940117553e-05,
+      "loss": 0.5327,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445311352765687e-05,
+      "loss": 0.5899,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944432791922287e-05,
+      "loss": 0.5457,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443343639577206e-05,
+      "loss": 0.5712,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442358513916875e-05,
+      "loss": 0.5647,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441372542330134e-05,
+      "loss": 0.5572,
+      "step": 1447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440385724905314e-05,
+      "loss": 0.5984,
+      "step": 1448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943939806173082e-05,
+      "loss": 0.5229,
+      "step": 1449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438409552895147e-05,
+      "loss": 0.5679,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437420198486836e-05,
+      "loss": 0.5166,
+      "step": 1451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436429998594536e-05,
+      "loss": 0.5332,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943543895330695e-05,
+      "loss": 0.5457,
+      "step": 1453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434447062712866e-05,
+      "loss": 0.5754,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433454326901148e-05,
+      "loss": 0.5697,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432460745960732e-05,
+      "loss": 0.5785,
+      "step": 1456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431466319980633e-05,
+      "loss": 0.5451,
+      "step": 1457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430471049049937e-05,
+      "loss": 0.5875,
+      "step": 1458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942947493325781e-05,
+      "loss": 0.5342,
+      "step": 1459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428477972693493e-05,
+      "loss": 0.5785,
+      "step": 1460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427480167446303e-05,
+      "loss": 0.5957,
+      "step": 1461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942648151760563e-05,
+      "loss": 0.5751,
+      "step": 1462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942548202326094e-05,
+      "loss": 0.541,
+      "step": 1463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424481684501787e-05,
+      "loss": 0.5179,
+      "step": 1464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423480501417775e-05,
+      "loss": 0.5588,
+      "step": 1465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422478474098604e-05,
+      "loss": 0.5767,
+      "step": 1466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421475602634047e-05,
+      "loss": 0.5415,
+      "step": 1467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942047188711395e-05,
+      "loss": 0.5537,
+      "step": 1468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419467327628226e-05,
+      "loss": 0.4885,
+      "step": 1469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418461924266886e-05,
+      "loss": 0.5684,
+      "step": 1470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417455677119987e-05,
+      "loss": 0.5614,
+      "step": 1471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416448586277687e-05,
+      "loss": 0.6007,
+      "step": 1472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941544065183021e-05,
+      "loss": 0.5064,
+      "step": 1473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941443187386785e-05,
+      "loss": 0.526,
+      "step": 1474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413422252480985e-05,
+      "loss": 0.5119,
+      "step": 1475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412411787760066e-05,
+      "loss": 0.5075,
+      "step": 1476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411400479795618e-05,
+      "loss": 0.5853,
+      "step": 1477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941038832867824e-05,
+      "loss": 0.537,
+      "step": 1478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409375334498616e-05,
+      "loss": 0.558,
+      "step": 1479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940836149734749e-05,
+      "loss": 0.5027,
+      "step": 1480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407346817315696e-05,
+      "loss": 0.5579,
+      "step": 1481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406331294494137e-05,
+      "loss": 0.5573,
+      "step": 1482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940531492897379e-05,
+      "loss": 0.5664,
+      "step": 1483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404297720845707e-05,
+      "loss": 0.5707,
+      "step": 1484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403279670201027e-05,
+      "loss": 0.527,
+      "step": 1485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402260777130946e-05,
+      "loss": 0.4883,
+      "step": 1486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401241041726756e-05,
+      "loss": 0.5321,
+      "step": 1487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400220464079803e-05,
+      "loss": 0.5371,
+      "step": 1488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399199044281523e-05,
+      "loss": 0.5289,
+      "step": 1489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939817678242342e-05,
+      "loss": 0.5134,
+      "step": 1490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397153678597084e-05,
+      "loss": 0.5597,
+      "step": 1491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939612973289417e-05,
+      "loss": 0.6025,
+      "step": 1492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939510494540641e-05,
+      "loss": 0.5952,
+      "step": 1493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394079316225616e-05,
+      "loss": 0.5713,
+      "step": 1494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939305284544367e-05,
+      "loss": 0.627,
+      "step": 1495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939202553315253e-05,
+      "loss": 0.5432,
+      "step": 1496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390997379444238e-05,
+      "loss": 0.5393,
+      "step": 1497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93899683844109e-05,
+      "loss": 0.5286,
+      "step": 1498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93889385481447e-05,
+      "loss": 0.454,
+      "step": 1499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387907870737907e-05,
+      "loss": 0.5559,
+      "step": 1500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938687635228285e-05,
+      "loss": 0.5252,
+      "step": 1501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385843992871947e-05,
+      "loss": 0.5971,
+      "step": 1502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384810792597682e-05,
+      "loss": 0.5597,
+      "step": 1503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383776751552622e-05,
+      "loss": 0.5522,
+      "step": 1504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382741869829394e-05,
+      "loss": 0.5843,
+      "step": 1505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381706147520726e-05,
+      "loss": 0.5538,
+      "step": 1506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380669584719397e-05,
+      "loss": 0.6076,
+      "step": 1507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379632181518277e-05,
+      "loss": 0.5305,
+      "step": 1508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378593938010302e-05,
+      "loss": 0.5951,
+      "step": 1509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377554854288486e-05,
+      "loss": 0.5833,
+      "step": 1510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376514930445918e-05,
+      "loss": 0.5192,
+      "step": 1511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375474166575768e-05,
+      "loss": 0.5593,
+      "step": 1512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937443256277127e-05,
+      "loss": 0.5796,
+      "step": 1513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937339011912575e-05,
+      "loss": 0.524,
+      "step": 1514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372346835732593e-05,
+      "loss": 0.5487,
+      "step": 1515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937130271268526e-05,
+      "loss": 0.5979,
+      "step": 1516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370257750077296e-05,
+      "loss": 0.5085,
+      "step": 1517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936921194800232e-05,
+      "loss": 0.5665,
+      "step": 1518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368165306554024e-05,
+      "loss": 0.5422,
+      "step": 1519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936711782582617e-05,
+      "loss": 0.5413,
+      "step": 1520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366069505912606e-05,
+      "loss": 0.5149,
+      "step": 1521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365020346907246e-05,
+      "loss": 0.5859,
+      "step": 1522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363970348904083e-05,
+      "loss": 0.5357,
+      "step": 1523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362919511997186e-05,
+      "loss": 0.5987,
+      "step": 1524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361867836280696e-05,
+      "loss": 0.5176,
+      "step": 1525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360815321848834e-05,
+      "loss": 0.5031,
+      "step": 1526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935976196879589e-05,
+      "loss": 0.5781,
+      "step": 1527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358707777216234e-05,
+      "loss": 0.5203,
+      "step": 1528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935765274720431e-05,
+      "loss": 0.5609,
+      "step": 1529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356596878854633e-05,
+      "loss": 0.5457,
+      "step": 1530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355540172261796e-05,
+      "loss": 0.5645,
+      "step": 1531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354482627520476e-05,
+      "loss": 0.5826,
+      "step": 1532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353424244725412e-05,
+      "loss": 0.5235,
+      "step": 1533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935236502397142e-05,
+      "loss": 0.5761,
+      "step": 1534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351304965353397e-05,
+      "loss": 0.5492,
+      "step": 1535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935024406896631e-05,
+      "loss": 0.5825,
+      "step": 1536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349182334905204e-05,
+      "loss": 0.5519,
+      "step": 1537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93481197632652e-05,
+      "loss": 0.4921,
+      "step": 1538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934705635414149e-05,
+      "loss": 0.5582,
+      "step": 1539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345992107629348e-05,
+      "loss": 0.5851,
+      "step": 1540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344927023824112e-05,
+      "loss": 0.4936,
+      "step": 1541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343861102821204e-05,
+      "loss": 0.5618,
+      "step": 1542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342794344716115e-05,
+      "loss": 0.5325,
+      "step": 1543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934172674960442e-05,
+      "loss": 0.5564,
+      "step": 1544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340658317581754e-05,
+      "loss": 0.5711,
+      "step": 1545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339589048743846e-05,
+      "loss": 0.5501,
+      "step": 1546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338518943186488e-05,
+      "loss": 0.5054,
+      "step": 1547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9337448001005547e-05,
+      "loss": 0.559,
+      "step": 1548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336376222296968e-05,
+      "loss": 0.5624,
+      "step": 1549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335303607156767e-05,
+      "loss": 0.5401,
+      "step": 1550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933423015568104e-05,
+      "loss": 0.5488,
+      "step": 1551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333155867965964e-05,
+      "loss": 0.4927,
+      "step": 1552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332080744107768e-05,
+      "loss": 0.6271,
+      "step": 1553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331004784202777e-05,
+      "loss": 0.5247,
+      "step": 1554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932992798834739e-05,
+      "loss": 0.612,
+      "step": 1555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932885035663807e-05,
+      "loss": 0.5706,
+      "step": 1556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932777188917136e-05,
+      "loss": 0.5328,
+      "step": 1557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932669258604388e-05,
+      "loss": 0.5274,
+      "step": 1558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325612447352326e-05,
+      "loss": 0.576,
+      "step": 1559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324531473193454e-05,
+      "loss": 0.5497,
+      "step": 1560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323449663664124e-05,
+      "loss": 0.5542,
+      "step": 1561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932236701886124e-05,
+      "loss": 0.5662,
+      "step": 1562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321283538881803e-05,
+      "loss": 0.5197,
+      "step": 1563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320199223822875e-05,
+      "loss": 0.5315,
+      "step": 1564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319114073781602e-05,
+      "loss": 0.5701,
+      "step": 1565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93180280888552e-05,
+      "loss": 0.5512,
+      "step": 1566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316941269140955e-05,
+      "loss": 0.5285,
+      "step": 1567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315853614736242e-05,
+      "loss": 0.5311,
+      "step": 1568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9314765125738498e-05,
+      "loss": 0.5516,
+      "step": 1569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313675802245236e-05,
+      "loss": 0.5713,
+      "step": 1570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312585644354054e-05,
+      "loss": 0.6115,
+      "step": 1571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311494652162613e-05,
+      "loss": 0.5344,
+      "step": 1572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310402825768655e-05,
+      "loss": 0.5072,
+      "step": 1573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309310165269995e-05,
+      "loss": 0.5266,
+      "step": 1574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930821667076452e-05,
+      "loss": 0.5887,
+      "step": 1575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93071223423502e-05,
+      "loss": 0.5933,
+      "step": 1576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306027180125065e-05,
+      "loss": 0.5441,
+      "step": 1577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930493118418724e-05,
+      "loss": 0.5796,
+      "step": 1578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303834354634903e-05,
+      "loss": 0.5954,
+      "step": 1579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302736691566327e-05,
+      "loss": 0.5793,
+      "step": 1580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301638195079845e-05,
+      "loss": 0.5709,
+      "step": 1581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930053886527387e-05,
+      "loss": 0.5417,
+      "step": 1582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299438702246893e-05,
+      "loss": 0.4817,
+      "step": 1583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298337706097465e-05,
+      "loss": 0.5691,
+      "step": 1584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297235876924233e-05,
+      "loss": 0.5549,
+      "step": 1585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296133214825906e-05,
+      "loss": 0.5936,
+      "step": 1586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295029719901267e-05,
+      "loss": 0.5501,
+      "step": 1587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929392539224918e-05,
+      "loss": 0.5028,
+      "step": 1588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929282023196858e-05,
+      "loss": 0.5647,
+      "step": 1589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929171423915847e-05,
+      "loss": 0.535,
+      "step": 1590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290607413917944e-05,
+      "loss": 0.4827,
+      "step": 1591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289499756346154e-05,
+      "loss": 0.5569,
+      "step": 1592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288391266542334e-05,
+      "loss": 0.5588,
+      "step": 1593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287281944605795e-05,
+      "loss": 0.5125,
+      "step": 1594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286171790635917e-05,
+      "loss": 0.5431,
+      "step": 1595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928506080473216e-05,
+      "loss": 0.5467,
+      "step": 1596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9283948986994047e-05,
+      "loss": 0.5213,
+      "step": 1597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282836337521192e-05,
+      "loss": 0.5248,
+      "step": 1598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281722856413277e-05,
+      "loss": 0.5479,
+      "step": 1599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280608543770054e-05,
+      "loss": 0.5028,
+      "step": 1600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279493399691353e-05,
+      "loss": 0.5819,
+      "step": 1601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278377424277074e-05,
+      "loss": 0.5791,
+      "step": 1602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92772606176272e-05,
+      "loss": 0.5577,
+      "step": 1603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276142979841782e-05,
+      "loss": 0.5346,
+      "step": 1604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927502451102095e-05,
+      "loss": 0.4961,
+      "step": 1605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92739052112649e-05,
+      "loss": 0.5517,
+      "step": 1606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272785080673917e-05,
+      "loss": 0.546,
+      "step": 1607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271664119348347e-05,
+      "loss": 0.5389,
+      "step": 1608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270542327388618e-05,
+      "loss": 0.5413,
+      "step": 1609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269419704895222e-05,
+      "loss": 0.546,
+      "step": 1610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926829625196874e-05,
+      "loss": 0.5127,
+      "step": 1611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267171968709818e-05,
+      "loss": 0.5603,
+      "step": 1612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266046855219183e-05,
+      "loss": 0.4784,
+      "step": 1613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264920911597628e-05,
+      "loss": 0.5631,
+      "step": 1614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926379413794602e-05,
+      "loss": 0.6019,
+      "step": 1615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262666534365315e-05,
+      "loss": 0.5519,
+      "step": 1616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9261538100956528e-05,
+      "loss": 0.5522,
+      "step": 1617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260408837820754e-05,
+      "loss": 0.572,
+      "step": 1618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925927874505916e-05,
+      "loss": 0.5133,
+      "step": 1619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258147822772995e-05,
+      "loss": 0.5858,
+      "step": 1620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925701607106357e-05,
+      "loss": 0.5421,
+      "step": 1621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925588349003228e-05,
+      "loss": 0.5078,
+      "step": 1622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254750079780587e-05,
+      "loss": 0.507,
+      "step": 1623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253615840410043e-05,
+      "loss": 0.5338,
+      "step": 1624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925248077202225e-05,
+      "loss": 0.5835,
+      "step": 1625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251344874718904e-05,
+      "loss": 0.5715,
+      "step": 1626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925020814860177e-05,
+      "loss": 0.6126,
+      "step": 1627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249070593772677e-05,
+      "loss": 0.6003,
+      "step": 1628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247932210333544e-05,
+      "loss": 0.5109,
+      "step": 1629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246792998386356e-05,
+      "loss": 0.5459,
+      "step": 1630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924565295803317e-05,
+      "loss": 0.5506,
+      "step": 1631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244512089376123e-05,
+      "loss": 0.5097,
+      "step": 1632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243370392517423e-05,
+      "loss": 0.5755,
+      "step": 1633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242227867559355e-05,
+      "loss": 0.5792,
+      "step": 1634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9241084514604273e-05,
+      "loss": 0.5718,
+      "step": 1635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239940333754608e-05,
+      "loss": 0.5694,
+      "step": 1636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.5485,
+      "step": 1637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237649488781634e-05,
+      "loss": 0.5372,
+      "step": 1638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236502824863554e-05,
+      "loss": 0.5451,
+      "step": 1639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923535533346136e-05,
+      "loss": 0.4529,
+      "step": 1640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234207014677853e-05,
+      "loss": 0.5103,
+      "step": 1641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923305786861591e-05,
+      "loss": 0.5085,
+      "step": 1642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231907895378478e-05,
+      "loss": 0.5945,
+      "step": 1643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230757095068587e-05,
+      "loss": 0.5547,
+      "step": 1644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922960546778933e-05,
+      "loss": 0.5526,
+      "step": 1645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228453013643883e-05,
+      "loss": 0.5431,
+      "step": 1646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922729973273549e-05,
+      "loss": 0.5326,
+      "step": 1647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226145625167475e-05,
+      "loss": 0.5865,
+      "step": 1648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224990691043227e-05,
+      "loss": 0.5007,
+      "step": 1649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223834930466218e-05,
+      "loss": 0.5048,
+      "step": 1650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222678343539997e-05,
+      "loss": 0.5973,
+      "step": 1651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221520930368172e-05,
+      "loss": 0.5283,
+      "step": 1652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220362691054435e-05,
+      "loss": 0.5593,
+      "step": 1653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219203625702552e-05,
+      "loss": 0.5401,
+      "step": 1654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921804373441637e-05,
+      "loss": 0.5166,
+      "step": 1655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216883017299786e-05,
+      "loss": 0.5001,
+      "step": 1656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215721474456802e-05,
+      "loss": 0.5602,
+      "step": 1657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214559105991467e-05,
+      "loss": 0.5767,
+      "step": 1658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213395912007928e-05,
+      "loss": 0.5057,
+      "step": 1659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921223189261038e-05,
+      "loss": 0.51,
+      "step": 1660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211067047903114e-05,
+      "loss": 0.5472,
+      "step": 1661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920990137799049e-05,
+      "loss": 0.5393,
+      "step": 1662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920873488297693e-05,
+      "loss": 0.5108,
+      "step": 1663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920756756296694e-05,
+      "loss": 0.5433,
+      "step": 1664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206399418065102e-05,
+      "loss": 0.5598,
+      "step": 1665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920523044837607e-05,
+      "loss": 0.5831,
+      "step": 1666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204060654004566e-05,
+      "loss": 0.5742,
+      "step": 1667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920289003505539e-05,
+      "loss": 0.5256,
+      "step": 1668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201718591633417e-05,
+      "loss": 0.5071,
+      "step": 1669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92005463238436e-05,
+      "loss": 0.5222,
+      "step": 1670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199373231790953e-05,
+      "loss": 0.5436,
+      "step": 1671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919819931558057e-05,
+      "loss": 0.5835,
+      "step": 1672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197024575317627e-05,
+      "loss": 0.581,
+      "step": 1673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195849011107367e-05,
+      "loss": 0.5582,
+      "step": 1674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194672623055103e-05,
+      "loss": 0.5159,
+      "step": 1675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919349541126623e-05,
+      "loss": 0.5165,
+      "step": 1676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919231737584621e-05,
+      "loss": 0.5671,
+      "step": 1677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919113851690058e-05,
+      "loss": 0.5773,
+      "step": 1678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189958834534952e-05,
+      "loss": 0.5457,
+      "step": 1679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188778328855017e-05,
+      "loss": 0.5573,
+      "step": 1680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918759699996653e-05,
+      "loss": 0.5611,
+      "step": 1681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186414847975327e-05,
+      "loss": 0.5299,
+      "step": 1682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918523187298731e-05,
+      "loss": 0.5448,
+      "step": 1683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184048075108467e-05,
+      "loss": 0.53,
+      "step": 1684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918286345444485e-05,
+      "loss": 0.5344,
+      "step": 1685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181678011102585e-05,
+      "loss": 0.5222,
+      "step": 1686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918049174518788e-05,
+      "loss": 0.5547,
+      "step": 1687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179304656807e-05,
+      "loss": 0.5672,
+      "step": 1688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178116746066307e-05,
+      "loss": 0.5483,
+      "step": 1689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176928013072218e-05,
+      "loss": 0.5613,
+      "step": 1690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175738457931226e-05,
+      "loss": 0.516,
+      "step": 1691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917454808074991e-05,
+      "loss": 0.5425,
+      "step": 1692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173356881634905e-05,
+      "loss": 0.5609,
+      "step": 1693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172164860692936e-05,
+      "loss": 0.4937,
+      "step": 1694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917097201803079e-05,
+      "loss": 0.5087,
+      "step": 1695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169778353755335e-05,
+      "loss": 0.5453,
+      "step": 1696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168583867973507e-05,
+      "loss": 0.5786,
+      "step": 1697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916738856079232e-05,
+      "loss": 0.5661,
+      "step": 1698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166192432318858e-05,
+      "loss": 0.5547,
+      "step": 1699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916499548266028e-05,
+      "loss": 0.5673,
+      "step": 1700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163797711923822e-05,
+      "loss": 0.4966,
+      "step": 1701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916259912021679e-05,
+      "loss": 0.5245,
+      "step": 1702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916139970764656e-05,
+      "loss": 0.5548,
+      "step": 1703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160199474320592e-05,
+      "loss": 0.5545,
+      "step": 1704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158998420346404e-05,
+      "loss": 0.5471,
+      "step": 1705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157796545831603e-05,
+      "loss": 0.5275,
+      "step": 1706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156593850883864e-05,
+      "loss": 0.595,
+      "step": 1707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915539033561093e-05,
+      "loss": 0.4939,
+      "step": 1708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154186000120626e-05,
+      "loss": 0.5857,
+      "step": 1709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915298084452084e-05,
+      "loss": 0.5288,
+      "step": 1710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151774868919554e-05,
+      "loss": 0.5133,
+      "step": 1711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150568073424792e-05,
+      "loss": 0.4948,
+      "step": 1712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149360458144682e-05,
+      "loss": 0.5251,
+      "step": 1713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914815202318741e-05,
+      "loss": 0.559,
+      "step": 1714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146942768661233e-05,
+      "loss": 0.5384,
+      "step": 1715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145732694674488e-05,
+      "loss": 0.6052,
+      "step": 1716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144521801335588e-05,
+      "loss": 0.4916,
+      "step": 1717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143310088753008e-05,
+      "loss": 0.5116,
+      "step": 1718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914209755703531e-05,
+      "loss": 0.5144,
+      "step": 1719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140884206291118e-05,
+      "loss": 0.5423,
+      "step": 1720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139670036629137e-05,
+      "loss": 0.5667,
+      "step": 1721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138455048158143e-05,
+      "loss": 0.4971,
+      "step": 1722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137239240986983e-05,
+      "loss": 0.5506,
+      "step": 1723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136022615224578e-05,
+      "loss": 0.5517,
+      "step": 1724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134805170979928e-05,
+      "loss": 0.5164,
+      "step": 1725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133586908362104e-05,
+      "loss": 0.5075,
+      "step": 1726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132367827480238e-05,
+      "loss": 0.5542,
+      "step": 1727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131147928443553e-05,
+      "loss": 0.4912,
+      "step": 1728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129927211361336e-05,
+      "loss": 0.5578,
+      "step": 1729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128705676342953e-05,
+      "loss": 0.5534,
+      "step": 1730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127483323497833e-05,
+      "loss": 0.5356,
+      "step": 1731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912626015293549e-05,
+      "loss": 0.5764,
+      "step": 1732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125036164765502e-05,
+      "loss": 0.5165,
+      "step": 1733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123811359097526e-05,
+      "loss": 0.5972,
+      "step": 1734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912258573604129e-05,
+      "loss": 0.5284,
+      "step": 1735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91213592957066e-05,
+      "loss": 0.5337,
+      "step": 1736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912013203820332e-05,
+      "loss": 0.5044,
+      "step": 1737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118903963641414e-05,
+      "loss": 0.5132,
+      "step": 1738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911767507213089e-05,
+      "loss": 0.5518,
+      "step": 1739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911644536378185e-05,
+      "loss": 0.5121,
+      "step": 1740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115214838704453e-05,
+      "loss": 0.5149,
+      "step": 1741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911398349700895e-05,
+      "loss": 0.5549,
+      "step": 1742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9112751338805647e-05,
+      "loss": 0.4904,
+      "step": 1743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911151836420494e-05,
+      "loss": 0.5711,
+      "step": 1744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911028457331728e-05,
+      "loss": 0.5178,
+      "step": 1745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109049966253207e-05,
+      "loss": 0.5819,
+      "step": 1746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107814543123325e-05,
+      "loss": 0.5934,
+      "step": 1747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910657830403831e-05,
+      "loss": 0.5977,
+      "step": 1748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910534124910892e-05,
+      "loss": 0.525,
+      "step": 1749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9104103378445983e-05,
+      "loss": 0.545,
+      "step": 1750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910286469216039e-05,
+      "loss": 0.5829,
+      "step": 1751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910162519036312e-05,
+      "loss": 0.5251,
+      "step": 1752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100384873165212e-05,
+      "loss": 0.5763,
+      "step": 1753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099143740677792e-05,
+      "loss": 0.5232,
+      "step": 1754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097901793012044e-05,
+      "loss": 0.5552,
+      "step": 1755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096659030279233e-05,
+      "loss": 0.5592,
+      "step": 1756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90954154525907e-05,
+      "loss": 0.5104,
+      "step": 1757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094171060057856e-05,
+      "loss": 0.5228,
+      "step": 1758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092925852792178e-05,
+      "loss": 0.5115,
+      "step": 1759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091679830905225e-05,
+      "loss": 0.554,
+      "step": 1760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909043299450863e-05,
+      "loss": 0.4944,
+      "step": 1761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908918534371409e-05,
+      "loss": 0.5049,
+      "step": 1762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908793687863338e-05,
+      "loss": 0.5458,
+      "step": 1763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086687599378355e-05,
+      "loss": 0.5583,
+      "step": 1764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085437506060925e-05,
+      "loss": 0.5315,
+      "step": 1765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084186598793094e-05,
+      "loss": 0.5104,
+      "step": 1766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082934877686927e-05,
+      "loss": 0.5388,
+      "step": 1767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908168234285456e-05,
+      "loss": 0.5375,
+      "step": 1768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080428994408207e-05,
+      "loss": 0.54,
+      "step": 1769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079174832460153e-05,
+      "loss": 0.5542,
+      "step": 1770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077919857122757e-05,
+      "loss": 0.5113,
+      "step": 1771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076664068508453e-05,
+      "loss": 0.5478,
+      "step": 1772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075407466729743e-05,
+      "loss": 0.6054,
+      "step": 1773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90741500518992e-05,
+      "loss": 0.5366,
+      "step": 1774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907289182412948e-05,
+      "loss": 0.5183,
+      "step": 1775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071632783533306e-05,
+      "loss": 0.5521,
+      "step": 1776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907037293022347e-05,
+      "loss": 0.5576,
+      "step": 1777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069112264312843e-05,
+      "loss": 0.4993,
+      "step": 1778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906785078591436e-05,
+      "loss": 0.5687,
+      "step": 1779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066588495141043e-05,
+      "loss": 0.5992,
+      "step": 1780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906532539210597e-05,
+      "loss": 0.5562,
+      "step": 1781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906406147692231e-05,
+      "loss": 0.5808,
+      "step": 1782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906279674970329e-05,
+      "loss": 0.532,
+      "step": 1783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061531210562218e-05,
+      "loss": 0.5238,
+      "step": 1784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9060264859612467e-05,
+      "loss": 0.5709,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058997696967488e-05,
+      "loss": 0.4925,
+      "step": 1786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905772972274081e-05,
+      "loss": 0.5042,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056460937046024e-05,
+      "loss": 0.5162,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055191339996798e-05,
+      "loss": 0.5091,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053920931706876e-05,
+      "loss": 0.5694,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905264971229007e-05,
+      "loss": 0.5124,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051377681860268e-05,
+      "loss": 0.5314,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050104840531427e-05,
+      "loss": 0.5277,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904883118841758e-05,
+      "loss": 0.5677,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047556725632834e-05,
+      "loss": 0.5868,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904628145229136e-05,
+      "loss": 0.4788,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045005368507418e-05,
+      "loss": 0.4833,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043728474395322e-05,
+      "loss": 0.5524,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042450770069466e-05,
+      "loss": 0.5556,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904117225564433e-05,
+      "loss": 0.5274,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039892931234434e-05,
+      "loss": 0.533,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903861279695441e-05,
+      "loss": 0.5116,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903733185291893e-05,
+      "loss": 0.5753,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036050099242756e-05,
+      "loss": 0.5637,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903476753604072e-05,
+      "loss": 0.5031,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033484163427726e-05,
+      "loss": 0.5607,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032199981518748e-05,
+      "loss": 0.5912,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903091499042883e-05,
+      "loss": 0.5303,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90296291902731e-05,
+      "loss": 0.5684,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028342581166744e-05,
+      "loss": 0.5194,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027055163225027e-05,
+      "loss": 0.5079,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025766936563293e-05,
+      "loss": 0.5724,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902447790129695e-05,
+      "loss": 0.5463,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023188057541477e-05,
+      "loss": 0.5444,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021897405412432e-05,
+      "loss": 0.5011,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020605945025443e-05,
+      "loss": 0.5799,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901931367649621e-05,
+      "loss": 0.5338,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018020599940502e-05,
+      "loss": 0.5385,
+      "step": 1818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901672671547417e-05,
+      "loss": 0.5254,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015432023213126e-05,
+      "loss": 0.5842,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901413652327336e-05,
+      "loss": 0.5629,
+      "step": 1821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012840215770935e-05,
+      "loss": 0.547,
+      "step": 1822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011543100821983e-05,
+      "loss": 0.469,
+      "step": 1823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010245178542713e-05,
+      "loss": 0.5671,
+      "step": 1824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9008946449049407e-05,
+      "loss": 0.4661,
+      "step": 1825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900764691245841e-05,
+      "loss": 0.5331,
+      "step": 1826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900634656888615e-05,
+      "loss": 0.5037,
+      "step": 1827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900504541844912e-05,
+      "loss": 0.4931,
+      "step": 1828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003743461263887e-05,
+      "loss": 0.5196,
+      "step": 1829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002440697447094e-05,
+      "loss": 0.5998,
+      "step": 1830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001137127115452e-05,
+      "loss": 0.5195,
+      "step": 1831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899983275038575e-05,
+      "loss": 0.5193,
+      "step": 1832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998527567374837e-05,
+      "loss": 0.5582,
+      "step": 1833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899722157819965e-05,
+      "loss": 0.5254,
+      "step": 1834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899591478297719e-05,
+      "loss": 0.5114,
+      "step": 1835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994607181824528e-05,
+      "loss": 0.5097,
+      "step": 1836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899329877485881e-05,
+      "loss": 0.5013,
+      "step": 1837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899198956219725e-05,
+      "loss": 0.5233,
+      "step": 1838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990679543957146e-05,
+      "loss": 0.5846,
+      "step": 1839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898936872025586e-05,
+      "loss": 0.5888,
+      "step": 1840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898805709121082e-05,
+      "loss": 0.4768,
+      "step": 1841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898674465693954e-05,
+      "loss": 0.5762,
+      "step": 1842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985431417559597e-05,
+      "loss": 0.5167,
+      "step": 1843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898411737318864e-05,
+      "loss": 0.4959,
+      "step": 1844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898280252394439e-05,
+      "loss": 0.5364,
+      "step": 1845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898148686994465e-05,
+      "loss": 0.5462,
+      "step": 1846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980170411307283e-05,
+      "loss": 0.5768,
+      "step": 1847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978853148150223e-05,
+      "loss": 0.5026,
+      "step": 1848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977535080591493e-05,
+      "loss": 0.5928,
+      "step": 1849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897621620874917e-05,
+      "loss": 0.5437,
+      "step": 1850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897489653274141e-05,
+      "loss": 0.5619,
+      "step": 1851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897357605268644e-05,
+      "loss": 0.5149,
+      "step": 1852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972254768702565e-05,
+      "loss": 0.5617,
+      "step": 1853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897093268090815e-05,
+      "loss": 0.5547,
+      "step": 1854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896960978942164e-05,
+      "loss": 0.54,
+      "step": 1855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968286094361555e-05,
+      "loss": 0.5224,
+      "step": 1856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966961595846482e-05,
+      "loss": 0.5241,
+      "step": 1857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965636293995072e-05,
+      "loss": 0.5608,
+      "step": 1858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964310188926072e-05,
+      "loss": 0.5722,
+      "step": 1859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962983280758274e-05,
+      "loss": 0.5066,
+      "step": 1860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961655569610557e-05,
+      "loss": 0.5196,
+      "step": 1861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896032705560187e-05,
+      "loss": 0.5367,
+      "step": 1862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958997738851232e-05,
+      "loss": 0.5234,
+      "step": 1863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957667619477732e-05,
+      "loss": 0.546,
+      "step": 1864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956336697600535e-05,
+      "loss": 0.5506,
+      "step": 1865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955004973338877e-05,
+      "loss": 0.575,
+      "step": 1866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895367244681207e-05,
+      "loss": 0.5594,
+      "step": 1867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952339118139478e-05,
+      "loss": 0.5364,
+      "step": 1868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951004987440567e-05,
+      "loss": 0.5656,
+      "step": 1869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949670054834853e-05,
+      "loss": 0.5547,
+      "step": 1870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894833432044193e-05,
+      "loss": 0.5438,
+      "step": 1871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946997784381473e-05,
+      "loss": 0.5405,
+      "step": 1872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945660446773208e-05,
+      "loss": 0.5258,
+      "step": 1873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894432230773695e-05,
+      "loss": 0.4854,
+      "step": 1874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942983367392584e-05,
+      "loss": 0.5061,
+      "step": 1875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941643625860057e-05,
+      "loss": 0.5517,
+      "step": 1876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89403030832594e-05,
+      "loss": 0.5181,
+      "step": 1877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938961739710708e-05,
+      "loss": 0.4822,
+      "step": 1878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893761959533415e-05,
+      "loss": 0.5265,
+      "step": 1879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893627665024997e-05,
+      "loss": 0.4922,
+      "step": 1880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8934932904578477e-05,
+      "loss": 0.5449,
+      "step": 1881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933588358440052e-05,
+      "loss": 0.5832,
+      "step": 1882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932243011955154e-05,
+      "loss": 0.5753,
+      "step": 1883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930896865244316e-05,
+      "loss": 0.5005,
+      "step": 1884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929549918428124e-05,
+      "loss": 0.5602,
+      "step": 1885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928202171627264e-05,
+      "loss": 0.5401,
+      "step": 1886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892685362496247e-05,
+      "loss": 0.4847,
+      "step": 1887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925504278554555e-05,
+      "loss": 0.5054,
+      "step": 1888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924154132524412e-05,
+      "loss": 0.5403,
+      "step": 1889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892280318699299e-05,
+      "loss": 0.5025,
+      "step": 1890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921451442081323e-05,
+      "loss": 0.54,
+      "step": 1891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892009889791051e-05,
+      "loss": 0.5467,
+      "step": 1892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8918745554601726e-05,
+      "loss": 0.5833,
+      "step": 1893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917391412276212e-05,
+      "loss": 0.5444,
+      "step": 1894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916036471055288e-05,
+      "loss": 0.5994,
+      "step": 1895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914680731060336e-05,
+      "loss": 0.5151,
+      "step": 1896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913324192412817e-05,
+      "loss": 0.5188,
+      "step": 1897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891196685523426e-05,
+      "loss": 0.5203,
+      "step": 1898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910608719646266e-05,
+      "loss": 0.5303,
+      "step": 1899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909249785770513e-05,
+      "loss": 0.4983,
+      "step": 1900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890789005372874e-05,
+      "loss": 0.521,
+      "step": 1901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906529523642767e-05,
+      "loss": 0.5196,
+      "step": 1902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905168195634483e-05,
+      "loss": 0.5234,
+      "step": 1903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903806069825842e-05,
+      "loss": 0.5589,
+      "step": 1904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890244314633888e-05,
+      "loss": 0.5597,
+      "step": 1905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901079425295694e-05,
+      "loss": 0.5299,
+      "step": 1906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899714906818464e-05,
+      "loss": 0.504,
+      "step": 1907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889834959102943e-05,
+      "loss": 0.5459,
+      "step": 1908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889698347805091e-05,
+      "loss": 0.5416,
+      "step": 1909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889561656800529e-05,
+      "loss": 0.4925,
+      "step": 1910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894248861015036e-05,
+      "loss": 0.5871,
+      "step": 1911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892880357202673e-05,
+      "loss": 0.5981,
+      "step": 1912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891511056690804e-05,
+      "loss": 0.5473,
+      "step": 1913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890140959602105e-05,
+      "loss": 0.5138,
+      "step": 1914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888877006605932e-05,
+      "loss": 0.5336,
+      "step": 1915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888739837618526e-05,
+      "loss": 0.5295,
+      "step": 1916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888602589010282e-05,
+      "loss": 0.5084,
+      "step": 1917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884652607934954e-05,
+      "loss": 0.5222,
+      "step": 1918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883278529804694e-05,
+      "loss": 0.5402,
+      "step": 1919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881903655835146e-05,
+      "loss": 0.5016,
+      "step": 1920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8880527986149473e-05,
+      "loss": 0.5376,
+      "step": 1921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887915152087093e-05,
+      "loss": 0.4936,
+      "step": 1922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887777426012282e-05,
+      "loss": 0.5456,
+      "step": 1923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876396204028543e-05,
+      "loss": 0.5517,
+      "step": 1924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875017352711547e-05,
+      "loss": 0.5512,
+      "step": 1925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873637706295363e-05,
+      "loss": 0.5029,
+      "step": 1926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88722572649036e-05,
+      "loss": 0.4678,
+      "step": 1927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870876028659916e-05,
+      "loss": 0.5816,
+      "step": 1928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886949399768806e-05,
+      "loss": 0.4983,
+      "step": 1929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8868111172111853e-05,
+      "loss": 0.5979,
+      "step": 1930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866727552055174e-05,
+      "loss": 0.5696,
+      "step": 1931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865343137641974e-05,
+      "loss": 0.4827,
+      "step": 1932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863957928996287e-05,
+      "loss": 0.5507,
+      "step": 1933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862571926242212e-05,
+      "loss": 0.5858,
+      "step": 1934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861185129503918e-05,
+      "loss": 0.6067,
+      "step": 1935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859797538905648e-05,
+      "loss": 0.5836,
+      "step": 1936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858409154571705e-05,
+      "loss": 0.5719,
+      "step": 1937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857019976626482e-05,
+      "loss": 0.5762,
+      "step": 1938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885563000519443e-05,
+      "loss": 0.5431,
+      "step": 1939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854239240400073e-05,
+      "loss": 0.6123,
+      "step": 1940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852847682368013e-05,
+      "loss": 0.5745,
+      "step": 1941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885145533122291e-05,
+      "loss": 0.5369,
+      "step": 1942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850062187089507e-05,
+      "loss": 0.538,
+      "step": 1943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884866825009261e-05,
+      "loss": 0.481,
+      "step": 1944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847273520357108e-05,
+      "loss": 0.5227,
+      "step": 1945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845877998007943e-05,
+      "loss": 0.5688,
+      "step": 1946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844481683170145e-05,
+      "loss": 0.546,
+      "step": 1947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8843084575968803e-05,
+      "loss": 0.5277,
+      "step": 1948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841686676529085e-05,
+      "loss": 0.5528,
+      "step": 1949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884028798497622e-05,
+      "loss": 0.5519,
+      "step": 1950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8838888501435523e-05,
+      "loss": 0.5338,
+      "step": 1951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883748822603237e-05,
+      "loss": 0.5726,
+      "step": 1952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836087158892205e-05,
+      "loss": 0.4509,
+      "step": 1953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883468530014055e-05,
+      "loss": 0.5182,
+      "step": 1954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833282649903e-05,
+      "loss": 0.5386,
+      "step": 1955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831879208305203e-05,
+      "loss": 0.5273,
+      "step": 1956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830474975472904e-05,
+      "loss": 0.54,
+      "step": 1957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829069951531904e-05,
+      "loss": 0.5186,
+      "step": 1958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882766413660808e-05,
+      "loss": 0.5837,
+      "step": 1959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882625753082736e-05,
+      "loss": 0.5926,
+      "step": 1960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824850134315776e-05,
+      "loss": 0.5239,
+      "step": 1961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823441947199413e-05,
+      "loss": 0.5361,
+      "step": 1962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822032969604423e-05,
+      "loss": 0.5379,
+      "step": 1963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820623201657037e-05,
+      "loss": 0.5184,
+      "step": 1964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.5929,
+      "step": 1965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881780129521034e-05,
+      "loss": 0.5194,
+      "step": 1966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881638915696384e-05,
+      "loss": 0.4913,
+      "step": 1967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814976228870565e-05,
+      "loss": 0.481,
+      "step": 1968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813562511057097e-05,
+      "loss": 0.5455,
+      "step": 1969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812148003650082e-05,
+      "loss": 0.545,
+      "step": 1970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810732706776255e-05,
+      "loss": 0.5786,
+      "step": 1971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809316620562402e-05,
+      "loss": 0.5042,
+      "step": 1972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807899745135392e-05,
+      "loss": 0.5153,
+      "step": 1973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880648208062216e-05,
+      "loss": 0.5238,
+      "step": 1974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880506362714971e-05,
+      "loss": 0.5289,
+      "step": 1975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880364438484512e-05,
+      "loss": 0.5343,
+      "step": 1976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802224353835542e-05,
+      "loss": 0.5507,
+      "step": 1977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800803534248186e-05,
+      "loss": 0.5084,
+      "step": 1978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799381926210348e-05,
+      "loss": 0.4823,
+      "step": 1979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797959529849385e-05,
+      "loss": 0.5855,
+      "step": 1980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879653634529273e-05,
+      "loss": 0.4954,
+      "step": 1981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879511237266788e-05,
+      "loss": 0.5351,
+      "step": 1982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879368761210241e-05,
+      "loss": 0.5599,
+      "step": 1983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792262063723958e-05,
+      "loss": 0.5999,
+      "step": 1984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879083572766024e-05,
+      "loss": 0.5597,
+      "step": 1985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789408604039038e-05,
+      "loss": 0.497,
+      "step": 1986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787980692988207e-05,
+      "loss": 0.5147,
+      "step": 1987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786551994635672e-05,
+      "loss": 0.5519,
+      "step": 1988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785122509109425e-05,
+      "loss": 0.5969,
+      "step": 1989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878369223653753e-05,
+      "loss": 0.4891,
+      "step": 1990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782261177048132e-05,
+      "loss": 0.5703,
+      "step": 1991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780829330769426e-05,
+      "loss": 0.5512,
+      "step": 1992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779396697829697e-05,
+      "loss": 0.4998,
+      "step": 1993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877796327835729e-05,
+      "loss": 0.5182,
+      "step": 1994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877652907248062e-05,
+      "loss": 0.5372,
+      "step": 1995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877509408032818e-05,
+      "loss": 0.568,
+      "step": 1996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773658302028525e-05,
+      "loss": 0.5682,
+      "step": 1997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8772221737710285e-05,
+      "loss": 0.5243,
+      "step": 1998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770784387502164e-05,
+      "loss": 0.5038,
+      "step": 1999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876934625153292e-05,
+      "loss": 0.5009,
+      "step": 2000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876790732993141e-05,
+      "loss": 0.5362,
+      "step": 2001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876646762282653e-05,
+      "loss": 0.4954,
+      "step": 2002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765027130347272e-05,
+      "loss": 0.5227,
+      "step": 2003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763585852622684e-05,
+      "loss": 0.4991,
+      "step": 2004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762143789781884e-05,
+      "loss": 0.4974,
+      "step": 2005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760700941954066e-05,
+      "loss": 0.514,
+      "step": 2006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759257309268498e-05,
+      "loss": 0.5513,
+      "step": 2007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757812891854507e-05,
+      "loss": 0.5087,
+      "step": 2008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756367689841496e-05,
+      "loss": 0.5446,
+      "step": 2009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875492170335894e-05,
+      "loss": 0.5652,
+      "step": 2010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875347493253638e-05,
+      "loss": 0.5544,
+      "step": 2011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752027377503438e-05,
+      "loss": 0.5573,
+      "step": 2012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875057903838979e-05,
+      "loss": 0.4887,
+      "step": 2013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749129915325195e-05,
+      "loss": 0.5591,
+      "step": 2014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747680008439472e-05,
+      "loss": 0.5113,
+      "step": 2015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746229317862527e-05,
+      "loss": 0.5351,
+      "step": 2016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8744777843724312e-05,
+      "loss": 0.5206,
+      "step": 2017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743325586154872e-05,
+      "loss": 0.4951,
+      "step": 2018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741872545284308e-05,
+      "loss": 0.5536,
+      "step": 2019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87404187212428e-05,
+      "loss": 0.548,
+      "step": 2020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738964114160586e-05,
+      "loss": 0.5357,
+      "step": 2021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873750872416799e-05,
+      "loss": 0.5166,
+      "step": 2022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873605255139539e-05,
+      "loss": 0.5296,
+      "step": 2023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873459559597325e-05,
+      "loss": 0.5508,
+      "step": 2024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733137858032093e-05,
+      "loss": 0.5441,
+      "step": 2025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873167933770252e-05,
+      "loss": 0.5342,
+      "step": 2026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730220035115188e-05,
+      "loss": 0.5223,
+      "step": 2027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872875995040084e-05,
+      "loss": 0.5291,
+      "step": 2028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727299083690284e-05,
+      "loss": 0.5448,
+      "step": 2029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725837435114395e-05,
+      "loss": 0.5478,
+      "step": 2030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872437500480412e-05,
+      "loss": 0.5592,
+      "step": 2031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722911792890476e-05,
+      "loss": 0.5332,
+      "step": 2032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721447799504545e-05,
+      "loss": 0.5741,
+      "step": 2033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719983024777493e-05,
+      "loss": 0.5221,
+      "step": 2034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871851746884054e-05,
+      "loss": 0.4697,
+      "step": 2035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717051131824984e-05,
+      "loss": 0.5026,
+      "step": 2036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87155840138622e-05,
+      "loss": 0.5559,
+      "step": 2037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871411611508361e-05,
+      "loss": 0.4987,
+      "step": 2038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8712647435620733e-05,
+      "loss": 0.5201,
+      "step": 2039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871117797560514e-05,
+      "loss": 0.4573,
+      "step": 2040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870970773516848e-05,
+      "loss": 0.5298,
+      "step": 2041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870823671444247e-05,
+      "loss": 0.5369,
+      "step": 2042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870676491355889e-05,
+      "loss": 0.4739,
+      "step": 2043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8705292332649608e-05,
+      "loss": 0.5525,
+      "step": 2044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870381897184654e-05,
+      "loss": 0.4926,
+      "step": 2045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702344831281688e-05,
+      "loss": 0.4629,
+      "step": 2046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700869911087115e-05,
+      "loss": 0.4798,
+      "step": 2047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699394211394957e-05,
+      "loss": 0.625,
+      "step": 2048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869791773233742e-05,
+      "loss": 0.5414,
+      "step": 2049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696440474046785e-05,
+      "loss": 0.4854,
+      "step": 2050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694962436655387e-05,
+      "loss": 0.5194,
+      "step": 2051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869348362029565e-05,
+      "loss": 0.5171,
+      "step": 2052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692004025100054e-05,
+      "loss": 0.5496,
+      "step": 2053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690523651201155e-05,
+      "loss": 0.5352,
+      "step": 2054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689042498731575e-05,
+      "loss": 0.5529,
+      "step": 2055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687560567824014e-05,
+      "loss": 0.533,
+      "step": 2056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686077858611235e-05,
+      "loss": 0.5062,
+      "step": 2057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684594371226064e-05,
+      "loss": 0.5465,
+      "step": 2058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868311010580141e-05,
+      "loss": 0.5429,
+      "step": 2059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681625062470244e-05,
+      "loss": 0.5475,
+      "step": 2060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680139241365614e-05,
+      "loss": 0.5056,
+      "step": 2061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678652642620627e-05,
+      "loss": 0.4692,
+      "step": 2062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867716526636847e-05,
+      "loss": 0.5477,
+      "step": 2063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675677112742388e-05,
+      "loss": 0.5697,
+      "step": 2064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674188181875706e-05,
+      "loss": 0.5619,
+      "step": 2065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672698473901815e-05,
+      "loss": 0.5455,
+      "step": 2066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867120798895418e-05,
+      "loss": 0.5445,
+      "step": 2067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8669716727166322e-05,
+      "loss": 0.5719,
+      "step": 2068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668224688671848e-05,
+      "loss": 0.5599,
+      "step": 2069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666731873604425e-05,
+      "loss": 0.5012,
+      "step": 2070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866523828209779e-05,
+      "loss": 0.5752,
+      "step": 2071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663743914285754e-05,
+      "loss": 0.4847,
+      "step": 2072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8662248770302198e-05,
+      "loss": 0.5177,
+      "step": 2073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8660752850281066e-05,
+      "loss": 0.5652,
+      "step": 2074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659256154356375e-05,
+      "loss": 0.5301,
+      "step": 2075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865775868266221e-05,
+      "loss": 0.5345,
+      "step": 2076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656260435332732e-05,
+      "loss": 0.4867,
+      "step": 2077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865476141250217e-05,
+      "loss": 0.558,
+      "step": 2078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653261614304804e-05,
+      "loss": 0.5006,
+      "step": 2079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651761040875015e-05,
+      "loss": 0.532,
+      "step": 2080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650259692347226e-05,
+      "loss": 0.4839,
+      "step": 2081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648757568855946e-05,
+      "loss": 0.524,
+      "step": 2082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647254670535744e-05,
+      "loss": 0.5461,
+      "step": 2083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864575099752127e-05,
+      "loss": 0.5323,
+      "step": 2084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644246549947226e-05,
+      "loss": 0.5239,
+      "step": 2085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86427413279484e-05,
+      "loss": 0.5465,
+      "step": 2086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641235331659643e-05,
+      "loss": 0.5659,
+      "step": 2087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863972856121587e-05,
+      "loss": 0.4889,
+      "step": 2088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863822101675207e-05,
+      "loss": 0.4967,
+      "step": 2089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636712698403307e-05,
+      "loss": 0.5177,
+      "step": 2090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863520360630471e-05,
+      "loss": 0.5854,
+      "step": 2091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633693740591465e-05,
+      "loss": 0.5618,
+      "step": 2092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632183101398847e-05,
+      "loss": 0.4893,
+      "step": 2093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630671688862194e-05,
+      "loss": 0.5638,
+      "step": 2094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629159503116905e-05,
+      "loss": 0.5013,
+      "step": 2095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627646544298463e-05,
+      "loss": 0.5676,
+      "step": 2096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86261328125424e-05,
+      "loss": 0.5908,
+      "step": 2097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624618307984338e-05,
+      "loss": 0.5196,
+      "step": 2098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623103030759954e-05,
+      "loss": 0.5435,
+      "step": 2099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621586981005e-05,
+      "loss": 0.5286,
+      "step": 2100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620070158855303e-05,
+      "loss": 0.5386,
+      "step": 2101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618552564446744e-05,
+      "loss": 0.543,
+      "step": 2102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617034197915284e-05,
+      "loss": 0.5105,
+      "step": 2103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615515059396955e-05,
+      "loss": 0.4834,
+      "step": 2104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613995149027852e-05,
+      "loss": 0.5562,
+      "step": 2105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861247446694414e-05,
+      "loss": 0.5265,
+      "step": 2106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861095301328206e-05,
+      "loss": 0.4603,
+      "step": 2107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860943078817791e-05,
+      "loss": 0.5642,
+      "step": 2108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607907791768067e-05,
+      "loss": 0.5068,
+      "step": 2109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8606384024188972e-05,
+      "loss": 0.5639,
+      "step": 2110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860485948557714e-05,
+      "loss": 0.5311,
+      "step": 2111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603334176069152e-05,
+      "loss": 0.485,
+      "step": 2112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601808095801653e-05,
+      "loss": 0.5107,
+      "step": 2113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600281244911367e-05,
+      "loss": 0.5083,
+      "step": 2114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598753623535085e-05,
+      "loss": 0.5338,
+      "step": 2115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597225231809655e-05,
+      "loss": 0.4774,
+      "step": 2116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595696069872013e-05,
+      "loss": 0.5094,
+      "step": 2117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594166137859148e-05,
+      "loss": 0.5112,
+      "step": 2118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8592635435908126e-05,
+      "loss": 0.5181,
+      "step": 2119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859110396415608e-05,
+      "loss": 0.546,
+      "step": 2120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589571722740213e-05,
+      "loss": 0.5427,
+      "step": 2121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588038711797794e-05,
+      "loss": 0.5874,
+      "step": 2122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858650493146617e-05,
+      "loss": 0.5031,
+      "step": 2123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584970381882745e-05,
+      "loss": 0.5277,
+      "step": 2124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858343506318499e-05,
+      "loss": 0.5045,
+      "step": 2125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858189897551046e-05,
+      "loss": 0.505,
+      "step": 2126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580362118996774e-05,
+      "loss": 0.538,
+      "step": 2127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857882449378161e-05,
+      "loss": 0.5013,
+      "step": 2128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577286100002723e-05,
+      "loss": 0.4564,
+      "step": 2129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575746937797933e-05,
+      "loss": 0.5281,
+      "step": 2130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857420700730514e-05,
+      "loss": 0.5268,
+      "step": 2131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572666308662295e-05,
+      "loss": 0.5494,
+      "step": 2132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857112484200743e-05,
+      "loss": 0.4811,
+      "step": 2133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856958260747864e-05,
+      "loss": 0.4787,
+      "step": 2134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85680396052141e-05,
+      "loss": 0.5065,
+      "step": 2135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566495835352033e-05,
+      "loss": 0.5493,
+      "step": 2136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564951298030754e-05,
+      "loss": 0.5319,
+      "step": 2137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563405993388626e-05,
+      "loss": 0.5107,
+      "step": 2138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85618599215641e-05,
+      "loss": 0.4887,
+      "step": 2139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856031308269568e-05,
+      "loss": 0.4787,
+      "step": 2140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558765476921942e-05,
+      "loss": 0.5325,
+      "step": 2141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557217104381543e-05,
+      "loss": 0.4541,
+      "step": 2142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555667965213192e-05,
+      "loss": 0.5266,
+      "step": 2143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554118059555678e-05,
+      "loss": 0.4914,
+      "step": 2144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552567387547852e-05,
+      "loss": 0.5064,
+      "step": 2145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855101594932864e-05,
+      "loss": 0.5539,
+      "step": 2146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854946374503703e-05,
+      "loss": 0.5219,
+      "step": 2147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854791077481208e-05,
+      "loss": 0.4965,
+      "step": 2148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546357038792918e-05,
+      "loss": 0.4641,
+      "step": 2149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544802537118745e-05,
+      "loss": 0.4682,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543247269928825e-05,
+      "loss": 0.5398,
+      "step": 2151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541691237362492e-05,
+      "loss": 0.5394,
+      "step": 2152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540134439559147e-05,
+      "loss": 0.4833,
+      "step": 2153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538576876658263e-05,
+      "loss": 0.5196,
+      "step": 2154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537018548799377e-05,
+      "loss": 0.5244,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85354594561221e-05,
+      "loss": 0.5729,
+      "step": 2156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533899598766106e-05,
+      "loss": 0.4886,
+      "step": 2157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532338976871147e-05,
+      "loss": 0.4876,
+      "step": 2158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853077759057703e-05,
+      "loss": 0.5014,
+      "step": 2159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529215440023634e-05,
+      "loss": 0.4985,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852765252535092e-05,
+      "loss": 0.4759,
+      "step": 2161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85260888466989e-05,
+      "loss": 0.4656,
+      "step": 2162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852452440420766e-05,
+      "loss": 0.5532,
+      "step": 2163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522959198017363e-05,
+      "loss": 0.5589,
+      "step": 2164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852139322826823e-05,
+      "loss": 0.4795,
+      "step": 2165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851982649510055e-05,
+      "loss": 0.5662,
+      "step": 2166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851825899865469e-05,
+      "loss": 0.5179,
+      "step": 2167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516690739071078e-05,
+      "loss": 0.4767,
+      "step": 2168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851512171649021e-05,
+      "loss": 0.5355,
+      "step": 2169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513551931052654e-05,
+      "loss": 0.4867,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851198138289904e-05,
+      "loss": 0.5457,
+      "step": 2171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510410072170082e-05,
+      "loss": 0.5791,
+      "step": 2172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508837999006544e-05,
+      "loss": 0.4972,
+      "step": 2173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8507265163549264e-05,
+      "loss": 0.5963,
+      "step": 2174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850569156593915e-05,
+      "loss": 0.4535,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504117206317183e-05,
+      "loss": 0.4868,
+      "step": 2176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502542084824403e-05,
+      "loss": 0.5487,
+      "step": 2177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500966201601927e-05,
+      "loss": 0.5363,
+      "step": 2178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499389556790932e-05,
+      "loss": 0.5359,
+      "step": 2179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849781215053267e-05,
+      "loss": 0.5557,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849623398296846e-05,
+      "loss": 0.546,
+      "step": 2181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494655054239675e-05,
+      "loss": 0.5203,
+      "step": 2182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493075364487783e-05,
+      "loss": 0.5091,
+      "step": 2183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491494913854303e-05,
+      "loss": 0.5261,
+      "step": 2184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8489913702480825e-05,
+      "loss": 0.5117,
+      "step": 2185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488331730509005e-05,
+      "loss": 0.4615,
+      "step": 2186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486748998080567e-05,
+      "loss": 0.485,
+      "step": 2187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848516550533731e-05,
+      "loss": 0.4973,
+      "step": 2188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84835812524211e-05,
+      "loss": 0.5844,
+      "step": 2189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481996239473855e-05,
+      "loss": 0.5007,
+      "step": 2190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848041046663759e-05,
+      "loss": 0.5451,
+      "step": 2191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847882393405436e-05,
+      "loss": 0.5356,
+      "step": 2192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477236641866306e-05,
+      "loss": 0.5367,
+      "step": 2193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847564859021563e-05,
+      "loss": 0.5614,
+      "step": 2194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84740597792446e-05,
+      "loss": 0.5018,
+      "step": 2195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847247020909556e-05,
+      "loss": 0.491,
+      "step": 2196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470879879910916e-05,
+      "loss": 0.5517,
+      "step": 2197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469288791833135e-05,
+      "loss": 0.5481,
+      "step": 2198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467696945004774e-05,
+      "loss": 0.5135,
+      "step": 2199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466104339568434e-05,
+      "loss": 0.515,
+      "step": 2200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84645109756668e-05,
+      "loss": 0.5308,
+      "step": 2201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846291685344261e-05,
+      "loss": 0.5462,
+      "step": 2202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846132197303869e-05,
+      "loss": 0.4766,
+      "step": 2203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459726334597916e-05,
+      "loss": 0.5318,
+      "step": 2204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458129938263242e-05,
+      "loss": 0.4984,
+      "step": 2205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456532784177684e-05,
+      "loss": 0.5014,
+      "step": 2206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454934872484334e-05,
+      "loss": 0.5029,
+      "step": 2207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845333620332634e-05,
+      "loss": 0.5325,
+      "step": 2208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451736776846927e-05,
+      "loss": 0.5381,
+      "step": 2209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450136593189384e-05,
+      "loss": 0.5417,
+      "step": 2210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448535652497073e-05,
+      "loss": 0.5631,
+      "step": 2211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446933954913416e-05,
+      "loss": 0.4745,
+      "step": 2212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445331500581905e-05,
+      "loss": 0.5841,
+      "step": 2213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443728289646105e-05,
+      "loss": 0.5386,
+      "step": 2214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442124322249646e-05,
+      "loss": 0.4685,
+      "step": 2215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844051959853622e-05,
+      "loss": 0.5521,
+      "step": 2216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84389141186496e-05,
+      "loss": 0.5033,
+      "step": 2217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437307882733607e-05,
+      "loss": 0.5551,
+      "step": 2218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843570089093215e-05,
+      "loss": 0.5459,
+      "step": 2219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434093143389194e-05,
+      "loss": 0.4815,
+      "step": 2220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843248464024878e-05,
+      "loss": 0.5769,
+      "step": 2221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430875381655e-05,
+      "loss": 0.5148,
+      "step": 2222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429265367752033e-05,
+      "loss": 0.5149,
+      "step": 2223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842765459868412e-05,
+      "loss": 0.4842,
+      "step": 2224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842604307459556e-05,
+      "loss": 0.5693,
+      "step": 2225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842443079563073e-05,
+      "loss": 0.5106,
+      "step": 2226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422817761934074e-05,
+      "loss": 0.5313,
+      "step": 2227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84212039736501e-05,
+      "loss": 0.5508,
+      "step": 2228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419589430923383e-05,
+      "loss": 0.5376,
+      "step": 2229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417974133898563e-05,
+      "loss": 0.5405,
+      "step": 2230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416358082720364e-05,
+      "loss": 0.5445,
+      "step": 2231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8414741277533558e-05,
+      "loss": 0.4902,
+      "step": 2232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413123718482993e-05,
+      "loss": 0.5562,
+      "step": 2233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411505405713584e-05,
+      "loss": 0.5175,
+      "step": 2234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840988633937031e-05,
+      "loss": 0.4604,
+      "step": 2235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408266519598224e-05,
+      "loss": 0.5217,
+      "step": 2236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406645946542446e-05,
+      "loss": 0.5448,
+      "step": 2237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405024620348157e-05,
+      "loss": 0.5305,
+      "step": 2238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403402541160608e-05,
+      "loss": 0.5591,
+      "step": 2239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401779709125117e-05,
+      "loss": 0.5103,
+      "step": 2240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840015612438708e-05,
+      "loss": 0.5049,
+      "step": 2241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839853178709194e-05,
+      "loss": 0.5633,
+      "step": 2242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396906697385227e-05,
+      "loss": 0.5127,
+      "step": 2243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839528085541253e-05,
+      "loss": 0.4836,
+      "step": 2244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393654261319504e-05,
+      "loss": 0.533,
+      "step": 2245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392026915251867e-05,
+      "loss": 0.5465,
+      "step": 2246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839039881735542e-05,
+      "loss": 0.54,
+      "step": 2247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388769967776014e-05,
+      "loss": 0.5487,
+      "step": 2248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387140366659585e-05,
+      "loss": 0.5254,
+      "step": 2249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385510014152114e-05,
+      "loss": 0.5077,
+      "step": 2250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383878910399674e-05,
+      "loss": 0.4783,
+      "step": 2251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838224705554838e-05,
+      "loss": 0.4918,
+      "step": 2252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380614449744442e-05,
+      "loss": 0.4813,
+      "step": 2253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378981093134108e-05,
+      "loss": 0.5,
+      "step": 2254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837734698586372e-05,
+      "loss": 0.517,
+      "step": 2255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837571212807967e-05,
+      "loss": 0.5245,
+      "step": 2256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374076519928422e-05,
+      "loss": 0.4647,
+      "step": 2257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372440161556508e-05,
+      "loss": 0.5124,
+      "step": 2258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837080305311053e-05,
+      "loss": 0.5632,
+      "step": 2259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369165194737148e-05,
+      "loss": 0.5786,
+      "step": 2260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83675265865831e-05,
+      "loss": 0.6158,
+      "step": 2261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365887228795184e-05,
+      "loss": 0.5501,
+      "step": 2262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364247121520268e-05,
+      "loss": 0.5235,
+      "step": 2263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362606264905288e-05,
+      "loss": 0.5281,
+      "step": 2264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360964659097242e-05,
+      "loss": 0.5397,
+      "step": 2265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359322304243202e-05,
+      "loss": 0.5215,
+      "step": 2266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357679200490308e-05,
+      "loss": 0.4882,
+      "step": 2267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356035347985757e-05,
+      "loss": 0.5404,
+      "step": 2268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835439074687682e-05,
+      "loss": 0.5525,
+      "step": 2269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352745397310837e-05,
+      "loss": 0.5535,
+      "step": 2270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351099299435212e-05,
+      "loss": 0.5529,
+      "step": 2271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8349452453397414e-05,
+      "loss": 0.5464,
+      "step": 2272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834780485934498e-05,
+      "loss": 0.5412,
+      "step": 2273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346156517425523e-05,
+      "loss": 0.5034,
+      "step": 2274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344507427786705e-05,
+      "loss": 0.5197,
+      "step": 2275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342857590576276e-05,
+      "loss": 0.4967,
+      "step": 2276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341207005942033e-05,
+      "loss": 0.4989,
+      "step": 2277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339555674031858e-05,
+      "loss": 0.5114,
+      "step": 2278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8337903594993683e-05,
+      "loss": 0.483,
+      "step": 2279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833625076897552e-05,
+      "loss": 0.5614,
+      "step": 2280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833459719612544e-05,
+      "loss": 0.5306,
+      "step": 2281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332942876591592e-05,
+      "loss": 0.5658,
+      "step": 2282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331287810522178e-05,
+      "loss": 0.5599,
+      "step": 2283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832963199806547e-05,
+      "loss": 0.5109,
+      "step": 2284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327975439369816e-05,
+      "loss": 0.5351,
+      "step": 2285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832631813458362e-05,
+      "loss": 0.5035,
+      "step": 2286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324660083855355e-05,
+      "loss": 0.4726,
+      "step": 2287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832300128733357e-05,
+      "loss": 0.5259,
+      "step": 2288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832134174516687e-05,
+      "loss": 0.4716,
+      "step": 2289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319681457503936e-05,
+      "loss": 0.4996,
+      "step": 2290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83180204244935e-05,
+      "loss": 0.5201,
+      "step": 2291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316358646284382e-05,
+      "loss": 0.491,
+      "step": 2292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.5266,
+      "step": 2293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313032854865657e-05,
+      "loss": 0.5355,
+      "step": 2294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311368841954004e-05,
+      "loss": 0.4827,
+      "step": 2295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8309704084439572e-05,
+      "loss": 0.5192,
+      "step": 2296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83080385824715e-05,
+      "loss": 0.5508,
+      "step": 2297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8306372336199e-05,
+      "loss": 0.5433,
+      "step": 2298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304705345771346e-05,
+      "loss": 0.5004,
+      "step": 2299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303037611337887e-05,
+      "loss": 0.4924,
+      "step": 2300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301369133048025e-05,
+      "loss": 0.5001,
+      "step": 2301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299699911051244e-05,
+      "loss": 0.5467,
+      "step": 2302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829802994549708e-05,
+      "loss": 0.4828,
+      "step": 2303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829635923653515e-05,
+      "loss": 0.5421,
+      "step": 2304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294687784315125e-05,
+      "loss": 0.4813,
+      "step": 2305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293015588986752e-05,
+      "loss": 0.5301,
+      "step": 2306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829134265069983e-05,
+      "loss": 0.547,
+      "step": 2307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828966896960425e-05,
+      "loss": 0.5082,
+      "step": 2308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287994545849948e-05,
+      "loss": 0.5434,
+      "step": 2309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828631937958693e-05,
+      "loss": 0.5432,
+      "step": 2310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284643470965273e-05,
+      "loss": 0.52,
+      "step": 2311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828296682013512e-05,
+      "loss": 0.4828,
+      "step": 2312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828128942724668e-05,
+      "loss": 0.5442,
+      "step": 2313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279611292450224e-05,
+      "loss": 0.5181,
+      "step": 2314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277932415896102e-05,
+      "loss": 0.5436,
+      "step": 2315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276252797734713e-05,
+      "loss": 0.4921,
+      "step": 2316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827457243811654e-05,
+      "loss": 0.4557,
+      "step": 2317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827289133719212e-05,
+      "loss": 0.5433,
+      "step": 2318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271209495112056e-05,
+      "loss": 0.5694,
+      "step": 2319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826952691202703e-05,
+      "loss": 0.607,
+      "step": 2320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826784358808777e-05,
+      "loss": 0.4835,
+      "step": 2321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826615952344509e-05,
+      "loss": 0.5335,
+      "step": 2322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264474718249868e-05,
+      "loss": 0.5048,
+      "step": 2323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262789172653034e-05,
+      "loss": 0.5226,
+      "step": 2324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82611028868056e-05,
+      "loss": 0.5229,
+      "step": 2325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825941586085863e-05,
+      "loss": 0.491,
+      "step": 2326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257728094963272e-05,
+      "loss": 0.5349,
+      "step": 2327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825603958927072e-05,
+      "loss": 0.55,
+      "step": 2328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254350343932256e-05,
+      "loss": 0.51,
+      "step": 2329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252660359099204e-05,
+      "loss": 0.5302,
+      "step": 2330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825096963492298e-05,
+      "loss": 0.4573,
+      "step": 2331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249278171555043e-05,
+      "loss": 0.5373,
+      "step": 2332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8247585969146936e-05,
+      "loss": 0.554,
+      "step": 2333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245893027850255e-05,
+      "loss": 0.5208,
+      "step": 2334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244199347816673e-05,
+      "loss": 0.541,
+      "step": 2335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824250492919792e-05,
+      "loss": 0.5052,
+      "step": 2336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240809772145797e-05,
+      "loss": 0.5103,
+      "step": 2337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239113876812175e-05,
+      "loss": 0.519,
+      "step": 2338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823741724334898e-05,
+      "loss": 0.5308,
+      "step": 2339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235719871908215e-05,
+      "loss": 0.4633,
+      "step": 2340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234021762641946e-05,
+      "loss": 0.5164,
+      "step": 2341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232322915702303e-05,
+      "loss": 0.4618,
+      "step": 2342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230623331241478e-05,
+      "loss": 0.5345,
+      "step": 2343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228923009411743e-05,
+      "loss": 0.5271,
+      "step": 2344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822722195036542e-05,
+      "loss": 0.4979,
+      "step": 2345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225520154254906e-05,
+      "loss": 0.5416,
+      "step": 2346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223817621232666e-05,
+      "loss": 0.494,
+      "step": 2347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222114351451223e-05,
+      "loss": 0.5418,
+      "step": 2348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822041034506317e-05,
+      "loss": 0.5436,
+      "step": 2349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218705602221173e-05,
+      "loss": 0.543,
+      "step": 2350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217000123077948e-05,
+      "loss": 0.4917,
+      "step": 2351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821529390778629e-05,
+      "loss": 0.5285,
+      "step": 2352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821358695649906e-05,
+      "loss": 0.5165,
+      "step": 2353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211879269369176e-05,
+      "loss": 0.546,
+      "step": 2354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210170846549632e-05,
+      "loss": 0.5133,
+      "step": 2355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820846168819348e-05,
+      "loss": 0.4926,
+      "step": 2356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206751794453837e-05,
+      "loss": 0.5263,
+      "step": 2357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8205041165483897e-05,
+      "loss": 0.5018,
+      "step": 2358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203329801436908e-05,
+      "loss": 0.5122,
+      "step": 2359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820161770246619e-05,
+      "loss": 0.5252,
+      "step": 2360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199904868725128e-05,
+      "loss": 0.5187,
+      "step": 2361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819819130036717e-05,
+      "loss": 0.5164,
+      "step": 2362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196476997545832e-05,
+      "loss": 0.502,
+      "step": 2363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81947619604147e-05,
+      "loss": 0.5087,
+      "step": 2364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193046189127416e-05,
+      "loss": 0.5385,
+      "step": 2365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191329683837697e-05,
+      "loss": 0.5283,
+      "step": 2366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189612444699323e-05,
+      "loss": 0.5015,
+      "step": 2367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187894471866133e-05,
+      "loss": 0.5249,
+      "step": 2368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186175765492044e-05,
+      "loss": 0.528,
+      "step": 2369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818445632573103e-05,
+      "loss": 0.5504,
+      "step": 2370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818273615273713e-05,
+      "loss": 0.4724,
+      "step": 2371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181015246664456e-05,
+      "loss": 0.4888,
+      "step": 2372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179293607667177e-05,
+      "loss": 0.5333,
+      "step": 2373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177571235899537e-05,
+      "loss": 0.5268,
+      "step": 2374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817584813151584e-05,
+      "loss": 0.4849,
+      "step": 2375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817412429467045e-05,
+      "loss": 0.4785,
+      "step": 2376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817239972551781e-05,
+      "loss": 0.4707,
+      "step": 2377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170674424212417e-05,
+      "loss": 0.4774,
+      "step": 2378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168948390908845e-05,
+      "loss": 0.5269,
+      "step": 2379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167221625761716e-05,
+      "loss": 0.5389,
+      "step": 2380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165494128925737e-05,
+      "loss": 0.4753,
+      "step": 2381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163765900555666e-05,
+      "loss": 0.5037,
+      "step": 2382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8162036940806337e-05,
+      "loss": 0.5339,
+      "step": 2383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160307249832644e-05,
+      "loss": 0.5194,
+      "step": 2384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815857682778954e-05,
+      "loss": 0.4992,
+      "step": 2385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156845674832066e-05,
+      "loss": 0.4856,
+      "step": 2386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81551137911153e-05,
+      "loss": 0.5781,
+      "step": 2387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153381176794403e-05,
+      "loss": 0.5273,
+      "step": 2388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151647832024597e-05,
+      "loss": 0.4834,
+      "step": 2389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814991375696117e-05,
+      "loss": 0.527,
+      "step": 2390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148178951759477e-05,
+      "loss": 0.5408,
+      "step": 2391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146443416574934e-05,
+      "loss": 0.5342,
+      "step": 2392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144707151563026e-05,
+      "loss": 0.5114,
+      "step": 2393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81429701568793e-05,
+      "loss": 0.5328,
+      "step": 2394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141232432679377e-05,
+      "loss": 0.528,
+      "step": 2395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813949397911893e-05,
+      "loss": 0.5137,
+      "step": 2396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137754796353708e-05,
+      "loss": 0.5475,
+      "step": 2397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136014884539518e-05,
+      "loss": 0.4921,
+      "step": 2398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8134274243832243e-05,
+      "loss": 0.5482,
+      "step": 2399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813253287438782e-05,
+      "loss": 0.4956,
+      "step": 2400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130790776362253e-05,
+      "loss": 0.5393,
+      "step": 2401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812904794991162e-05,
+      "loss": 0.55,
+      "step": 2402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812730439519205e-05,
+      "loss": 0.4899,
+      "step": 2403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125560112359756e-05,
+      "loss": 0.4991,
+      "step": 2404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123815101570996e-05,
+      "loss": 0.5222,
+      "step": 2405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8122069362982108e-05,
+      "loss": 0.4657,
+      "step": 2406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812032289674949e-05,
+      "loss": 0.4857,
+      "step": 2407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118575703029602e-05,
+      "loss": 0.5817,
+      "step": 2408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116827781978977e-05,
+      "loss": 0.5111,
+      "step": 2409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115079133754205e-05,
+      "loss": 0.4931,
+      "step": 2410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113329758511944e-05,
+      "loss": 0.5216,
+      "step": 2411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8111579656408918e-05,
+      "loss": 0.5917,
+      "step": 2412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8109828827601924e-05,
+      "loss": 0.5027,
+      "step": 2413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108077272247805e-05,
+      "loss": 0.5316,
+      "step": 2414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106324990503485e-05,
+      "loss": 0.5679,
+      "step": 2415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810457198252595e-05,
+      "loss": 0.5353,
+      "step": 2416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102818248472244e-05,
+      "loss": 0.5377,
+      "step": 2417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101063788499488e-05,
+      "loss": 0.501,
+      "step": 2418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099308602764857e-05,
+      "loss": 0.5118,
+      "step": 2419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097552691425595e-05,
+      "loss": 0.5227,
+      "step": 2420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095796054639015e-05,
+      "loss": 0.5675,
+      "step": 2421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094038692562487e-05,
+      "loss": 0.547,
+      "step": 2422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092280605353453e-05,
+      "loss": 0.5461,
+      "step": 2423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090521793169417e-05,
+      "loss": 0.5653,
+      "step": 2424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808876225616795e-05,
+      "loss": 0.5718,
+      "step": 2425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087001994506683e-05,
+      "loss": 0.5413,
+      "step": 2426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8085241008343315e-05,
+      "loss": 0.5653,
+      "step": 2427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083479297835612e-05,
+      "loss": 0.495,
+      "step": 2428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081716863141407e-05,
+      "loss": 0.5363,
+      "step": 2429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079953704418584e-05,
+      "loss": 0.5303,
+      "step": 2430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807818982182511e-05,
+      "loss": 0.5529,
+      "step": 2431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8076425215519006e-05,
+      "loss": 0.5248,
+      "step": 2432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8074659885658357e-05,
+      "loss": 0.4855,
+      "step": 2433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072893832401322e-05,
+      "loss": 0.4704,
+      "step": 2434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8071127055906114e-05,
+      "loss": 0.4984,
+      "step": 2435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069359556331018e-05,
+      "loss": 0.5006,
+      "step": 2436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067591333834382e-05,
+      "loss": 0.5521,
+      "step": 2437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806582238857462e-05,
+      "loss": 0.5351,
+      "step": 2438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064052720710205e-05,
+      "loss": 0.5394,
+      "step": 2439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062282330399677e-05,
+      "loss": 0.5594,
+      "step": 2440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060511217801652e-05,
+      "loss": 0.5508,
+      "step": 2441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058739383074795e-05,
+      "loss": 0.518,
+      "step": 2442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805696682637784e-05,
+      "loss": 0.5392,
+      "step": 2443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055193547869594e-05,
+      "loss": 0.5625,
+      "step": 2444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053419547708916e-05,
+      "loss": 0.5355,
+      "step": 2445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805164482605474e-05,
+      "loss": 0.5553,
+      "step": 2446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049869383066053e-05,
+      "loss": 0.5543,
+      "step": 2447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8048093218901925e-05,
+      "loss": 0.4995,
+      "step": 2448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046316333721475e-05,
+      "loss": 0.5169,
+      "step": 2449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044538727683895e-05,
+      "loss": 0.5174,
+      "step": 2450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042760400948426e-05,
+      "loss": 0.4639,
+      "step": 2451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80409813536744e-05,
+      "loss": 0.5108,
+      "step": 2452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039201586021192e-05,
+      "loss": 0.4763,
+      "step": 2453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037421098148247e-05,
+      "loss": 0.4937,
+      "step": 2454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035639890215084e-05,
+      "loss": 0.5477,
+      "step": 2455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803385796238127e-05,
+      "loss": 0.4962,
+      "step": 2456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803207531480645e-05,
+      "loss": 0.537,
+      "step": 2457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030291947650325e-05,
+      "loss": 0.4453,
+      "step": 2458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028507861072672e-05,
+      "loss": 0.4996,
+      "step": 2459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802672305523332e-05,
+      "loss": 0.5184,
+      "step": 2460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024937530292158e-05,
+      "loss": 0.531,
+      "step": 2461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802315128640916e-05,
+      "loss": 0.5079,
+      "step": 2462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021364323744352e-05,
+      "loss": 0.544,
+      "step": 2463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019576642457825e-05,
+      "loss": 0.52,
+      "step": 2464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017788242709725e-05,
+      "loss": 0.5049,
+      "step": 2465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015999124660284e-05,
+      "loss": 0.6129,
+      "step": 2466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014209288469778e-05,
+      "loss": 0.5097,
+      "step": 2467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8012418734298567e-05,
+      "loss": 0.4911,
+      "step": 2468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801062746230705e-05,
+      "loss": 0.5138,
+      "step": 2469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800883547265571e-05,
+      "loss": 0.552,
+      "step": 2470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8007042765505095e-05,
+      "loss": 0.4919,
+      "step": 2471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.80052493410158e-05,
+      "loss": 0.5267,
+      "step": 2472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8003455199348504e-05,
+      "loss": 0.5106,
+      "step": 2473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001660340663934e-05,
+      "loss": 0.47,
+      "step": 2474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999864765122895e-05,
+      "loss": 0.5649,
+      "step": 2475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998068472886245e-05,
+      "loss": 0.5886,
+      "step": 2476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996271464114915e-05,
+      "loss": 0.4932,
+      "step": 2477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994473738969895e-05,
+      "loss": 0.4805,
+      "step": 2478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799267529761224e-05,
+      "loss": 0.4752,
+      "step": 2479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990876140203067e-05,
+      "loss": 0.5581,
+      "step": 2480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989076266903566e-05,
+      "loss": 0.5381,
+      "step": 2481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987275677874977e-05,
+      "loss": 0.5385,
+      "step": 2482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798547437327862e-05,
+      "loss": 0.5651,
+      "step": 2483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983672353275865e-05,
+      "loss": 0.5253,
+      "step": 2484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981869618028157e-05,
+      "loss": 0.4921,
+      "step": 2485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980066167696994e-05,
+      "loss": 0.506,
+      "step": 2486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797826200244395e-05,
+      "loss": 0.4815,
+      "step": 2487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7976457122430656e-05,
+      "loss": 0.507,
+      "step": 2488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974651527818808e-05,
+      "loss": 0.6067,
+      "step": 2489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972845218770167e-05,
+      "loss": 0.5194,
+      "step": 2490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971038195446556e-05,
+      "loss": 0.4485,
+      "step": 2491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969230458009867e-05,
+      "loss": 0.4597,
+      "step": 2492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796742200662205e-05,
+      "loss": 0.5412,
+      "step": 2493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7965612841445114e-05,
+      "loss": 0.4989,
+      "step": 2494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796380296264115e-05,
+      "loss": 0.5343,
+      "step": 2495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79619923703723e-05,
+      "loss": 0.5019,
+      "step": 2496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960181064800774e-05,
+      "loss": 0.4889,
+      "step": 2497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958369046088837e-05,
+      "loss": 0.5371,
+      "step": 2498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795655631439883e-05,
+      "loss": 0.5063,
+      "step": 2499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795474286989315e-05,
+      "loss": 0.4861,
+      "step": 2500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952928712734266e-05,
+      "loss": 0.4817,
+      "step": 2501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79511138430847e-05,
+      "loss": 0.5663,
+      "step": 2502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949298261107046e-05,
+      "loss": 0.5339,
+      "step": 2503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794748196696396e-05,
+      "loss": 0.4919,
+      "step": 2504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945664960818158e-05,
+      "loss": 0.5109,
+      "step": 2505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943847242832424e-05,
+      "loss": 0.5251,
+      "step": 2506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794202881316961e-05,
+      "loss": 0.5177,
+      "step": 2507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940209671992616e-05,
+      "loss": 0.5786,
+      "step": 2508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793838981946442e-05,
+      "loss": 0.502,
+      "step": 2509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936569255748065e-05,
+      "loss": 0.4624,
+      "step": 2510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793474798100665e-05,
+      "loss": 0.5115,
+      "step": 2511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932925995403335e-05,
+      "loss": 0.568,
+      "step": 2512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931103299101354e-05,
+      "loss": 0.5159,
+      "step": 2513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929279892264e-05,
+      "loss": 0.5144,
+      "step": 2514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927455775054626e-05,
+      "loss": 0.5617,
+      "step": 2515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925630947636655e-05,
+      "loss": 0.4619,
+      "step": 2516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792380541017357e-05,
+      "loss": 0.5759,
+      "step": 2517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7921979162828912e-05,
+      "loss": 0.5295,
+      "step": 2518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7920152205766302e-05,
+      "loss": 0.5728,
+      "step": 2519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918324539149407e-05,
+      "loss": 0.5006,
+      "step": 2520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7916496163141967e-05,
+      "loss": 0.4994,
+      "step": 2521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7914667077907786e-05,
+      "loss": 0.5376,
+      "step": 2522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912837283610724e-05,
+      "loss": 0.5198,
+      "step": 2523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7911006780414712e-05,
+      "loss": 0.5154,
+      "step": 2524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790917556848374e-05,
+      "loss": 0.4569,
+      "step": 2525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907343647981866e-05,
+      "loss": 0.5561,
+      "step": 2526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790551101907321e-05,
+      "loss": 0.5267,
+      "step": 2527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903677681921953e-05,
+      "loss": 0.5744,
+      "step": 2528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7901843636692338e-05,
+      "loss": 0.5111,
+      "step": 2529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900008883548676e-05,
+      "loss": 0.5807,
+      "step": 2530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898173422655343e-05,
+      "loss": 0.4789,
+      "step": 2531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789633725417677e-05,
+      "loss": 0.4597,
+      "step": 2532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894500378277463e-05,
+      "loss": 0.4628,
+      "step": 2533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892662795121976e-05,
+      "loss": 0.5008,
+      "step": 2534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789082450487494e-05,
+      "loss": 0.4798,
+      "step": 2535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888985507701052e-05,
+      "loss": 0.4663,
+      "step": 2536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887145803765054e-05,
+      "loss": 0.4977,
+      "step": 2537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7885305393231763e-05,
+      "loss": 0.4427,
+      "step": 2538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883464276266064e-05,
+      "loss": 0.5436,
+      "step": 2539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881622453032897e-05,
+      "loss": 0.4925,
+      "step": 2540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787977992369727e-05,
+      "loss": 0.5337,
+      "step": 2541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877936688424248e-05,
+      "loss": 0.5288,
+      "step": 2542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7876092747378967e-05,
+      "loss": 0.4677,
+      "step": 2543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874248100726624e-05,
+      "loss": 0.5177,
+      "step": 2544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872402748632477e-05,
+      "loss": 0.5261,
+      "step": 2545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870556691261848e-05,
+      "loss": 0.5402,
+      "step": 2546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786870992878012e-05,
+      "loss": 0.5705,
+      "step": 2547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866862461352744e-05,
+      "loss": 0.5391,
+      "step": 2548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786501428914523e-05,
+      "loss": 0.5154,
+      "step": 2549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863165412323156e-05,
+      "loss": 0.5431,
+      "step": 2550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861315831052162e-05,
+      "loss": 0.5398,
+      "step": 2551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785946554549794e-05,
+      "loss": 0.4628,
+      "step": 2552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857614555826264e-05,
+      "loss": 0.5625,
+      "step": 2553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855762862202954e-05,
+      "loss": 0.4719,
+      "step": 2554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853910464793906e-05,
+      "loss": 0.5289,
+      "step": 2555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785205736376507e-05,
+      "loss": 0.511,
+      "step": 2556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850203559282464e-05,
+      "loss": 0.5324,
+      "step": 2557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7848349051512167e-05,
+      "loss": 0.5076,
+      "step": 2558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846493840620322e-05,
+      "loss": 0.5289,
+      "step": 2559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784463792677313e-05,
+      "loss": 0.461,
+      "step": 2560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7842781310136868e-05,
+      "loss": 0.5085,
+      "step": 2561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840923990877865e-05,
+      "loss": 0.4738,
+      "step": 2562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783906596916251e-05,
+      "loss": 0.5307,
+      "step": 2563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783720724515726e-05,
+      "loss": 0.4518,
+      "step": 2564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7835347819028642e-05,
+      "loss": 0.5543,
+      "step": 2565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833487690943236e-05,
+      "loss": 0.5195,
+      "step": 2566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783162686106769e-05,
+      "loss": 0.5165,
+      "step": 2567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782976532956871e-05,
+      "loss": 0.5018,
+      "step": 2568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827903096613063e-05,
+      "loss": 0.5404,
+      "step": 2569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826040162367598e-05,
+      "loss": 0.5525,
+      "step": 2570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824176526999196e-05,
+      "loss": 0.4633,
+      "step": 2571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782231219067483e-05,
+      "loss": 0.5218,
+      "step": 2572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782044715356152e-05,
+      "loss": 0.5021,
+      "step": 2573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818581415826343e-05,
+      "loss": 0.541,
+      "step": 2574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781671497763646e-05,
+      "loss": 0.5325,
+      "step": 2575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781484783915907e-05,
+      "loss": 0.5172,
+      "step": 2576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781298000056146e-05,
+      "loss": 0.4914,
+      "step": 2577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811111462010956e-05,
+      "loss": 0.5307,
+      "step": 2578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7809242223674968e-05,
+      "loss": 0.5338,
+      "step": 2579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807372285720945e-05,
+      "loss": 0.5517,
+      "step": 2580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805501648316426e-05,
+      "loss": 0.5271,
+      "step": 2581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803630311628987e-05,
+      "loss": 0.5644,
+      "step": 2582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801758275826282e-05,
+      "loss": 0.4655,
+      "step": 2583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779988554107603e-05,
+      "loss": 0.5237,
+      "step": 2584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798012107545995e-05,
+      "loss": 0.5498,
+      "step": 2585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779613797540402e-05,
+      "loss": 0.5028,
+      "step": 2586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794263144818015e-05,
+      "loss": 0.5291,
+      "step": 2587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779238761595593e-05,
+      "loss": 0.4686,
+      "step": 2588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790511388985795e-05,
+      "loss": 0.5125,
+      "step": 2589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788634464075702e-05,
+      "loss": 0.5264,
+      "step": 2590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786756841393795e-05,
+      "loss": 0.496,
+      "step": 2591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784878521108292e-05,
+      "loss": 0.4889,
+      "step": 2592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778299950338747e-05,
+      "loss": 0.5907,
+      "step": 2593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778111978839966e-05,
+      "loss": 0.5314,
+      "step": 2594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779239376313273e-05,
+      "loss": 0.4847,
+      "step": 2595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777358267296766e-05,
+      "loss": 0.5387,
+      "step": 2596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775476461518668e-05,
+      "loss": 0.4452,
+      "step": 2597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773593959147563e-05,
+      "loss": 0.5232,
+      "step": 2598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771710760352104e-05,
+      "loss": 0.5038,
+      "step": 2599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769826865301003e-05,
+      "loss": 0.488,
+      "step": 2600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767942274163035e-05,
+      "loss": 0.527,
+      "step": 2601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776605698710704e-05,
+      "loss": 0.5653,
+      "step": 2602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764171004301913e-05,
+      "loss": 0.445,
+      "step": 2603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7762284325916622e-05,
+      "loss": 0.471,
+      "step": 2604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776039695212019e-05,
+      "loss": 0.5218,
+      "step": 2605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7758508883081702e-05,
+      "loss": 0.4933,
+      "step": 2606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7756620118970307e-05,
+      "loss": 0.4787,
+      "step": 2607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775473065995522e-05,
+      "loss": 0.4979,
+      "step": 2608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752840506205714e-05,
+      "loss": 0.4955,
+      "step": 2609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775094965789112e-05,
+      "loss": 0.5619,
+      "step": 2610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749058115180842e-05,
+      "loss": 0.4845,
+      "step": 2611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747165878244335e-05,
+      "loss": 0.5226,
+      "step": 2612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774527294725113e-05,
+      "loss": 0.4929,
+      "step": 2613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743379322370802e-05,
+      "loss": 0.4771,
+      "step": 2614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741485003773005e-05,
+      "loss": 0.4946,
+      "step": 2615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739589991627443e-05,
+      "loss": 0.5191,
+      "step": 2616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737694286103892e-05,
+      "loss": 0.501,
+      "step": 2617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735797887372182e-05,
+      "loss": 0.517,
+      "step": 2618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773390079560221e-05,
+      "loss": 0.6053,
+      "step": 2619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732003010963934e-05,
+      "loss": 0.4764,
+      "step": 2620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.5244,
+      "step": 2621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728205363762607e-05,
+      "loss": 0.4855,
+      "step": 2622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772630550153978e-05,
+      "loss": 0.5252,
+      "step": 2623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724404947129097e-05,
+      "loss": 0.4253,
+      "step": 2624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722503700700836e-05,
+      "loss": 0.5055,
+      "step": 2625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772060176242531e-05,
+      "loss": 0.512,
+      "step": 2626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771869913247292e-05,
+      "loss": 0.5135,
+      "step": 2627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716795811014127e-05,
+      "loss": 0.508,
+      "step": 2628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7714891798219432e-05,
+      "loss": 0.5238,
+      "step": 2629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7712987094259418e-05,
+      "loss": 0.4987,
+      "step": 2630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711081699304728e-05,
+      "loss": 0.4936,
+      "step": 2631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770917561352606e-05,
+      "loss": 0.5018,
+      "step": 2632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770726883709418e-05,
+      "loss": 0.5549,
+      "step": 2633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705361370179908e-05,
+      "loss": 0.5242,
+      "step": 2634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703453212954138e-05,
+      "loss": 0.5736,
+      "step": 2635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701544365587813e-05,
+      "loss": 0.4995,
+      "step": 2636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699634828251945e-05,
+      "loss": 0.5291,
+      "step": 2637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769772460111761e-05,
+      "loss": 0.5177,
+      "step": 2638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769581368435594e-05,
+      "loss": 0.515,
+      "step": 2639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769390207813813e-05,
+      "loss": 0.477,
+      "step": 2640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769198978263544e-05,
+      "loss": 0.517,
+      "step": 2641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690076798019185e-05,
+      "loss": 0.5049,
+      "step": 2642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688163124460755e-05,
+      "loss": 0.5198,
+      "step": 2643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686248762131584e-05,
+      "loss": 0.561,
+      "step": 2644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768433371120318e-05,
+      "loss": 0.4747,
+      "step": 2645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7682417971847113e-05,
+      "loss": 0.5278,
+      "step": 2646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768050154423501e-05,
+      "loss": 0.5227,
+      "step": 2647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678584428538556e-05,
+      "loss": 0.4857,
+      "step": 2648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7676666624929507e-05,
+      "loss": 0.4788,
+      "step": 2649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674748133579674e-05,
+      "loss": 0.4753,
+      "step": 2650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767282895466093e-05,
+      "loss": 0.497,
+      "step": 2651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670909088345216e-05,
+      "loss": 0.5025,
+      "step": 2652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668988534804524e-05,
+      "loss": 0.5306,
+      "step": 2653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7667067294210923e-05,
+      "loss": 0.5192,
+      "step": 2654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665145366736526e-05,
+      "loss": 0.4961,
+      "step": 2655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663222752553517e-05,
+      "loss": 0.48,
+      "step": 2656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766129945183414e-05,
+      "loss": 0.499,
+      "step": 2657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659375464750702e-05,
+      "loss": 0.5318,
+      "step": 2658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657450791475566e-05,
+      "loss": 0.4436,
+      "step": 2659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765552543218117e-05,
+      "loss": 0.5832,
+      "step": 2660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653599387039993e-05,
+      "loss": 0.5537,
+      "step": 2661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651672656224592e-05,
+      "loss": 0.5343,
+      "step": 2662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649745239907576e-05,
+      "loss": 0.5112,
+      "step": 2663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764781713826163e-05,
+      "loss": 0.5423,
+      "step": 2664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645888351459476e-05,
+      "loss": 0.5157,
+      "step": 2665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764395887967392e-05,
+      "loss": 0.5781,
+      "step": 2666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642028723077812e-05,
+      "loss": 0.5001,
+      "step": 2667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764009788184408e-05,
+      "loss": 0.5321,
+      "step": 2668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638166356145703e-05,
+      "loss": 0.4587,
+      "step": 2669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636234146155723e-05,
+      "loss": 0.4827,
+      "step": 2670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763430125204724e-05,
+      "loss": 0.5731,
+      "step": 2671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763236767399343e-05,
+      "loss": 0.5456,
+      "step": 2672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630433412167506e-05,
+      "loss": 0.4828,
+      "step": 2673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628498466742764e-05,
+      "loss": 0.4962,
+      "step": 2674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626562837892545e-05,
+      "loss": 0.5615,
+      "step": 2675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624626525790266e-05,
+      "loss": 0.46,
+      "step": 2676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622689530609397e-05,
+      "loss": 0.5656,
+      "step": 2677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762075185252347e-05,
+      "loss": 0.5195,
+      "step": 2678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618813491706082e-05,
+      "loss": 0.5376,
+      "step": 2679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761687444833088e-05,
+      "loss": 0.4803,
+      "step": 2680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614934722571586e-05,
+      "loss": 0.524,
+      "step": 2681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761299431460197e-05,
+      "loss": 0.5218,
+      "step": 2682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761105322459588e-05,
+      "loss": 0.5306,
+      "step": 2683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609111452727212e-05,
+      "loss": 0.5164,
+      "step": 2684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607168999169923e-05,
+      "loss": 0.5151,
+      "step": 2685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605225864098036e-05,
+      "loss": 0.4676,
+      "step": 2686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760328204768563e-05,
+      "loss": 0.4931,
+      "step": 2687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7601337550106857e-05,
+      "loss": 0.5045,
+      "step": 2688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7599392371535916e-05,
+      "loss": 0.4686,
+      "step": 2689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597446512147075e-05,
+      "loss": 0.4552,
+      "step": 2690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7595499972114657e-05,
+      "loss": 0.5118,
+      "step": 2691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593552751613052e-05,
+      "loss": 0.5051,
+      "step": 2692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7591604850816705e-05,
+      "loss": 0.4561,
+      "step": 2693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589656269900132e-05,
+      "loss": 0.4824,
+      "step": 2694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75877070090379e-05,
+      "loss": 0.5029,
+      "step": 2695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585757068404642e-05,
+      "loss": 0.5084,
+      "step": 2696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758380644817504e-05,
+      "loss": 0.4872,
+      "step": 2697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758185514852386e-05,
+      "loss": 0.5202,
+      "step": 2698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579903169625916e-05,
+      "loss": 0.4856,
+      "step": 2699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577950511656076e-05,
+      "loss": 0.4764,
+      "step": 2700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575997174789275e-05,
+      "loss": 0.5613,
+      "step": 2701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574043159200512e-05,
+      "loss": 0.5393,
+      "step": 2702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572088465064847e-05,
+      "loss": 0.5781,
+      "step": 2703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570133092557394e-05,
+      "loss": 0.525,
+      "step": 2704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568177041853337e-05,
+      "loss": 0.4528,
+      "step": 2705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566220313127907e-05,
+      "loss": 0.5464,
+      "step": 2706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564262906556412e-05,
+      "loss": 0.4737,
+      "step": 2707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562304822314205e-05,
+      "loss": 0.5075,
+      "step": 2708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560346060576718e-05,
+      "loss": 0.5181,
+      "step": 2709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755838662151943e-05,
+      "loss": 0.3999,
+      "step": 2710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556426505317882e-05,
+      "loss": 0.5165,
+      "step": 2711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554465712147677e-05,
+      "loss": 0.5057,
+      "step": 2712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755250424218448e-05,
+      "loss": 0.5662,
+      "step": 2713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550542095604018e-05,
+      "loss": 0.5862,
+      "step": 2714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548579272582076e-05,
+      "loss": 0.5339,
+      "step": 2715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546615773294497e-05,
+      "loss": 0.521,
+      "step": 2716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544651597917194e-05,
+      "loss": 0.4666,
+      "step": 2717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754268674662613e-05,
+      "loss": 0.5069,
+      "step": 2718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7540721219597337e-05,
+      "loss": 0.5381,
+      "step": 2719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75387550170069e-05,
+      "loss": 0.5372,
+      "step": 2720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753678813903097e-05,
+      "loss": 0.5342,
+      "step": 2721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534820585845752e-05,
+      "loss": 0.5394,
+      "step": 2722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753285235762752e-05,
+      "loss": 0.4965,
+      "step": 2723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530883454552606e-05,
+      "loss": 0.5249,
+      "step": 2724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.75289138767974e-05,
+      "loss": 0.5083,
+      "step": 2725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526943624538352e-05,
+      "loss": 0.4687,
+      "step": 2726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524972697951973e-05,
+      "loss": 0.4942,
+      "step": 2727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752300109721484e-05,
+      "loss": 0.5283,
+      "step": 2728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521028822503578e-05,
+      "loss": 0.5283,
+      "step": 2729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519055873994887e-05,
+      "loss": 0.5192,
+      "step": 2730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517082251865517e-05,
+      "loss": 0.5317,
+      "step": 2731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515107956292286e-05,
+      "loss": 0.5106,
+      "step": 2732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751313298745206e-05,
+      "loss": 0.4949,
+      "step": 2733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511157345521783e-05,
+      "loss": 0.5196,
+      "step": 2734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7509181030678442e-05,
+      "loss": 0.5386,
+      "step": 2735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7507204043099095e-05,
+      "loss": 0.4636,
+      "step": 2736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750522638296086e-05,
+      "loss": 0.517,
+      "step": 2737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503248050440907e-05,
+      "loss": 0.4959,
+      "step": 2738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501269045716472e-05,
+      "loss": 0.504,
+      "step": 2739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499289368964862e-05,
+      "loss": 0.5159,
+      "step": 2740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497309020363418e-05,
+      "loss": 0.5277,
+      "step": 2741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495328000089564e-05,
+      "loss": 0.5161,
+      "step": 2742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749334630832078e-05,
+      "loss": 0.5205,
+      "step": 2743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491363945234595e-05,
+      "loss": 0.5433,
+      "step": 2744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748938091100861e-05,
+      "loss": 0.5316,
+      "step": 2745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748739720582048e-05,
+      "loss": 0.507,
+      "step": 2746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485412829847926e-05,
+      "loss": 0.4854,
+      "step": 2747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483427783268724e-05,
+      "loss": 0.5206,
+      "step": 2748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481442066260704e-05,
+      "loss": 0.4798,
+      "step": 2749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7479455679001774e-05,
+      "loss": 0.515,
+      "step": 2750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477468621669887e-05,
+      "loss": 0.4895,
+      "step": 2751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475480894443056e-05,
+      "loss": 0.4839,
+      "step": 2752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473492497499366e-05,
+      "loss": 0.4845,
+      "step": 2753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471503431016952e-05,
+      "loss": 0.524,
+      "step": 2754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746951369517401e-05,
+      "loss": 0.5021,
+      "step": 2755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467523290148793e-05,
+      "loss": 0.4977,
+      "step": 2756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7465532216119628e-05,
+      "loss": 0.4821,
+      "step": 2757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7463540473264886e-05,
+      "loss": 0.5687,
+      "step": 2758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461548061763007e-05,
+      "loss": 0.4717,
+      "step": 2759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459554981792483e-05,
+      "loss": 0.5327,
+      "step": 2760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457561233531876e-05,
+      "loss": 0.482,
+      "step": 2761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455566817159803e-05,
+      "loss": 0.5162,
+      "step": 2762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453571732854938e-05,
+      "loss": 0.5005,
+      "step": 2763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745157598079602e-05,
+      "loss": 0.5355,
+      "step": 2764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7449579561161844e-05,
+      "loss": 0.5132,
+      "step": 2765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447582474131262e-05,
+      "loss": 0.4689,
+      "step": 2766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.74455847198832e-05,
+      "loss": 0.5001,
+      "step": 2767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443586298596626e-05,
+      "loss": 0.5051,
+      "step": 2768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441587210450577e-05,
+      "loss": 0.475,
+      "step": 2769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439587455624148e-05,
+      "loss": 0.5301,
+      "step": 2770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437587034296494e-05,
+      "loss": 0.4824,
+      "step": 2771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435585946646833e-05,
+      "loss": 0.5471,
+      "step": 2772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7433584192854435e-05,
+      "loss": 0.5305,
+      "step": 2773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431581773098635e-05,
+      "loss": 0.527,
+      "step": 2774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742957868755883e-05,
+      "loss": 0.5149,
+      "step": 2775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427574936414467e-05,
+      "loss": 0.5153,
+      "step": 2776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7425570519845068e-05,
+      "loss": 0.4969,
+      "step": 2777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7423565438030198e-05,
+      "loss": 0.514,
+      "step": 2778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421559691149487e-05,
+      "loss": 0.4952,
+      "step": 2779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419553279382637e-05,
+      "loss": 0.4658,
+      "step": 2780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7417546202909395e-05,
+      "loss": 0.519,
+      "step": 2781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7415538461909567e-05,
+      "loss": 0.515,
+      "step": 2782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7413530056563028e-05,
+      "loss": 0.5287,
+      "step": 2783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411520987049705e-05,
+      "loss": 0.5083,
+      "step": 2784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409511253549592e-05,
+      "loss": 0.5265,
+      "step": 2785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7407500856242735e-05,
+      "loss": 0.569,
+      "step": 2786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7405489795309244e-05,
+      "loss": 0.5522,
+      "step": 2787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403478070929286e-05,
+      "loss": 0.5244,
+      "step": 2788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401465683283085e-05,
+      "loss": 0.5076,
+      "step": 2789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739945263255093e-05,
+      "loss": 0.5066,
+      "step": 2790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739743891891317e-05,
+      "loss": 0.512,
+      "step": 2791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739542454255021e-05,
+      "loss": 0.5224,
+      "step": 2792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393409503642514e-05,
+      "loss": 0.5074,
+      "step": 2793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391393802370603e-05,
+      "loss": 0.5467,
+      "step": 2794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389377438915062e-05,
+      "loss": 0.4741,
+      "step": 2795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387360413456534e-05,
+      "loss": 0.5036,
+      "step": 2796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385342726175728e-05,
+      "loss": 0.5378,
+      "step": 2797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383324377253393e-05,
+      "loss": 0.5323,
+      "step": 2798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381305366870362e-05,
+      "loss": 0.5216,
+      "step": 2799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737928569520751e-05,
+      "loss": 0.5394,
+      "step": 2800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377265362445772e-05,
+      "loss": 0.4783,
+      "step": 2801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737524436876615e-05,
+      "loss": 0.534,
+      "step": 2802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373222714349703e-05,
+      "loss": 0.529,
+      "step": 2803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737120039937755e-05,
+      "loss": 0.5097,
+      "step": 2804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736917742403086e-05,
+      "loss": 0.506,
+      "step": 2805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367153788490875e-05,
+      "loss": 0.5565,
+      "step": 2806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365129492938884e-05,
+      "loss": 0.4817,
+      "step": 2807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363104537556247e-05,
+      "loss": 0.5171,
+      "step": 2808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736107892252437e-05,
+      "loss": 0.5054,
+      "step": 2809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359052648024728e-05,
+      "loss": 0.5214,
+      "step": 2810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7357025714238854e-05,
+      "loss": 0.4871,
+      "step": 2811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354998121348332e-05,
+      "loss": 0.5125,
+      "step": 2812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352969869534818e-05,
+      "loss": 0.4927,
+      "step": 2813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350940958980013e-05,
+      "loss": 0.5589,
+      "step": 2814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348911389865687e-05,
+      "loss": 0.5066,
+      "step": 2815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7346881162373668e-05,
+      "loss": 0.5014,
+      "step": 2816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344850276685837e-05,
+      "loss": 0.5543,
+      "step": 2817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342818732984144e-05,
+      "loss": 0.5379,
+      "step": 2818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340786531450586e-05,
+      "loss": 0.4858,
+      "step": 2819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338753672267227e-05,
+      "loss": 0.5004,
+      "step": 2820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336720155616186e-05,
+      "loss": 0.5178,
+      "step": 2821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7334685981679647e-05,
+      "loss": 0.5254,
+      "step": 2822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332651150639843e-05,
+      "loss": 0.5283,
+      "step": 2823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330615662679076e-05,
+      "loss": 0.4872,
+      "step": 2824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73285795179797e-05,
+      "loss": 0.4729,
+      "step": 2825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326542716724127e-05,
+      "loss": 0.4926,
+      "step": 2826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324505259094843e-05,
+      "loss": 0.5223,
+      "step": 2827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322467145274365e-05,
+      "loss": 0.5095,
+      "step": 2828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732042837544529e-05,
+      "loss": 0.5245,
+      "step": 2829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318388949790275e-05,
+      "loss": 0.4778,
+      "step": 2830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316348868492022e-05,
+      "loss": 0.5123,
+      "step": 2831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7314308131733298e-05,
+      "loss": 0.5091,
+      "step": 2832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312266739696936e-05,
+      "loss": 0.5103,
+      "step": 2833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310224692565815e-05,
+      "loss": 0.4475,
+      "step": 2834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7308181990522883e-05,
+      "loss": 0.434,
+      "step": 2835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306138633751136e-05,
+      "loss": 0.4579,
+      "step": 2836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304094622433646e-05,
+      "loss": 0.518,
+      "step": 2837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302049956753525e-05,
+      "loss": 0.4799,
+      "step": 2838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730000463689395e-05,
+      "loss": 0.4504,
+      "step": 2839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729795866303816e-05,
+      "loss": 0.5069,
+      "step": 2840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295912035369458e-05,
+      "loss": 0.539,
+      "step": 2841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293864754071192e-05,
+      "loss": 0.5399,
+      "step": 2842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291816819326767e-05,
+      "loss": 0.4544,
+      "step": 2843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728976823131967e-05,
+      "loss": 0.5498,
+      "step": 2844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728771899023342e-05,
+      "loss": 0.5105,
+      "step": 2845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285669096251608e-05,
+      "loss": 0.5241,
+      "step": 2846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728361854955788e-05,
+      "loss": 0.5426,
+      "step": 2847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281567350335944e-05,
+      "loss": 0.5686,
+      "step": 2848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279515498769565e-05,
+      "loss": 0.5111,
+      "step": 2849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727746299504256e-05,
+      "loss": 0.5138,
+      "step": 2850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275409839338814e-05,
+      "loss": 0.5298,
+      "step": 2851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273356031842265e-05,
+      "loss": 0.5261,
+      "step": 2852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271301572736904e-05,
+      "loss": 0.455,
+      "step": 2853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269246462206797e-05,
+      "loss": 0.4618,
+      "step": 2854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267190700436053e-05,
+      "loss": 0.5126,
+      "step": 2855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265134287608846e-05,
+      "loss": 0.4932,
+      "step": 2856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263077223909407e-05,
+      "loss": 0.5035,
+      "step": 2857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261019509522022e-05,
+      "loss": 0.4849,
+      "step": 2858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7258961144631047e-05,
+      "loss": 0.4769,
+      "step": 2859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256902129420874e-05,
+      "loss": 0.5223,
+      "step": 2860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254842464075977e-05,
+      "loss": 0.4775,
+      "step": 2861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725278214878088e-05,
+      "loss": 0.4877,
+      "step": 2862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250721183720153e-05,
+      "loss": 0.5052,
+      "step": 2863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7248659569078446e-05,
+      "loss": 0.461,
+      "step": 2864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246597305040448e-05,
+      "loss": 0.523,
+      "step": 2865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724453439179092e-05,
+      "loss": 0.5054,
+      "step": 2866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242470829514674e-05,
+      "loss": 0.4942,
+      "step": 2867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240406618396574e-05,
+      "loss": 0.5116,
+      "step": 2868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723834175862156e-05,
+      "loss": 0.4886,
+      "step": 2869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236276250374612e-05,
+      "loss": 0.5486,
+      "step": 2870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234210093840777e-05,
+      "loss": 0.5576,
+      "step": 2871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232143289205164e-05,
+      "loss": 0.5436,
+      "step": 2872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723007583665293e-05,
+      "loss": 0.5065,
+      "step": 2873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7228007736369293e-05,
+      "loss": 0.5346,
+      "step": 2874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722593898853954e-05,
+      "loss": 0.5202,
+      "step": 2875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223869593348997e-05,
+      "loss": 0.4942,
+      "step": 2876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7221799550983062e-05,
+      "loss": 0.5702,
+      "step": 2877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219728861627186e-05,
+      "loss": 0.5357,
+      "step": 2878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217657525466882e-05,
+      "loss": 0.5304,
+      "step": 2879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215585542687715e-05,
+      "loss": 0.5286,
+      "step": 2880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213512913475313e-05,
+      "loss": 0.4875,
+      "step": 2881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211439638015356e-05,
+      "loss": 0.4702,
+      "step": 2882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209365716493585e-05,
+      "loss": 0.5245,
+      "step": 2883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207291149095806e-05,
+      "loss": 0.4901,
+      "step": 2884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720521593600787e-05,
+      "loss": 0.465,
+      "step": 2885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.72031400774157e-05,
+      "loss": 0.492,
+      "step": 2886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201063573505255e-05,
+      "loss": 0.5546,
+      "step": 2887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719898642446258e-05,
+      "loss": 0.4367,
+      "step": 2888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196908630473755e-05,
+      "loss": 0.5404,
+      "step": 2889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719483019172493e-05,
+      "loss": 0.5088,
+      "step": 2890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719275110840231e-05,
+      "loss": 0.5165,
+      "step": 2891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7190671380692153e-05,
+      "loss": 0.5037,
+      "step": 2892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188591008780783e-05,
+      "loss": 0.4769,
+      "step": 2893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186509992854572e-05,
+      "loss": 0.5223,
+      "step": 2894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718442833309996e-05,
+      "loss": 0.4729,
+      "step": 2895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718234602970344e-05,
+      "loss": 0.4949,
+      "step": 2896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180263082851556e-05,
+      "loss": 0.504,
+      "step": 2897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717817949273092e-05,
+      "loss": 0.5311,
+      "step": 2898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176095259528204e-05,
+      "loss": 0.5463,
+      "step": 2899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717401038343012e-05,
+      "loss": 0.4684,
+      "step": 2900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171924864623456e-05,
+      "loss": 0.4955,
+      "step": 2901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169838703295048e-05,
+      "loss": 0.5266,
+      "step": 2902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7167751899631794e-05,
+      "loss": 0.472,
+      "step": 2903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716566445382064e-05,
+      "loss": 0.4976,
+      "step": 2904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716357636604861e-05,
+      "loss": 0.4858,
+      "step": 2905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161487636502762e-05,
+      "loss": 0.4767,
+      "step": 2906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715939826537023e-05,
+      "loss": 0.5481,
+      "step": 2907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157308252838187e-05,
+      "loss": 0.5267,
+      "step": 2908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155217599093886e-05,
+      "loss": 0.5066,
+      "step": 2909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153126304324614e-05,
+      "loss": 0.4821,
+      "step": 2910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151034368717737e-05,
+      "loss": 0.5107,
+      "step": 2911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148941792460664e-05,
+      "loss": 0.5448,
+      "step": 2912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146848575740862e-05,
+      "loss": 0.4763,
+      "step": 2913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144754718745863e-05,
+      "loss": 0.4793,
+      "step": 2914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142660221663255e-05,
+      "loss": 0.4817,
+      "step": 2915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714056508468068e-05,
+      "loss": 0.4955,
+      "step": 2916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138469307985832e-05,
+      "loss": 0.518,
+      "step": 2917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7136372891766476e-05,
+      "loss": 0.4995,
+      "step": 2918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713427583621042e-05,
+      "loss": 0.5433,
+      "step": 2919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713217814150554e-05,
+      "loss": 0.4748,
+      "step": 2920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713007980783976e-05,
+      "loss": 0.5346,
+      "step": 2921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127980835401075e-05,
+      "loss": 0.4832,
+      "step": 2922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125881224377526e-05,
+      "loss": 0.5211,
+      "step": 2923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712378097495721e-05,
+      "loss": 0.4757,
+      "step": 2924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121680087328288e-05,
+      "loss": 0.4888,
+      "step": 2925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711957856167898e-05,
+      "loss": 0.5038,
+      "step": 2926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711747639819755e-05,
+      "loss": 0.5704,
+      "step": 2927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711537359707233e-05,
+      "loss": 0.5111,
+      "step": 2928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113270158491705e-05,
+      "loss": 0.4718,
+      "step": 2929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111166082644127e-05,
+      "loss": 0.4819,
+      "step": 2930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7109061369718086e-05,
+      "loss": 0.5227,
+      "step": 2931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710695601990215e-05,
+      "loss": 0.5478,
+      "step": 2932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710485003338493e-05,
+      "loss": 0.491,
+      "step": 2933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7102743410355095e-05,
+      "loss": 0.5189,
+      "step": 2934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100636151001375e-05,
+      "loss": 0.444,
+      "step": 2935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709852825551256e-05,
+      "loss": 0.4662,
+      "step": 2936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7096419724077492e-05,
+      "loss": 0.4661,
+      "step": 2937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709431055688507e-05,
+      "loss": 0.5053,
+      "step": 2938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709220075412425e-05,
+      "loss": 0.581,
+      "step": 2939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709009031598405e-05,
+      "loss": 0.5029,
+      "step": 2940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087979242653535e-05,
+      "loss": 0.4639,
+      "step": 2941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708586753432184e-05,
+      "loss": 0.493,
+      "step": 2942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708375519117814e-05,
+      "loss": 0.5596,
+      "step": 2943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081642213411687e-05,
+      "loss": 0.4257,
+      "step": 2944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707952860121177e-05,
+      "loss": 0.4841,
+      "step": 2945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707741435476775e-05,
+      "loss": 0.4935,
+      "step": 2946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7075299474269044e-05,
+      "loss": 0.554,
+      "step": 2947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707318395990511e-05,
+      "loss": 0.4658,
+      "step": 2948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.4626,
+      "step": 2949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068951030339732e-05,
+      "loss": 0.4721,
+      "step": 2950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706683361551751e-05,
+      "loss": 0.4947,
+      "step": 2951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064715567588508e-05,
+      "loss": 0.5151,
+      "step": 2952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706259688674248e-05,
+      "loss": 0.4687,
+      "step": 2953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060477573169232e-05,
+      "loss": 0.4774,
+      "step": 2954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058357627058635e-05,
+      "loss": 0.5609,
+      "step": 2955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056237048600608e-05,
+      "loss": 0.4701,
+      "step": 2956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705411583798513e-05,
+      "loss": 0.4533,
+      "step": 2957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705199399540224e-05,
+      "loss": 0.5212,
+      "step": 2958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7049871521042026e-05,
+      "loss": 0.5199,
+      "step": 2959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047748415094646e-05,
+      "loss": 0.4733,
+      "step": 2960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045624677750296e-05,
+      "loss": 0.5012,
+      "step": 2961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043500309199244e-05,
+      "loss": 0.4826,
+      "step": 2962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041375309631804e-05,
+      "loss": 0.466,
+      "step": 2963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039249679238358e-05,
+      "loss": 0.5022,
+      "step": 2964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037123418209334e-05,
+      "loss": 0.4661,
+      "step": 2965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034996526735222e-05,
+      "loss": 0.4899,
+      "step": 2966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032869005006564e-05,
+      "loss": 0.458,
+      "step": 2967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030740853213966e-05,
+      "loss": 0.4874,
+      "step": 2968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702861207154808e-05,
+      "loss": 0.5278,
+      "step": 2969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702648266019963e-05,
+      "loss": 0.5381,
+      "step": 2970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024352619359376e-05,
+      "loss": 0.4866,
+      "step": 2971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022221949218147e-05,
+      "loss": 0.5201,
+      "step": 2972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7020090649966835e-05,
+      "loss": 0.5096,
+      "step": 2973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701795872179637e-05,
+      "loss": 0.4579,
+      "step": 2974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701582616489775e-05,
+      "loss": 0.524,
+      "step": 2975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7013692979462033e-05,
+      "loss": 0.5055,
+      "step": 2976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701155916568032e-05,
+      "loss": 0.5304,
+      "step": 2977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7009424723743787e-05,
+      "loss": 0.5512,
+      "step": 2978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7007289653843643e-05,
+      "loss": 0.501,
+      "step": 2979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005153956171177e-05,
+      "loss": 0.5459,
+      "step": 2980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700301763091771e-05,
+      "loss": 0.5364,
+      "step": 2981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000880678274643e-05,
+      "loss": 0.494,
+      "step": 2982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6998743098433422e-05,
+      "loss": 0.5077,
+      "step": 2983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699660489158554e-05,
+      "loss": 0.5388,
+      "step": 2984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994466057922564e-05,
+      "loss": 0.4991,
+      "step": 2985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992326597636108e-05,
+      "loss": 0.4899,
+      "step": 2986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6990186510917843e-05,
+      "loss": 0.5077,
+      "step": 2987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698804579795949e-05,
+      "loss": 0.4774,
+      "step": 2988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985904458952843e-05,
+      "loss": 0.5269,
+      "step": 2989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983762494089732e-05,
+      "loss": 0.4928,
+      "step": 2990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698161990356205e-05,
+      "loss": 0.5361,
+      "step": 2991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979476687561763e-05,
+      "loss": 0.4756,
+      "step": 2992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977332846280865e-05,
+      "loss": 0.5594,
+      "step": 2993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6975188379911424e-05,
+      "loss": 0.4962,
+      "step": 2994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973043288645558e-05,
+      "loss": 0.5076,
+      "step": 2995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6970897572675447e-05,
+      "loss": 0.5223,
+      "step": 2996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968751232193315e-05,
+      "loss": 0.4709,
+      "step": 2997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966604267391458e-05,
+      "loss": 0.4741,
+      "step": 2998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696445667846221e-05,
+      "loss": 0.6031,
+      "step": 2999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696230846559798e-05,
+      "loss": 0.4792,
+      "step": 3000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6960159628991215e-05,
+      "loss": 0.5198,
+      "step": 3001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958010168834425e-05,
+      "loss": 0.4772,
+      "step": 3002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955860085320187e-05,
+      "loss": 0.5548,
+      "step": 3003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953709378641117e-05,
+      "loss": 0.5022,
+      "step": 3004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951558048989893e-05,
+      "loss": 0.5403,
+      "step": 3005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694940609655925e-05,
+      "loss": 0.4763,
+      "step": 3006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947253521541975e-05,
+      "loss": 0.5273,
+      "step": 3007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6945100324130925e-05,
+      "loss": 0.5363,
+      "step": 3008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6942946504518986e-05,
+      "loss": 0.5095,
+      "step": 3009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694079206289913e-05,
+      "loss": 0.5043,
+      "step": 3010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693863699946436e-05,
+      "loss": 0.4519,
+      "step": 3011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6936481314407748e-05,
+      "loss": 0.5022,
+      "step": 3012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934325007922418e-05,
+      "loss": 0.5111,
+      "step": 3013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932168080201555e-05,
+      "loss": 0.4879,
+      "step": 3014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930010531438384e-05,
+      "loss": 0.4982,
+      "step": 3015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692785236182621e-05,
+      "loss": 0.5532,
+      "step": 3016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6925693571558372e-05,
+      "loss": 0.51,
+      "step": 3017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692353416082827e-05,
+      "loss": 0.4817,
+      "step": 3018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921374129829374e-05,
+      "loss": 0.5622,
+      "step": 3019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6919213478755186e-05,
+      "loss": 0.4805,
+      "step": 3020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691705220779928e-05,
+      "loss": 0.472,
+      "step": 3021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691489031715528e-05,
+      "loss": 0.4818,
+      "step": 3022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912727807016867e-05,
+      "loss": 0.514,
+      "step": 3023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691056467757778e-05,
+      "loss": 0.5437,
+      "step": 3024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908400929031804e-05,
+      "loss": 0.5346,
+      "step": 3025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6906236561572794e-05,
+      "loss": 0.4745,
+      "step": 3026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690407157539464e-05,
+      "loss": 0.5401,
+      "step": 3027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901905970691316e-05,
+      "loss": 0.5195,
+      "step": 3028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899739747656823e-05,
+      "loss": 0.5226,
+      "step": 3029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897572906485233e-05,
+      "loss": 0.4846,
+      "step": 3030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689540544737067e-05,
+      "loss": 0.4833,
+      "step": 3031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893237370507316e-05,
+      "loss": 0.5078,
+      "step": 3032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68910686760894e-05,
+      "loss": 0.5479,
+      "step": 3033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688889936431122e-05,
+      "loss": 0.5674,
+      "step": 3034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886729435367117e-05,
+      "loss": 0.4737,
+      "step": 3035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884558889451486e-05,
+      "loss": 0.4539,
+      "step": 3036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882387726758793e-05,
+      "loss": 0.535,
+      "step": 3037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688021594748354e-05,
+      "loss": 0.5083,
+      "step": 3038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68780435518203e-05,
+      "loss": 0.4846,
+      "step": 3039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875870539963694e-05,
+      "loss": 0.5418,
+      "step": 3040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687369691210839e-05,
+      "loss": 0.5099,
+      "step": 3041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871522668449135e-05,
+      "loss": 0.4888,
+      "step": 3042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6869347809180708e-05,
+      "loss": 0.4959,
+      "step": 3043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686717233449795e-05,
+      "loss": 0.4614,
+      "step": 3044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864996244595757e-05,
+      "loss": 0.4792,
+      "step": 3045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862819539669088e-05,
+      "loss": 0.5211,
+      "step": 3046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860642219912946e-05,
+      "loss": 0.5032,
+      "step": 3047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6858464285522392e-05,
+      "loss": 0.5079,
+      "step": 3048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856285736692553e-05,
+      "loss": 0.5237,
+      "step": 3049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685410657361859e-05,
+      "loss": 0.515,
+      "step": 3050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685192679649574e-05,
+      "loss": 0.5073,
+      "step": 3051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684974640551928e-05,
+      "loss": 0.533,
+      "step": 3052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847565400884555e-05,
+      "loss": 0.515,
+      "step": 3053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845383782786948e-05,
+      "loss": 0.47,
+      "step": 3054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843201551421912e-05,
+      "loss": 0.5578,
+      "step": 3055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841018706984953e-05,
+      "loss": 0.5345,
+      "step": 3056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838835249671622e-05,
+      "loss": 0.5082,
+      "step": 3057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683665117967754e-05,
+      "loss": 0.5486,
+      "step": 3058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6834466497198362e-05,
+      "loss": 0.5099,
+      "step": 3059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832281202429825e-05,
+      "loss": 0.5101,
+      "step": 3060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6830095295567695e-05,
+      "loss": 0.5167,
+      "step": 3061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827908776807808e-05,
+      "loss": 0.523,
+      "step": 3062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825721646346052e-05,
+      "loss": 0.5179,
+      "step": 3063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823533904378365e-05,
+      "loss": 0.481,
+      "step": 3064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682134555110075e-05,
+      "loss": 0.4907,
+      "step": 3065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819156586709253e-05,
+      "loss": 0.5409,
+      "step": 3066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6816967011399977e-05,
+      "loss": 0.5211,
+      "step": 3067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681477682536909e-05,
+      "loss": 0.4907,
+      "step": 3068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681258602881281e-05,
+      "loss": 0.489,
+      "step": 3069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810394621927393e-05,
+      "loss": 0.541,
+      "step": 3070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808202604909177e-05,
+      "loss": 0.483,
+      "step": 3071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806009977954533e-05,
+      "loss": 0.4476,
+      "step": 3072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803816741259898e-05,
+      "loss": 0.4885,
+      "step": 3073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801622895021764e-05,
+      "loss": 0.5757,
+      "step": 3074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799428439436672e-05,
+      "loss": 0.4624,
+      "step": 3075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679723337470122e-05,
+      "loss": 0.487,
+      "step": 3076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679503770101206e-05,
+      "loss": 0.5136,
+      "step": 3077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6792841418565896e-05,
+      "loss": 0.443,
+      "step": 3078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790644527559498e-05,
+      "loss": 0.5025,
+      "step": 3079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788447028189672e-05,
+      "loss": 0.5782,
+      "step": 3080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786248920653294e-05,
+      "loss": 0.518,
+      "step": 3081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6784050205147288e-05,
+      "loss": 0.4917,
+      "step": 3082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781850881868637e-05,
+      "loss": 0.4715,
+      "step": 3083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677965095101437e-05,
+      "loss": 0.5212,
+      "step": 3084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777450412781578e-05,
+      "loss": 0.4964,
+      "step": 3085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6775249267367402e-05,
+      "loss": 0.4979,
+      "step": 3086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773047514969044e-05,
+      "loss": 0.6099,
+      "step": 3087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677084515578375e-05,
+      "loss": 0.5265,
+      "step": 3088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768642190008825e-05,
+      "loss": 0.4383,
+      "step": 3089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6766438617841637e-05,
+      "loss": 0.4247,
+      "step": 3090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764234439479596e-05,
+      "loss": 0.4523,
+      "step": 3091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676202965512017e-05,
+      "loss": 0.534,
+      "step": 3092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6759824264960882e-05,
+      "loss": 0.5394,
+      "step": 3093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757618269199315e-05,
+      "loss": 0.5342,
+      "step": 3094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675541166803309e-05,
+      "loss": 0.499,
+      "step": 3095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753204461659904e-05,
+      "loss": 0.4934,
+      "step": 3096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750996650277492e-05,
+      "loss": 0.4672,
+      "step": 3097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674878823408365e-05,
+      "loss": 0.4851,
+      "step": 3098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746579213276224e-05,
+      "loss": 0.4812,
+      "step": 3099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674436958805312e-05,
+      "loss": 0.5183,
+      "step": 3100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742159358612294e-05,
+      "loss": 0.5263,
+      "step": 3101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739948525151754e-05,
+      "loss": 0.54,
+      "step": 3102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673773708786957e-05,
+      "loss": 0.5276,
+      "step": 3103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673552504696386e-05,
+      "loss": 0.5194,
+      "step": 3104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733312402632792e-05,
+      "loss": 0.4893,
+      "step": 3105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731099155074603e-05,
+      "loss": 0.5275,
+      "step": 3106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672888530448757e-05,
+      "loss": 0.4992,
+      "step": 3107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726670851070026e-05,
+      "loss": 0.5342,
+      "step": 3108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672445579502036e-05,
+      "loss": 0.5332,
+      "step": 3109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672224013653702e-05,
+      "loss": 0.5036,
+      "step": 3110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67200238758185e-05,
+      "loss": 0.5031,
+      "step": 3111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717807013063356e-05,
+      "loss": 0.5219,
+      "step": 3112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715589548470187e-05,
+      "loss": 0.4946,
+      "step": 3113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671337148223765e-05,
+      "loss": 0.4902,
+      "step": 3114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711152814564472e-05,
+      "loss": 0.5096,
+      "step": 3115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708933545649408e-05,
+      "loss": 0.5182,
+      "step": 3116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706713675691283e-05,
+      "loss": 0.4658,
+      "step": 3117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670449320488897e-05,
+      "loss": 0.4921,
+      "step": 3118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67022721334414e-05,
+      "loss": 0.516,
+      "step": 3119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700050461547554e-05,
+      "loss": 0.4742,
+      "step": 3120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6697828189406465e-05,
+      "loss": 0.5115,
+      "step": 3121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669560531721723e-05,
+      "loss": 0.5117,
+      "step": 3122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693381845178985e-05,
+      "loss": 0.5092,
+      "step": 3123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691157773490937e-05,
+      "loss": 0.4885,
+      "step": 3124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6688933102352326e-05,
+      "loss": 0.4988,
+      "step": 3125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668670783196247e-05,
+      "loss": 0.4441,
+      "step": 3126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668448196252071e-05,
+      "loss": 0.4783,
+      "step": 3127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668225549422647e-05,
+      "loss": 0.529,
+      "step": 3128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680028427279217e-05,
+      "loss": 0.4934,
+      "step": 3129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677800761878467e-05,
+      "loss": 0.5393,
+      "step": 3130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675572498223794e-05,
+      "loss": 0.432,
+      "step": 3131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673343636514824e-05,
+      "loss": 0.4474,
+      "step": 3132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6671114176951236e-05,
+      "loss": 0.5216,
+      "step": 3133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668884119732768e-05,
+      "loss": 0.5142,
+      "step": 3134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66666534650592e-05,
+      "loss": 0.5059,
+      "step": 3135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664422213130382e-05,
+      "loss": 0.5466,
+      "step": 3136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662190364146207e-05,
+      "loss": 0.4946,
+      "step": 3137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6659957918306615e-05,
+      "loss": 0.485,
+      "step": 3138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6657724875811614e-05,
+      "loss": 0.5377,
+      "step": 3139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665549123686126e-05,
+      "loss": 0.5015,
+      "step": 3140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653257001655652e-05,
+      "loss": 0.4652,
+      "step": 3141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665102217039497e-05,
+      "loss": 0.5536,
+      "step": 3142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648786743279404e-05,
+      "loss": 0.5331,
+      "step": 3143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646550720509248e-05,
+      "loss": 0.4707,
+      "step": 3144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6644314102284803e-05,
+      "loss": 0.487,
+      "step": 3145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642076888806456e-05,
+      "loss": 0.551,
+      "step": 3146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639839080274637e-05,
+      "loss": 0.4793,
+      "step": 3147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6637600676889817e-05,
+      "loss": 0.4993,
+      "step": 3148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663536167885254e-05,
+      "loss": 0.5186,
+      "step": 3149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663312208636339e-05,
+      "loss": 0.485,
+      "step": 3150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630881899623012e-05,
+      "loss": 0.476,
+      "step": 3151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66286411188321e-05,
+      "loss": 0.492,
+      "step": 3152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626399744191402e-05,
+      "loss": 0.49,
+      "step": 3153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662415777590172e-05,
+      "loss": 0.4837,
+      "step": 3154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621915214163907e-05,
+      "loss": 0.4337,
+      "step": 3155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6619672059178868e-05,
+      "loss": 0.5123,
+      "step": 3156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661742831114757e-05,
+      "loss": 0.4963,
+      "step": 3157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661518397027102e-05,
+      "loss": 0.4709,
+      "step": 3158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612939036750295e-05,
+      "loss": 0.4857,
+      "step": 3159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610693510786503e-05,
+      "loss": 0.5094,
+      "step": 3160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6608447392580825e-05,
+      "loss": 0.4197,
+      "step": 3161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660620068233449e-05,
+      "loss": 0.4522,
+      "step": 3162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6603953380248764e-05,
+      "loss": 0.5569,
+      "step": 3163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601705486524996e-05,
+      "loss": 0.4873,
+      "step": 3164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659945700136456e-05,
+      "loss": 0.5356,
+      "step": 3165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65972079249689e-05,
+      "loss": 0.4962,
+      "step": 3166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594958257539505e-05,
+      "loss": 0.4853,
+      "step": 3167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592707999277914e-05,
+      "loss": 0.5078,
+      "step": 3168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659045715038574e-05,
+      "loss": 0.4827,
+      "step": 3169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588205711064615e-05,
+      "loss": 0.4546,
+      "step": 3170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658595368151625e-05,
+      "loss": 0.5588,
+      "step": 3171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583701061942403e-05,
+      "loss": 0.4909,
+      "step": 3172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581447852544877e-05,
+      "loss": 0.4681,
+      "step": 3173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657919405352554e-05,
+      "loss": 0.5107,
+      "step": 3174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576939665086303e-05,
+      "loss": 0.4919,
+      "step": 3175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574684687429133e-05,
+      "loss": 0.4752,
+      "step": 3176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657242912075605e-05,
+      "loss": 0.5043,
+      "step": 3177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657017296526913e-05,
+      "loss": 0.5414,
+      "step": 3178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567916221170492e-05,
+      "loss": 0.5529,
+      "step": 3179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656565888866232e-05,
+      "loss": 0.4718,
+      "step": 3180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6563400967946844e-05,
+      "loss": 0.4683,
+      "step": 3181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561142459226344e-05,
+      "loss": 0.498,
+      "step": 3182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558883362703162e-05,
+      "loss": 0.5539,
+      "step": 3183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655662367857968e-05,
+      "loss": 0.5283,
+      "step": 3184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554363407058346e-05,
+      "loss": 0.4789,
+      "step": 3185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655210254834165e-05,
+      "loss": 0.4937,
+      "step": 3186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6549841102632144e-05,
+      "loss": 0.4803,
+      "step": 3187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654757907013242e-05,
+      "loss": 0.477,
+      "step": 3188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654531645104514e-05,
+      "loss": 0.4973,
+      "step": 3189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543053245572997e-05,
+      "loss": 0.4768,
+      "step": 3190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540789453918755e-05,
+      "loss": 0.499,
+      "step": 3191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538525076285227e-05,
+      "loss": 0.5474,
+      "step": 3192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536260112875265e-05,
+      "loss": 0.5084,
+      "step": 3193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653399456389179e-05,
+      "loss": 0.5092,
+      "step": 3194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6531728429537766e-05,
+      "loss": 0.4827,
+      "step": 3195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529461710016218e-05,
+      "loss": 0.4821,
+      "step": 3196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6527194405530217e-05,
+      "loss": 0.5058,
+      "step": 3197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6524926516282884e-05,
+      "loss": 0.4948,
+      "step": 3198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6522658042477397e-05,
+      "loss": 0.5529,
+      "step": 3199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652038898431698e-05,
+      "loss": 0.5263,
+      "step": 3200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6518119342004923e-05,
+      "loss": 0.5336,
+      "step": 3201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6515849115744555e-05,
+      "loss": 0.4824,
+      "step": 3202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513578305739263e-05,
+      "loss": 0.4536,
+      "step": 3203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511306912192487e-05,
+      "loss": 0.533,
+      "step": 3204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509034935307716e-05,
+      "loss": 0.5187,
+      "step": 3205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650676237528849e-05,
+      "loss": 0.4945,
+      "step": 3206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650448923233841e-05,
+      "loss": 0.5282,
+      "step": 3207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650221550666112e-05,
+      "loss": 0.5069,
+      "step": 3208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499941198460317e-05,
+      "loss": 0.4451,
+      "step": 3209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649766630793976e-05,
+      "loss": 0.5021,
+      "step": 3210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649539083530325e-05,
+      "loss": 0.5351,
+      "step": 3211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493114780754636e-05,
+      "loss": 0.4492,
+      "step": 3212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649083814449784e-05,
+      "loss": 0.4566,
+      "step": 3213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6488560926736807e-05,
+      "loss": 0.465,
+      "step": 3214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486283127675565e-05,
+      "loss": 0.4742,
+      "step": 3215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484004747518164e-05,
+      "loss": 0.4908,
+      "step": 3216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648172578646873e-05,
+      "loss": 0.5259,
+      "step": 3217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479446244731426e-05,
+      "loss": 0.5068,
+      "step": 3218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647716612251048e-05,
+      "loss": 0.4734,
+      "step": 3219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474885420010156e-05,
+      "loss": 0.5599,
+      "step": 3220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472604137434785e-05,
+      "loss": 0.5006,
+      "step": 3221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470322274988742e-05,
+      "loss": 0.5345,
+      "step": 3222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6468039832876454e-05,
+      "loss": 0.5296,
+      "step": 3223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64657568113024e-05,
+      "loss": 0.4843,
+      "step": 3224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463473210471118e-05,
+      "loss": 0.4731,
+      "step": 3225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461189030587187e-05,
+      "loss": 0.4414,
+      "step": 3226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6458904271855245e-05,
+      "loss": 0.5305,
+      "step": 3227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6456618934479982e-05,
+      "loss": 0.4946,
+      "step": 3228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645433301866614e-05,
+      "loss": 0.4616,
+      "step": 3229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6452046524618496e-05,
+      "loss": 0.5094,
+      "step": 3230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449759452541915e-05,
+      "loss": 0.4884,
+      "step": 3231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6447471802641274e-05,
+      "loss": 0.4348,
+      "step": 3232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445183575121534e-05,
+      "loss": 0.47,
+      "step": 3233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442894770187684e-05,
+      "loss": 0.5378,
+      "step": 3234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440605388044777e-05,
+      "loss": 0.4873,
+      "step": 3235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6438315428897914e-05,
+      "loss": 0.5184,
+      "step": 3236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436024892952256e-05,
+      "loss": 0.5183,
+      "step": 3237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433733780413e-05,
+      "loss": 0.473,
+      "step": 3238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643144209148541e-05,
+      "loss": 0.507,
+      "step": 3239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429149826374792e-05,
+      "loss": 0.5451,
+      "step": 3240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6426856985286505e-05,
+      "loss": 0.4788,
+      "step": 3241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424563568425963e-05,
+      "loss": 0.4785,
+      "step": 3242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422269575998632e-05,
+      "loss": 0.5513,
+      "step": 3243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641997500821002e-05,
+      "loss": 0.4699,
+      "step": 3244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64176798652657e-05,
+      "loss": 0.5016,
+      "step": 3245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415384147371286e-05,
+      "loss": 0.5394,
+      "step": 3246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6413087854732455e-05,
+      "loss": 0.5299,
+      "step": 3247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641079098755492e-05,
+      "loss": 0.479,
+      "step": 3248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640849354604446e-05,
+      "loss": 0.5182,
+      "step": 3249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406195530406894e-05,
+      "loss": 0.5019,
+      "step": 3250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403896940848104e-05,
+      "loss": 0.4678,
+      "step": 3251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640159777757401e-05,
+      "loss": 0.5077,
+      "step": 3252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399298040790598e-05,
+      "loss": 0.4625,
+      "step": 3253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639699773070389e-05,
+      "loss": 0.4704,
+      "step": 3254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394696847519974e-05,
+      "loss": 0.5102,
+      "step": 3255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639239539144498e-05,
+      "loss": 0.5089,
+      "step": 3256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639009336268509e-05,
+      "loss": 0.5151,
+      "step": 3257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638779076144654e-05,
+      "loss": 0.5118,
+      "step": 3258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385487587935623e-05,
+      "loss": 0.4782,
+      "step": 3259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383183842358667e-05,
+      "loss": 0.5134,
+      "step": 3260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380879524922066e-05,
+      "loss": 0.4918,
+      "step": 3261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637857463583226e-05,
+      "loss": 0.4814,
+      "step": 3262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637626917529574e-05,
+      "loss": 0.4916,
+      "step": 3263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637396314351905e-05,
+      "loss": 0.4753,
+      "step": 3264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371656540708782e-05,
+      "loss": 0.4773,
+      "step": 3265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636934936707158e-05,
+      "loss": 0.5166,
+      "step": 3266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367041622814144e-05,
+      "loss": 0.4844,
+      "step": 3267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364733308143213e-05,
+      "loss": 0.4235,
+      "step": 3268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63624244232656e-05,
+      "loss": 0.5505,
+      "step": 3269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636011496838814e-05,
+      "loss": 0.5095,
+      "step": 3270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357804943717742e-05,
+      "loss": 0.5621,
+      "step": 3271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355494349461355e-05,
+      "loss": 0.4683,
+      "step": 3272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635318318582598e-05,
+      "loss": 0.4675,
+      "step": 3273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6350871453018675e-05,
+      "loss": 0.5198,
+      "step": 3274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348559151246538e-05,
+      "loss": 0.4836,
+      "step": 3275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6346246280716733e-05,
+      "loss": 0.5192,
+      "step": 3276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.4879,
+      "step": 3277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341618834212974e-05,
+      "loss": 0.4895,
+      "step": 3278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633930425865359e-05,
+      "loss": 0.5032,
+      "step": 3279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336989115165665e-05,
+      "loss": 0.4732,
+      "step": 3280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633467340395661e-05,
+      "loss": 0.4999,
+      "step": 3281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332357125233884e-05,
+      "loss": 0.5113,
+      "step": 3282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6330040279205004e-05,
+      "loss": 0.52,
+      "step": 3283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6327722866077524e-05,
+      "loss": 0.4938,
+      "step": 3284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632540488605906e-05,
+      "loss": 0.4968,
+      "step": 3285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323086339357287e-05,
+      "loss": 0.4655,
+      "step": 3286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320767226179902e-05,
+      "loss": 0.5209,
+      "step": 3287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318447546734687e-05,
+      "loss": 0.5212,
+      "step": 3288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6316127301229448e-05,
+      "loss": 0.4841,
+      "step": 3289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313806489872058e-05,
+      "loss": 0.3998,
+      "step": 3290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631148511287043e-05,
+      "loss": 0.4989,
+      "step": 3291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630916317043254e-05,
+      "loss": 0.4775,
+      "step": 3292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63068406627664e-05,
+      "loss": 0.5704,
+      "step": 3293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630451759008008e-05,
+      "loss": 0.4983,
+      "step": 3294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302193952581708e-05,
+      "loss": 0.5422,
+      "step": 3295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629986975047945e-05,
+      "loss": 0.5243,
+      "step": 3296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297544983981524e-05,
+      "loss": 0.5426,
+      "step": 3297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629521965329621e-05,
+      "loss": 0.4874,
+      "step": 3298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6292893758631824e-05,
+      "loss": 0.4943,
+      "step": 3299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6290567300196746e-05,
+      "loss": 0.5185,
+      "step": 3300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288240278199393e-05,
+      "loss": 0.4165,
+      "step": 3301
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285912692848243e-05,
+      "loss": 0.504,
+      "step": 3302
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628358454435182e-05,
+      "loss": 0.4882,
+      "step": 3303
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281255832918705e-05,
+      "loss": 0.4881,
+      "step": 3304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278926558757514e-05,
+      "loss": 0.4977,
+      "step": 3305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276596722076927e-05,
+      "loss": 0.4555,
+      "step": 3306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627426632308567e-05,
+      "loss": 0.4759,
+      "step": 3307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271935361992525e-05,
+      "loss": 0.5583,
+      "step": 3308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626960383900631e-05,
+      "loss": 0.5034,
+      "step": 3309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267271754335908e-05,
+      "loss": 0.4826,
+      "step": 3310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626493910819025e-05,
+      "loss": 0.4823,
+      "step": 3311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626260590077831e-05,
+      "loss": 0.4958,
+      "step": 3312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6260272132309115e-05,
+      "loss": 0.4616,
+      "step": 3313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6257937802991747e-05,
+      "loss": 0.5267,
+      "step": 3314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255602913035336e-05,
+      "loss": 0.4743,
+      "step": 3315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253267462649052e-05,
+      "loss": 0.5357,
+      "step": 3316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250931452042136e-05,
+      "loss": 0.4766,
+      "step": 3317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248594881423866e-05,
+      "loss": 0.5107,
+      "step": 3318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246257751003565e-05,
+      "loss": 0.5087,
+      "step": 3319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6243920060990618e-05,
+      "loss": 0.5096,
+      "step": 3320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241581811594453e-05,
+      "loss": 0.4994,
+      "step": 3321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623924300302455e-05,
+      "loss": 0.4978,
+      "step": 3322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6236903635490445e-05,
+      "loss": 0.5004,
+      "step": 3323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623456370920171e-05,
+      "loss": 0.5737,
+      "step": 3324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623222322436798e-05,
+      "loss": 0.5399,
+      "step": 3325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229882181198934e-05,
+      "loss": 0.5264,
+      "step": 3326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62275405799043e-05,
+      "loss": 0.5495,
+      "step": 3327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225198420693865e-05,
+      "loss": 0.5048,
+      "step": 3328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222855703777457e-05,
+      "loss": 0.4385,
+      "step": 3329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220512429364953e-05,
+      "loss": 0.4831,
+      "step": 3330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218168597666287e-05,
+      "loss": 0.4685,
+      "step": 3331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215824208891434e-05,
+      "loss": 0.5156,
+      "step": 3332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6213479263250433e-05,
+      "loss": 0.4911,
+      "step": 3333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6211133760953358e-05,
+      "loss": 0.5279,
+      "step": 3334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208787702210335e-05,
+      "loss": 0.5328,
+      "step": 3335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206441087231556e-05,
+      "loss": 0.5276,
+      "step": 3336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204093916227242e-05,
+      "loss": 0.4962,
+      "step": 3337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6201746189407674e-05,
+      "loss": 0.4881,
+      "step": 3338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199397906983176e-05,
+      "loss": 0.5088,
+      "step": 3339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6197049069164138e-05,
+      "loss": 0.5051,
+      "step": 3340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619469967616098e-05,
+      "loss": 0.4795,
+      "step": 3341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619234972818419e-05,
+      "loss": 0.5295,
+      "step": 3342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189999225444283e-05,
+      "loss": 0.5135,
+      "step": 3343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187648168151844e-05,
+      "loss": 0.4928,
+      "step": 3344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185296556517504e-05,
+      "loss": 0.5009,
+      "step": 3345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182944390751932e-05,
+      "loss": 0.4881,
+      "step": 3346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6180591671065865e-05,
+      "loss": 0.5231,
+      "step": 3347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617823839767007e-05,
+      "loss": 0.4999,
+      "step": 3348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617588457077538e-05,
+      "loss": 0.505,
+      "step": 3349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173530190592665e-05,
+      "loss": 0.5195,
+      "step": 3350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6171175257332855e-05,
+      "loss": 0.4884,
+      "step": 3351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168819771206922e-05,
+      "loss": 0.5004,
+      "step": 3352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166463732425894e-05,
+      "loss": 0.4788,
+      "step": 3353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616410714120084e-05,
+      "loss": 0.5247,
+      "step": 3354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161749997742882e-05,
+      "loss": 0.4889,
+      "step": 3355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159392302263207e-05,
+      "loss": 0.5208,
+      "step": 3356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615703405497302e-05,
+      "loss": 0.4711,
+      "step": 3357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154675256083604e-05,
+      "loss": 0.5347,
+      "step": 3358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615231590580627e-05,
+      "loss": 0.4929,
+      "step": 3359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61499560043524e-05,
+      "loss": 0.5241,
+      "step": 3360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6147595551933405e-05,
+      "loss": 0.4739,
+      "step": 3361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145234548760756e-05,
+      "loss": 0.4815,
+      "step": 3362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142872995045976e-05,
+      "loss": 0.4901,
+      "step": 3363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140510891000635e-05,
+      "loss": 0.5247,
+      "step": 3364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613814823683634e-05,
+      "loss": 0.4943,
+      "step": 3365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135785032764764e-05,
+      "loss": 0.4742,
+      "step": 3366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133421278997624e-05,
+      "loss": 0.4897,
+      "step": 3367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613105697574668e-05,
+      "loss": 0.4785,
+      "step": 3368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128692123223753e-05,
+      "loss": 0.4995,
+      "step": 3369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61263267216407e-05,
+      "loss": 0.5297,
+      "step": 3370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612396077120944e-05,
+      "loss": 0.537,
+      "step": 3371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121594272141927e-05,
+      "loss": 0.4855,
+      "step": 3372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611922722465018e-05,
+      "loss": 0.4955,
+      "step": 3373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116859628946256e-05,
+      "loss": 0.5052,
+      "step": 3374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611449148524226e-05,
+      "loss": 0.5254,
+      "step": 3375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112122793750358e-05,
+      "loss": 0.5309,
+      "step": 3376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109753554682752e-05,
+      "loss": 0.517,
+      "step": 3377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61073837682517e-05,
+      "loss": 0.5206,
+      "step": 3378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610501343466951e-05,
+      "loss": 0.5294,
+      "step": 3379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102642554148533e-05,
+      "loss": 0.5221,
+      "step": 3380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6100271126901175e-05,
+      "loss": 0.4815,
+      "step": 3381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097899153139885e-05,
+      "loss": 0.5122,
+      "step": 3382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609552663307717e-05,
+      "loss": 0.5165,
+      "step": 3383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6093153566925574e-05,
+      "loss": 0.4541,
+      "step": 3384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6090779954897702e-05,
+      "loss": 0.5161,
+      "step": 3385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60884057972062e-05,
+      "loss": 0.4794,
+      "step": 3386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608603109406376e-05,
+      "loss": 0.5172,
+      "step": 3387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083655845683143e-05,
+      "loss": 0.5339,
+      "step": 3388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081280052277125e-05,
+      "loss": 0.5501,
+      "step": 3389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078903714058564e-05,
+      "loss": 0.467,
+      "step": 3390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076526831240342e-05,
+      "loss": 0.5056,
+      "step": 3391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074149404035407e-05,
+      "loss": 0.5209,
+      "step": 3392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607177143265675e-05,
+      "loss": 0.4607,
+      "step": 3393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069392917317407e-05,
+      "loss": 0.4951,
+      "step": 3394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067013858230468e-05,
+      "loss": 0.5048,
+      "step": 3395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064634255609063e-05,
+      "loss": 0.5134,
+      "step": 3396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062254109666383e-05,
+      "loss": 0.4945,
+      "step": 3397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059873420615658e-05,
+      "loss": 0.492,
+      "step": 3398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057492188670173e-05,
+      "loss": 0.4795,
+      "step": 3399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605511041404326e-05,
+      "loss": 0.5321,
+      "step": 3400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605272809694829e-05,
+      "loss": 0.4471,
+      "step": 3401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050345237598702e-05,
+      "loss": 0.5192,
+      "step": 3402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047961836207967e-05,
+      "loss": 0.4516,
+      "step": 3403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045577892989612e-05,
+      "loss": 0.5356,
+      "step": 3404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043193408157208e-05,
+      "loss": 0.496,
+      "step": 3405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040808381924384e-05,
+      "loss": 0.5004,
+      "step": 3406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038422814504807e-05,
+      "loss": 0.4608,
+      "step": 3407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036036706112193e-05,
+      "loss": 0.5685,
+      "step": 3408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603365005696031e-05,
+      "loss": 0.4884,
+      "step": 3409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603126286726298e-05,
+      "loss": 0.5189,
+      "step": 3410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028875137234066e-05,
+      "loss": 0.5018,
+      "step": 3411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026486867087477e-05,
+      "loss": 0.4902,
+      "step": 3412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602409805703718e-05,
+      "loss": 0.5299,
+      "step": 3413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021708707297178e-05,
+      "loss": 0.4442,
+      "step": 3414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019318818081534e-05,
+      "loss": 0.5722,
+      "step": 3415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016928389604353e-05,
+      "loss": 0.4956,
+      "step": 3416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6014537422079796e-05,
+      "loss": 0.5247,
+      "step": 3417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012145915722053e-05,
+      "loss": 0.4757,
+      "step": 3418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009753870745386e-05,
+      "loss": 0.4888,
+      "step": 3419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600736128736409e-05,
+      "loss": 0.4913,
+      "step": 3420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004968165792516e-05,
+      "loss": 0.4777,
+      "step": 3421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002574506245057e-05,
+      "loss": 0.5236,
+      "step": 3422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600018030893616e-05,
+      "loss": 0.5293,
+      "step": 3423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599778557408032e-05,
+      "loss": 0.4594,
+      "step": 3424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599539030189207e-05,
+      "loss": 0.4366,
+      "step": 3425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992994492586e-05,
+      "loss": 0.5,
+      "step": 3426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990598146376753e-05,
+      "loss": 0.5244,
+      "step": 3427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988201263479007e-05,
+      "loss": 0.4949,
+      "step": 3428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985803844107502e-05,
+      "loss": 0.5147,
+      "step": 3429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5983405888477016e-05,
+      "loss": 0.4495,
+      "step": 3430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981007396802377e-05,
+      "loss": 0.4674,
+      "step": 3431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5978608369298462e-05,
+      "loss": 0.5203,
+      "step": 3432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976208806180196e-05,
+      "loss": 0.5369,
+      "step": 3433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973808707662554e-05,
+      "loss": 0.4255,
+      "step": 3434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971408073960558e-05,
+      "loss": 0.4609,
+      "step": 3435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5969006905289277e-05,
+      "loss": 0.525,
+      "step": 3436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5966605201863822e-05,
+      "loss": 0.4949,
+      "step": 3437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5964202963899368e-05,
+      "loss": 0.5056,
+      "step": 3438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961800191611118e-05,
+      "loss": 0.5329,
+      "step": 3439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5959396885214337e-05,
+      "loss": 0.4814,
+      "step": 3440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956993044924334e-05,
+      "loss": 0.475,
+      "step": 3441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954588670956467e-05,
+      "loss": 0.436,
+      "step": 3442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5952183763526137e-05,
+      "loss": 0.4948,
+      "step": 3443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949778322848794e-05,
+      "loss": 0.5235,
+      "step": 3444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5947372349139943e-05,
+      "loss": 0.5184,
+      "step": 3445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5944965842615133e-05,
+      "loss": 0.5247,
+      "step": 3446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942558803489952e-05,
+      "loss": 0.5111,
+      "step": 3447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940151231980047e-05,
+      "loss": 0.4642,
+      "step": 3448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593774312830111e-05,
+      "loss": 0.4964,
+      "step": 3449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5935334492668873e-05,
+      "loss": 0.5336,
+      "step": 3450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593292532529913e-05,
+      "loss": 0.465,
+      "step": 3451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593051562640771e-05,
+      "loss": 0.5217,
+      "step": 3452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5928105396210494e-05,
+      "loss": 0.5322,
+      "step": 3453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925694634923415e-05,
+      "loss": 0.5345,
+      "step": 3454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923283342762444e-05,
+      "loss": 0.5185,
+      "step": 3455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5920871519943605e-05,
+      "loss": 0.5501,
+      "step": 3456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5918459166682977e-05,
+      "loss": 0.4603,
+      "step": 3457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916046283196668e-05,
+      "loss": 0.524,
+      "step": 3458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591363286970086e-05,
+      "loss": 0.464,
+      "step": 3459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911218926411746e-05,
+      "loss": 0.4706,
+      "step": 3460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5908804453545608e-05,
+      "loss": 0.4243,
+      "step": 3461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906389451318742e-05,
+      "loss": 0.5219,
+      "step": 3462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590397391994751e-05,
+      "loss": 0.4889,
+      "step": 3463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590155785964831e-05,
+      "loss": 0.4412,
+      "step": 3464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58991412706376e-05,
+      "loss": 0.4611,
+      "step": 3465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896724153131877e-05,
+      "loss": 0.5472,
+      "step": 3466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589430650734768e-05,
+      "loss": 0.5326,
+      "step": 3467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891888333501615e-05,
+      "loss": 0.5076,
+      "step": 3468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588946963181031e-05,
+      "loss": 0.5653,
+      "step": 3469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5887050402490462e-05,
+      "loss": 0.4228,
+      "step": 3470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58846306457588e-05,
+      "loss": 0.4724,
+      "step": 3471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882210361832108e-05,
+      "loss": 0.5672,
+      "step": 3472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587978955092721e-05,
+      "loss": 0.4632,
+      "step": 3473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877368213260992e-05,
+      "loss": 0.5016,
+      "step": 3474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587494634905038e-05,
+      "loss": 0.5288,
+      "step": 3475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872523958512336e-05,
+      "loss": 0.518,
+      "step": 3476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587010104186388e-05,
+      "loss": 0.5221,
+      "step": 3477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586767759932208e-05,
+      "loss": 0.5161,
+      "step": 3478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586525363110405e-05,
+      "loss": 0.471,
+      "step": 3479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586282913742694e-05,
+      "loss": 0.4752,
+      "step": 3480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5860404118507972e-05,
+      "loss": 0.4872,
+      "step": 3481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585797857456439e-05,
+      "loss": 0.5107,
+      "step": 3482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855552505813493e-05,
+      "loss": 0.456,
+      "step": 3483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853125912472637e-05,
+      "loss": 0.4851,
+      "step": 3484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585069879475921e-05,
+      "loss": 0.4424,
+      "step": 3485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848271152890653e-05,
+      "loss": 0.4806,
+      "step": 3486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584584298708446e-05,
+      "loss": 0.5451,
+      "step": 3487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843414297558165e-05,
+      "loss": 0.5405,
+      "step": 3488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5840985084529347e-05,
+      "loss": 0.5143,
+      "step": 3489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583855534821564e-05,
+      "loss": 0.4524,
+      "step": 3490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836125088834716e-05,
+      "loss": 0.5114,
+      "step": 3491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58336943066043e-05,
+      "loss": 0.4756,
+      "step": 3492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831263001742168e-05,
+      "loss": 0.4651,
+      "step": 3493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5828831174466125e-05,
+      "loss": 0.4909,
+      "step": 3494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826398824994047e-05,
+      "loss": 0.5112,
+      "step": 3495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5823965953543834e-05,
+      "loss": 0.5023,
+      "step": 3496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821532560333452e-05,
+      "loss": 0.4883,
+      "step": 3497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58190986455809e-05,
+      "loss": 0.4967,
+      "step": 3498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816664209504227e-05,
+      "loss": 0.4869,
+      "step": 3499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814229252321535e-05,
+      "loss": 0.5163,
+      "step": 3500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5811793774250964e-05,
+      "loss": 0.4854,
+      "step": 3501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580935777551071e-05,
+      "loss": 0.5162,
+      "step": 3502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5806921256319004e-05,
+      "loss": 0.4863,
+      "step": 3503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804484216894137e-05,
+      "loss": 0.4797,
+      "step": 3504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802046657454432e-05,
+      "loss": 0.4446,
+      "step": 3505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5799608578218272e-05,
+      "loss": 0.5576,
+      "step": 3506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797169979404075e-05,
+      "loss": 0.4652,
+      "step": 3507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794730861230322e-05,
+      "loss": 0.5286,
+      "step": 3508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5792291223915518e-05,
+      "loss": 0.4797,
+      "step": 3509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789851067678233e-05,
+      "loss": 0.5041,
+      "step": 3510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5787410392737073e-05,
+      "loss": 0.5123,
+      "step": 3511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57849691993107e-05,
+      "loss": 0.4954,
+      "step": 3512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578252748761781e-05,
+      "loss": 0.4861,
+      "step": 3513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578008525787716e-05,
+      "loss": 0.4855,
+      "step": 3514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777642510307536e-05,
+      "loss": 0.4664,
+      "step": 3515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577519924512779e-05,
+      "loss": 0.5084,
+      "step": 3516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57727554625568e-05,
+      "loss": 0.4961,
+      "step": 3517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577031116281351e-05,
+      "loss": 0.4419,
+      "step": 3518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767866346116896e-05,
+      "loss": 0.5535,
+      "step": 3519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5765421012685986e-05,
+      "loss": 0.5419,
+      "step": 3520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5762975162739854e-05,
+      "loss": 0.4869,
+      "step": 3521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5760528796497623e-05,
+      "loss": 0.439,
+      "step": 3522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758081914178457e-05,
+      "loss": 0.5301,
+      "step": 3523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755634516001563e-05,
+      "loss": 0.4361,
+      "step": 3524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753186602186207e-05,
+      "loss": 0.5034,
+      "step": 3525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750738172951692e-05,
+      "loss": 0.5344,
+      "step": 3526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574828922851737e-05,
+      "loss": 0.531,
+      "step": 3527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574583976910263e-05,
+      "loss": 0.4466,
+      "step": 3528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574338979492693e-05,
+      "loss": 0.5397,
+      "step": 3529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5740939306209746e-05,
+      "loss": 0.4612,
+      "step": 3530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573848830317062e-05,
+      "loss": 0.5097,
+      "step": 3531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736036786029135e-05,
+      "loss": 0.4515,
+      "step": 3532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733584755004912e-05,
+      "loss": 0.4912,
+      "step": 3533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5731132210317634e-05,
+      "loss": 0.5038,
+      "step": 3534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572867915218701e-05,
+      "loss": 0.4919,
+      "step": 3535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726225580832813e-05,
+      "loss": 0.5112,
+      "step": 3536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723771496474854e-05,
+      "loss": 0.4434,
+      "step": 3537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572131689933299e-05,
+      "loss": 0.513,
+      "step": 3538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718861789627125e-05,
+      "loss": 0.487,
+      "step": 3539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716406167577206e-05,
+      "loss": 0.5144,
+      "step": 3540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571395003340323e-05,
+      "loss": 0.4689,
+      "step": 3541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711493387325237e-05,
+      "loss": 0.4927,
+      "step": 3542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570903622956332e-05,
+      "loss": 0.5062,
+      "step": 3543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706578560337603e-05,
+      "loss": 0.5125,
+      "step": 3544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704120379868272e-05,
+      "loss": 0.4676,
+      "step": 3545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701661688375547e-05,
+      "loss": 0.5049,
+      "step": 3546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699202486079702e-05,
+      "loss": 0.5057,
+      "step": 3547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569674277320105e-05,
+      "loss": 0.5098,
+      "step": 3548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694282549959953e-05,
+      "loss": 0.5512,
+      "step": 3549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691821816576822e-05,
+      "loss": 0.5297,
+      "step": 3550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568936057327211e-05,
+      "loss": 0.5403,
+      "step": 3551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568689882026631e-05,
+      "loss": 0.4743,
+      "step": 3552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5684436557779975e-05,
+      "loss": 0.4422,
+      "step": 3553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5681973786033693e-05,
+      "loss": 0.4869,
+      "step": 3554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567951050524809e-05,
+      "loss": 0.52,
+      "step": 3555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677046715643865e-05,
+      "loss": 0.4996,
+      "step": 3556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5674582417441734e-05,
+      "loss": 0.502,
+      "step": 3557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567211761086247e-05,
+      "loss": 0.4537,
+      "step": 3558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566965229612689e-05,
+      "loss": 0.4903,
+      "step": 3559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5667186473455866e-05,
+      "loss": 0.4899,
+      "step": 3560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56647201430703e-05,
+      "loss": 0.4619,
+      "step": 3561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662253305191147e-05,
+      "loss": 0.496,
+      "step": 3562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659785960039412e-05,
+      "loss": 0.5378,
+      "step": 3563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657318107836133e-05,
+      "loss": 0.4916,
+      "step": 3564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654849748802406e-05,
+      "loss": 0.4527,
+      "step": 3565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565238088315937e-05,
+      "loss": 0.5234,
+      "step": 3566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5649911511128206e-05,
+      "loss": 0.5104,
+      "step": 3567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647441632930138e-05,
+      "loss": 0.4973,
+      "step": 3568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564497124878644e-05,
+      "loss": 0.4941,
+      "step": 3569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564250035891843e-05,
+      "loss": 0.4545,
+      "step": 3570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640028963547468e-05,
+      "loss": 0.4906,
+      "step": 3571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5637557062894968e-05,
+      "loss": 0.504,
+      "step": 3572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563508465718238e-05,
+      "loss": 0.5154,
+      "step": 3573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5632611746631205e-05,
+      "loss": 0.5104,
+      "step": 3574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630138331462988e-05,
+      "loss": 0.4721,
+      "step": 3575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5627664411899316e-05,
+      "loss": 0.4485,
+      "step": 3576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5625189988161824e-05,
+      "loss": 0.5016,
+      "step": 3577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622715060472192e-05,
+      "loss": 0.5419,
+      "step": 3578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562023962905215e-05,
+      "loss": 0.5102,
+      "step": 3579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617763694123456e-05,
+      "loss": 0.4748,
+      "step": 3580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561528725590794e-05,
+      "loss": 0.4729,
+      "step": 3581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5612810314627452e-05,
+      "loss": 0.5212,
+      "step": 3582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610332870503904e-05,
+      "loss": 0.4979,
+      "step": 3583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560785492375924e-05,
+      "loss": 0.4257,
+      "step": 3584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560537647461546e-05,
+      "loss": 0.5083,
+      "step": 3585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56028975232946e-05,
+      "loss": 0.4523,
+      "step": 3586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560041807001875e-05,
+      "loss": 0.5051,
+      "step": 3587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597938115010038e-05,
+      "loss": 0.5363,
+      "step": 3588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5595457658490643e-05,
+      "loss": 0.5161,
+      "step": 3589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5592976700682777e-05,
+      "loss": 0.4601,
+      "step": 3590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5590495241808715e-05,
+      "loss": 0.4964,
+      "step": 3591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588013282090763e-05,
+      "loss": 0.5131,
+      "step": 3592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585530821751274e-05,
+      "loss": 0.4626,
+      "step": 3593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5583047861012652e-05,
+      "loss": 0.482,
+      "step": 3594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580564400097337e-05,
+      "loss": 0.495,
+      "step": 3595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557808043922782e-05,
+      "loss": 0.4689,
+      "step": 3596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575595978626634e-05,
+      "loss": 0.4562,
+      "step": 3597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573111018516366e-05,
+      "loss": 0.4877,
+      "step": 3598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570625559119626e-05,
+      "loss": 0.463,
+      "step": 3599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5568139600659097e-05,
+      "loss": 0.4722,
+      "step": 3600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556565314335748e-05,
+      "loss": 0.5195,
+      "step": 3601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563166187437543e-05,
+      "loss": 0.5232,
+      "step": 3602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556067873312208e-05,
+      "loss": 0.4986,
+      "step": 3603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5558190780633945e-05,
+      "loss": 0.456,
+      "step": 3604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.4783,
+      "step": 3605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553213382031257e-05,
+      "loss": 0.5238,
+      "step": 3606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550723936362624e-05,
+      "loss": 0.4763,
+      "step": 3607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554823399341315e-05,
+      "loss": 0.3807,
+      "step": 3608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554574355340591e-05,
+      "loss": 0.3882,
+      "step": 3609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5543252616564013e-05,
+      "loss": 0.3369,
+      "step": 3610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554076118311062e-05,
+      "loss": 0.3476,
+      "step": 3611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5538269253268935e-05,
+      "loss": 0.3635,
+      "step": 3612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535776827262204e-05,
+      "loss": 0.37,
+      "step": 3613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533283905313723e-05,
+      "loss": 0.4043,
+      "step": 3614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530790487646825e-05,
+      "loss": 0.3859,
+      "step": 3615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528296574484898e-05,
+      "loss": 0.3645,
+      "step": 3616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552580216605136e-05,
+      "loss": 0.4039,
+      "step": 3617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552330726256969e-05,
+      "loss": 0.3798,
+      "step": 3618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520811864263395e-05,
+      "loss": 0.3594,
+      "step": 3619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518315971356036e-05,
+      "loss": 0.3977,
+      "step": 3620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5515819584071216e-05,
+      "loss": 0.3679,
+      "step": 3621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513322702632586e-05,
+      "loss": 0.3998,
+      "step": 3622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510825327263833e-05,
+      "loss": 0.3608,
+      "step": 3623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5508327458188696e-05,
+      "loss": 0.3575,
+      "step": 3624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5505829095630954e-05,
+      "loss": 0.3815,
+      "step": 3625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550333023981443e-05,
+      "loss": 0.3538,
+      "step": 3626
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5500830890962998e-05,
+      "loss": 0.3753,
+      "step": 3627
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549833104930057e-05,
+      "loss": 0.4002,
+      "step": 3628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495830715051093e-05,
+      "loss": 0.3781,
+      "step": 3629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493329888438576e-05,
+      "loss": 0.3872,
+      "step": 3630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5490828569687066e-05,
+      "loss": 0.4045,
+      "step": 3631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548832675902065e-05,
+      "loss": 0.3821,
+      "step": 3632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485824456663464e-05,
+      "loss": 0.3646,
+      "step": 3633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548332166283968e-05,
+      "loss": 0.4021,
+      "step": 3634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548081837777352e-05,
+      "loss": 0.3729,
+      "step": 3635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478314601689257e-05,
+      "loss": 0.4144,
+      "step": 3636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547581033481119e-05,
+      "loss": 0.3967,
+      "step": 3637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547330557736368e-05,
+      "loss": 0.3795,
+      "step": 3638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470800329571122e-05,
+      "loss": 0.4119,
+      "step": 3639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468294591657954e-05,
+      "loss": 0.3862,
+      "step": 3640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546578836384867e-05,
+      "loss": 0.3602,
+      "step": 3641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546328164636779e-05,
+      "loss": 0.3684,
+      "step": 3642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460774439439892e-05,
+      "loss": 0.4394,
+      "step": 3643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545826674328959e-05,
+      "loss": 0.3758,
+      "step": 3644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455758558141546e-05,
+      "loss": 0.3718,
+      "step": 3645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453249884220466e-05,
+      "loss": 0.3799,
+      "step": 3646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450740721751096e-05,
+      "loss": 0.3806,
+      "step": 3647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448231070958227e-05,
+      "loss": 0.3695,
+      "step": 3648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445720932066698e-05,
+      "loss": 0.3572,
+      "step": 3649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544321030530139e-05,
+      "loss": 0.379,
+      "step": 3650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440699190887216e-05,
+      "loss": 0.3842,
+      "step": 3651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5438187589049156e-05,
+      "loss": 0.3916,
+      "step": 3652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5435675500012212e-05,
+      "loss": 0.3511,
+      "step": 3653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5433162924001445e-05,
+      "loss": 0.4066,
+      "step": 3654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430649861241944e-05,
+      "loss": 0.3668,
+      "step": 3655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428136311958854e-05,
+      "loss": 0.368,
+      "step": 3656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5425622276377364e-05,
+      "loss": 0.4149,
+      "step": 3657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54231077547227e-05,
+      "loss": 0.4251,
+      "step": 3658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5420592747220134e-05,
+      "loss": 0.3639,
+      "step": 3659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5418077254094977e-05,
+      "loss": 0.3704,
+      "step": 3660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54155612755726e-05,
+      "loss": 0.3567,
+      "step": 3661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5413044811878393e-05,
+      "loss": 0.4092,
+      "step": 3662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5410527863237807e-05,
+      "loss": 0.3615,
+      "step": 3663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408010429876335e-05,
+      "loss": 0.3895,
+      "step": 3664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405492512019506e-05,
+      "loss": 0.3809,
+      "step": 3665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54029741098929e-05,
+      "loss": 0.3696,
+      "step": 3666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540045522372213e-05,
+      "loss": 0.3546,
+      "step": 3667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5397935853732862e-05,
+      "loss": 0.4108,
+      "step": 3668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395416000150808e-05,
+      "loss": 0.3757,
+      "step": 3669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392895663201715e-05,
+      "loss": 0.3428,
+      "step": 3670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539037484311137e-05,
+      "loss": 0.3489,
+      "step": 3671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387853540105612e-05,
+      "loss": 0.4323,
+      "step": 3672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538533175441033e-05,
+      "loss": 0.3826,
+      "step": 3673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382809486251436e-05,
+      "loss": 0.3413,
+      "step": 3674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380286735854903e-05,
+      "loss": 0.3977,
+      "step": 3675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5377763503446733e-05,
+      "loss": 0.3931,
+      "step": 3676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375239789252986e-05,
+      "loss": 0.3703,
+      "step": 3677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537271559349975e-05,
+      "loss": 0.3795,
+      "step": 3678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5370190916413172e-05,
+      "loss": 0.4008,
+      "step": 3679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536766575821943e-05,
+      "loss": 0.3883,
+      "step": 3680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365140119144748e-05,
+      "loss": 0.4196,
+      "step": 3681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362613999415397e-05,
+      "loss": 0.3573,
+      "step": 3682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360087399257687e-05,
+      "loss": 0.3881,
+      "step": 3683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357560318897968e-05,
+      "loss": 0.4066,
+      "step": 3684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535503275856264e-05,
+      "loss": 0.3865,
+      "step": 3685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535250471847815e-05,
+      "loss": 0.355,
+      "step": 3686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349976198870974e-05,
+      "loss": 0.3909,
+      "step": 3687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5347447199967636e-05,
+      "loss": 0.3716,
+      "step": 3688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344917721994715e-05,
+      "loss": 0.3816,
+      "step": 3689
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5342387765178812e-05,
+      "loss": 0.3731,
+      "step": 3690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5339857329746587e-05,
+      "loss": 0.3593,
+      "step": 3691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337326415924742e-05,
+      "loss": 0.3438,
+      "step": 3692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533479502394001e-05,
+      "loss": 0.3837,
+      "step": 3693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332263154019177e-05,
+      "loss": 0.4218,
+      "step": 3694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329730806389073e-05,
+      "loss": 0.3673,
+      "step": 3695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532719798127656e-05,
+      "loss": 0.4098,
+      "step": 3696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5324664678908556e-05,
+      "loss": 0.3678,
+      "step": 3697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5322130899512018e-05,
+      "loss": 0.3985,
+      "step": 3698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5319596643313935e-05,
+      "loss": 0.3663,
+      "step": 3699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531706191054135e-05,
+      "loss": 0.3716,
+      "step": 3700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531452670142135e-05,
+      "loss": 0.4026,
+      "step": 3701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531199101618106e-05,
+      "loss": 0.3833,
+      "step": 3702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309454855047645e-05,
+      "loss": 0.3939,
+      "step": 3703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306918218248315e-05,
+      "loss": 0.3816,
+      "step": 3704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304381106010324e-05,
+      "loss": 0.3978,
+      "step": 3705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530184351856097e-05,
+      "loss": 0.3703,
+      "step": 3706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299305456127592e-05,
+      "loss": 0.4133,
+      "step": 3707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529676691893757e-05,
+      "loss": 0.4467,
+      "step": 3708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5294227907218325e-05,
+      "loss": 0.3734,
+      "step": 3709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529168842119733e-05,
+      "loss": 0.4102,
+      "step": 3710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289148461102087e-05,
+      "loss": 0.3473,
+      "step": 3711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286608027160144e-05,
+      "loss": 0.4059,
+      "step": 3712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5284067119599105e-05,
+      "loss": 0.338,
+      "step": 3713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5281525738646602e-05,
+      "loss": 0.3509,
+      "step": 3714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527898388453031e-05,
+      "loss": 0.4047,
+      "step": 3715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5276441557477953e-05,
+      "loss": 0.3582,
+      "step": 3716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273898757717295e-05,
+      "loss": 0.3758,
+      "step": 3717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5271355485476142e-05,
+      "loss": 0.3996,
+      "step": 3718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268811740982334e-05,
+      "loss": 0.3983,
+      "step": 3719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526626752446377e-05,
+      "loss": 0.4183,
+      "step": 3720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526372283614838e-05,
+      "loss": 0.371,
+      "step": 3721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5261177676264135e-05,
+      "loss": 0.3603,
+      "step": 3722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5258632045039057e-05,
+      "loss": 0.4526,
+      "step": 3723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52560859427012e-05,
+      "loss": 0.3723,
+      "step": 3724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253539369478674e-05,
+      "loss": 0.3901,
+      "step": 3725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250992325599616e-05,
+      "loss": 0.3934,
+      "step": 3726
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5248444811292214e-05,
+      "loss": 0.4202,
+      "step": 3727
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245896826784689e-05,
+      "loss": 0.4059,
+      "step": 3728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243348372305322e-05,
+      "loss": 0.3956,
+      "step": 3729
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524079944808242e-05,
+      "loss": 0.3933,
+      "step": 3730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238250054344334e-05,
+      "loss": 0.4039,
+      "step": 3731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5235700191319466e-05,
+      "loss": 0.3871,
+      "step": 3732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5233149859236251e-05,
+      "loss": 0.3875,
+      "step": 3733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230599058323174e-05,
+      "loss": 0.3875,
+      "step": 3734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522804778880875e-05,
+      "loss": 0.3805,
+      "step": 3735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225496050921548e-05,
+      "loss": 0.3816,
+      "step": 3736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222943844890173e-05,
+      "loss": 0.3953,
+      "step": 3737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220391170943271e-05,
+      "loss": 0.4267,
+      "step": 3738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521783802930954e-05,
+      "loss": 0.4094,
+      "step": 3739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52152844202177e-05,
+      "loss": 0.3433,
+      "step": 3740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5212730343896533e-05,
+      "loss": 0.413,
+      "step": 3741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210175800574858e-05,
+      "loss": 0.3683,
+      "step": 3742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5207620790481526e-05,
+      "loss": 0.3736,
+      "step": 3743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205065313845433e-05,
+      "loss": 0.3831,
+      "step": 3744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202509370895528e-05,
+      "loss": 0.3733,
+      "step": 3745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199952961860795e-05,
+      "loss": 0.381,
+      "step": 3746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5197396086970253e-05,
+      "loss": 0.4031,
+      "step": 3747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194838746452972e-05,
+      "loss": 0.3865,
+      "step": 3748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192280940538058e-05,
+      "loss": 0.3414,
+      "step": 3749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189722669454661e-05,
+      "loss": 0.3859,
+      "step": 3750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187163933431976e-05,
+      "loss": 0.3822,
+      "step": 3751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518460473269923e-05,
+      "loss": 0.3964,
+      "step": 3752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182045067485705e-05,
+      "loss": 0.3918,
+      "step": 3753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5179484938020712e-05,
+      "loss": 0.4121,
+      "step": 3754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176924344533612e-05,
+      "loss": 0.3787,
+      "step": 3755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5174363287253804e-05,
+      "loss": 0.4306,
+      "step": 3756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171801766410727e-05,
+      "loss": 0.4069,
+      "step": 3757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169239782233868e-05,
+      "loss": 0.3919,
+      "step": 3758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516667733495275e-05,
+      "loss": 0.3975,
+      "step": 3759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164114424796936e-05,
+      "loss": 0.3187,
+      "step": 3760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161551051996032e-05,
+      "loss": 0.3738,
+      "step": 3761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158987216779695e-05,
+      "loss": 0.4135,
+      "step": 3762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156422919377607e-05,
+      "loss": 0.3618,
+      "step": 3763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5153858160019502e-05,
+      "loss": 0.3363,
+      "step": 3764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151292938935151e-05,
+      "loss": 0.3676,
+      "step": 3765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148727256354374e-05,
+      "loss": 0.4127,
+      "step": 3766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5146161112507022e-05,
+      "loss": 0.4022,
+      "step": 3767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514359450762299e-05,
+      "loss": 0.4439,
+      "step": 3768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5141027441932217e-05,
+      "loss": 0.3544,
+      "step": 3769
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513845991566469e-05,
+      "loss": 0.3643,
+      "step": 3770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513589192905042e-05,
+      "loss": 0.3947,
+      "step": 3771
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133323482319475e-05,
+      "loss": 0.3952,
+      "step": 3772
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130754575701953e-05,
+      "loss": 0.416,
+      "step": 3773
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128185209428004e-05,
+      "loss": 0.3853,
+      "step": 3774
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5125615383727811e-05,
+      "loss": 0.38,
+      "step": 3775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5123045098831601e-05,
+      "loss": 0.3401,
+      "step": 3776
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512047435496964e-05,
+      "loss": 0.3777,
+      "step": 3777
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511790315237224e-05,
+      "loss": 0.4055,
+      "step": 3778
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115331491269749e-05,
+      "loss": 0.4051,
+      "step": 3779
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511275937189256e-05,
+      "loss": 0.3266,
+      "step": 3780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110186794471105e-05,
+      "loss": 0.3906,
+      "step": 3781
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5107613759235858e-05,
+      "loss": 0.4061,
+      "step": 3782
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105040266417325e-05,
+      "loss": 0.3636,
+      "step": 3783
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102466316246073e-05,
+      "loss": 0.4433,
+      "step": 3784
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099891908952693e-05,
+      "loss": 0.409,
+      "step": 3785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097317044767823e-05,
+      "loss": 0.3736,
+      "step": 3786
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509474172392214e-05,
+      "loss": 0.4072,
+      "step": 3787
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5092165946646362e-05,
+      "loss": 0.4047,
+      "step": 3788
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089589713171254e-05,
+      "loss": 0.3823,
+      "step": 3789
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508701302372761e-05,
+      "loss": 0.3853,
+      "step": 3790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508443587854628e-05,
+      "loss": 0.4158,
+      "step": 3791
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5081858277858134e-05,
+      "loss": 0.3825,
+      "step": 3792
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079280221894108e-05,
+      "loss": 0.3685,
+      "step": 3793
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507670171088516e-05,
+      "loss": 0.3944,
+      "step": 3794
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5074122745062293e-05,
+      "loss": 0.3736,
+      "step": 3795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5071543324656562e-05,
+      "loss": 0.3882,
+      "step": 3796
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068963449899039e-05,
+      "loss": 0.3939,
+      "step": 3797
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066383121020864e-05,
+      "loss": 0.3848,
+      "step": 3798
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063802338253197e-05,
+      "loss": 0.3847,
+      "step": 3799
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061221101827248e-05,
+      "loss": 0.4162,
+      "step": 3800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505863941197427e-05,
+      "loss": 0.3653,
+      "step": 3801
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056057268925546e-05,
+      "loss": 0.3865,
+      "step": 3802
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053474672912414e-05,
+      "loss": 0.3565,
+      "step": 3803
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050891624166235e-05,
+      "loss": 0.398,
+      "step": 3804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504830812291843e-05,
+      "loss": 0.3517,
+      "step": 3805
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045724169400445e-05,
+      "loss": 0.331,
+      "step": 3806
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043139763843773e-05,
+      "loss": 0.3671,
+      "step": 3807
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504055490647995e-05,
+      "loss": 0.3776,
+      "step": 3808
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037969597540546e-05,
+      "loss": 0.3798,
+      "step": 3809
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035383837257178e-05,
+      "loss": 0.3749,
+      "step": 3810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032797625861496e-05,
+      "loss": 0.371,
+      "step": 3811
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.50302109635852e-05,
+      "loss": 0.3494,
+      "step": 3812
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502762385066002e-05,
+      "loss": 0.4245,
+      "step": 3813
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025036287317737e-05,
+      "loss": 0.4293,
+      "step": 3814
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022448273790163e-05,
+      "loss": 0.4452,
+      "step": 3815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5019859810309152e-05,
+      "loss": 0.3777,
+      "step": 3816
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017270897106607e-05,
+      "loss": 0.3881,
+      "step": 3817
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014681534414462e-05,
+      "loss": 0.4011,
+      "step": 3818
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501209172246469e-05,
+      "loss": 0.4159,
+      "step": 3819
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009501461489313e-05,
+      "loss": 0.3753,
+      "step": 3820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006910751720385e-05,
+      "loss": 0.3597,
+      "step": 3821
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004319593390007e-05,
+      "loss": 0.4137,
+      "step": 3822
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5001727986730316e-05,
+      "loss": 0.3529,
+      "step": 3823
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499913593197349e-05,
+      "loss": 0.3688,
+      "step": 3824
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4996543429351747e-05,
+      "loss": 0.3987,
+      "step": 3825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993950479097343e-05,
+      "loss": 0.3865,
+      "step": 3826
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499135708144258e-05,
+      "loss": 0.4321,
+      "step": 3827
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988763236619797e-05,
+      "loss": 0.4053,
+      "step": 3828
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986168944861366e-05,
+      "loss": 0.4222,
+      "step": 3829
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4983574206399714e-05,
+      "loss": 0.3817,
+      "step": 3830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4980979021467295e-05,
+      "loss": 0.3735,
+      "step": 3831
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978383390296611e-05,
+      "loss": 0.4047,
+      "step": 3832
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975787313120194e-05,
+      "loss": 0.4175,
+      "step": 3833
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973190790170628e-05,
+      "loss": 0.366,
+      "step": 3834
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970593821680532e-05,
+      "loss": 0.3858,
+      "step": 3835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967996407882562e-05,
+      "loss": 0.3618,
+      "step": 3836
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965398549009416e-05,
+      "loss": 0.3734,
+      "step": 3837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962800245293833e-05,
+      "loss": 0.4213,
+      "step": 3838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4960201496968596e-05,
+      "loss": 0.395,
+      "step": 3839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957602304266514e-05,
+      "loss": 0.3677,
+      "step": 3840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955002667420448e-05,
+      "loss": 0.4147,
+      "step": 3841
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4952402586663296e-05,
+      "loss": 0.3764,
+      "step": 3842
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4949802062227998e-05,
+      "loss": 0.389,
+      "step": 3843
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494720109434753e-05,
+      "loss": 0.4038,
+      "step": 3844
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944599683254903e-05,
+      "loss": 0.4022,
+      "step": 3845
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941997829183182e-05,
+      "loss": 0.4056,
+      "step": 3846
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4939395532365455e-05,
+      "loss": 0.3405,
+      "step": 3847
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936792793034865e-05,
+      "loss": 0.3666,
+      "step": 3848
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934189611424578e-05,
+      "loss": 0.3658,
+      "step": 3849
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931585987767822e-05,
+      "loss": 0.3834,
+      "step": 3850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928981922297842e-05,
+      "loss": 0.42,
+      "step": 3851
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926377415247935e-05,
+      "loss": 0.3611,
+      "step": 3852
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4923772466851434e-05,
+      "loss": 0.3854,
+      "step": 3853
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4921167077341715e-05,
+      "loss": 0.3412,
+      "step": 3854
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918561246952188e-05,
+      "loss": 0.4044,
+      "step": 3855
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915954975916305e-05,
+      "loss": 0.3691,
+      "step": 3856
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491334826446756e-05,
+      "loss": 0.3965,
+      "step": 3857
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910741112839481e-05,
+      "loss": 0.3619,
+      "step": 3858
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908133521265647e-05,
+      "loss": 0.395,
+      "step": 3859
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905525489979657e-05,
+      "loss": 0.3902,
+      "step": 3860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4902917019215167e-05,
+      "loss": 0.4021,
+      "step": 3861
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900308109205866e-05,
+      "loss": 0.3314,
+      "step": 3862
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489769876018548e-05,
+      "loss": 0.363,
+      "step": 3863
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895088972387777e-05,
+      "loss": 0.4038,
+      "step": 3864
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4892478746046562e-05,
+      "loss": 0.3952,
+      "step": 3865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889868081395689e-05,
+      "loss": 0.406,
+      "step": 3866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4887256978669033e-05,
+      "loss": 0.3767,
+      "step": 3867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884645438100524e-05,
+      "loss": 0.391,
+      "step": 3868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882033459924128e-05,
+      "loss": 0.3499,
+      "step": 3869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879421044373845e-05,
+      "loss": 0.3621,
+      "step": 3870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876808191683718e-05,
+      "loss": 0.3702,
+      "step": 3871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4874194902087828e-05,
+      "loss": 0.3735,
+      "step": 3872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871581175820296e-05,
+      "loss": 0.4032,
+      "step": 3873
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868967013115281e-05,
+      "loss": 0.3825,
+      "step": 3874
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4866352414206984e-05,
+      "loss": 0.3504,
+      "step": 3875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486373737932964e-05,
+      "loss": 0.3814,
+      "step": 3876
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861121908717529e-05,
+      "loss": 0.389,
+      "step": 3877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858506002604965e-05,
+      "loss": 0.375,
+      "step": 3878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4855889661226302e-05,
+      "loss": 0.3606,
+      "step": 3879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853272884815937e-05,
+      "loss": 0.3476,
+      "step": 3880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850655673608298e-05,
+      "loss": 0.3641,
+      "step": 3881
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848038027837866e-05,
+      "loss": 0.3778,
+      "step": 3882
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845419947739143e-05,
+      "loss": 0.4388,
+      "step": 3883
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4842801433546683e-05,
+      "loss": 0.4113,
+      "step": 3884
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840182485495072e-05,
+      "loss": 0.3531,
+      "step": 3885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837563103818942e-05,
+      "loss": 0.3673,
+      "step": 3886
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834943288752954e-05,
+      "loss": 0.404,
+      "step": 3887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832323040531817e-05,
+      "loss": 0.4337,
+      "step": 3888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829702359390274e-05,
+      "loss": 0.3934,
+      "step": 3889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827081245563107e-05,
+      "loss": 0.3716,
+      "step": 3890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4824459699285139e-05,
+      "loss": 0.4172,
+      "step": 3891
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482183772079123e-05,
+      "loss": 0.3978,
+      "step": 3892
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819215310316278e-05,
+      "loss": 0.3796,
+      "step": 3893
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4816592468095219e-05,
+      "loss": 0.3428,
+      "step": 3894
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813969194363034e-05,
+      "loss": 0.3971,
+      "step": 3895
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811345489354737e-05,
+      "loss": 0.3832,
+      "step": 3896
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808721353305376e-05,
+      "loss": 0.4097,
+      "step": 3897
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806096786450052e-05,
+      "loss": 0.3725,
+      "step": 3898
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4803471789023891e-05,
+      "loss": 0.3512,
+      "step": 3899
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4800846361262062e-05,
+      "loss": 0.3968,
+      "step": 3900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4798220503399775e-05,
+      "loss": 0.3949,
+      "step": 3901
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795594215672272e-05,
+      "loss": 0.3889,
+      "step": 3902
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792967498314848e-05,
+      "loss": 0.3532,
+      "step": 3903
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790340351562818e-05,
+      "loss": 0.3687,
+      "step": 3904
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4787712775651546e-05,
+      "loss": 0.3712,
+      "step": 3905
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785084770816433e-05,
+      "loss": 0.3932,
+      "step": 3906
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782456337292922e-05,
+      "loss": 0.3619,
+      "step": 3907
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477982747531648e-05,
+      "loss": 0.4083,
+      "step": 3908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477719818512263e-05,
+      "loss": 0.3666,
+      "step": 3909
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4774568466946927e-05,
+      "loss": 0.381,
+      "step": 3910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477193832102496e-05,
+      "loss": 0.4441,
+      "step": 3911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769307747592365e-05,
+      "loss": 0.4109,
+      "step": 3912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766676746884803e-05,
+      "loss": 0.3176,
+      "step": 3913
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764045319137987e-05,
+      "loss": 0.391,
+      "step": 3914
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476141346458766e-05,
+      "loss": 0.407,
+      "step": 3915
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758781183469611e-05,
+      "loss": 0.4049,
+      "step": 3916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756148476019654e-05,
+      "loss": 0.4091,
+      "step": 3917
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753515342473652e-05,
+      "loss": 0.3863,
+      "step": 3918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750881783067507e-05,
+      "loss": 0.3935,
+      "step": 3919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748247798037154e-05,
+      "loss": 0.3786,
+      "step": 3920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4745613387618566e-05,
+      "loss": 0.4062,
+      "step": 3921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742978552047756e-05,
+      "loss": 0.3973,
+      "step": 3922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4740343291560776e-05,
+      "loss": 0.3856,
+      "step": 3923
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4737707606393715e-05,
+      "loss": 0.3909,
+      "step": 3924
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735071496782699e-05,
+      "loss": 0.413,
+      "step": 3925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732434962963891e-05,
+      "loss": 0.3623,
+      "step": 3926
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47297980051735e-05,
+      "loss": 0.3886,
+      "step": 3927
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727160623647763e-05,
+      "loss": 0.4196,
+      "step": 3928
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724522818622957e-05,
+      "loss": 0.3583,
+      "step": 3929
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721884590335403e-05,
+      "loss": 0.4179,
+      "step": 3930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719245939021453e-05,
+      "loss": 0.3974,
+      "step": 3931
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716606864917501e-05,
+      "loss": 0.4072,
+      "step": 3932
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.3846,
+      "step": 3933
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471132744928535e-05,
+      "loss": 0.3891,
+      "step": 3934
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470868710823013e-05,
+      "loss": 0.3872,
+      "step": 3935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706046345330854e-05,
+      "loss": 0.3788,
+      "step": 3936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703405160824107e-05,
+      "loss": 0.3739,
+      "step": 3937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4700763554946509e-05,
+      "loss": 0.3851,
+      "step": 3938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469812152793472e-05,
+      "loss": 0.4233,
+      "step": 3939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695479080025428e-05,
+      "loss": 0.368,
+      "step": 3940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4692836211455373e-05,
+      "loss": 0.3711,
+      "step": 3941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690192922461321e-05,
+      "loss": 0.3902,
+      "step": 3942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687549213280084e-05,
+      "loss": 0.3442,
+      "step": 3943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684905084148506e-05,
+      "loss": 0.4101,
+      "step": 3944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468226053530347e-05,
+      "loss": 0.4419,
+      "step": 3945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4679615566981895e-05,
+      "loss": 0.4031,
+      "step": 3946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676970179420745e-05,
+      "loss": 0.3639,
+      "step": 3947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674324372857012e-05,
+      "loss": 0.3926,
+      "step": 3948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4671678147527732e-05,
+      "loss": 0.4021,
+      "step": 3949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4669031503669971e-05,
+      "loss": 0.3694,
+      "step": 3950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666384441520847e-05,
+      "loss": 0.4349,
+      "step": 3951
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46637369613175e-05,
+      "loss": 0.4268,
+      "step": 3952
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4661089063297113e-05,
+      "loss": 0.3975,
+      "step": 3953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658440747696912e-05,
+      "loss": 0.3484,
+      "step": 3954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655792014754153e-05,
+      "loss": 0.4222,
+      "step": 3955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653142864706129e-05,
+      "loss": 0.369,
+      "step": 3956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4650493297790178e-05,
+      "loss": 0.4104,
+      "step": 3957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647843314243667e-05,
+      "loss": 0.4276,
+      "step": 3958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645192914304006e-05,
+      "loss": 0.4046,
+      "step": 3959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642542098208641e-05,
+      "loss": 0.4015,
+      "step": 3960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463989086619505e-05,
+      "loss": 0.4279,
+      "step": 3961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4637239218500759e-05,
+      "loss": 0.4088,
+      "step": 3962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463458715536332e-05,
+      "loss": 0.3898,
+      "step": 3963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4631934677020332e-05,
+      "loss": 0.3721,
+      "step": 3964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629281783709422e-05,
+      "loss": 0.3562,
+      "step": 3965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626628475668256e-05,
+      "loss": 0.4377,
+      "step": 3966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462397475313455e-05,
+      "loss": 0.3892,
+      "step": 3967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621320616346041e-05,
+      "loss": 0.3977,
+      "step": 3968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618666065540512e-05,
+      "loss": 0.3886,
+      "step": 3969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616011100955772e-05,
+      "loss": 0.4186,
+      "step": 3970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4613355722829682e-05,
+      "loss": 0.3717,
+      "step": 3971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610699931400132e-05,
+      "loss": 0.3526,
+      "step": 3972
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460804372690505e-05,
+      "loss": 0.3843,
+      "step": 3973
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4605387109582401e-05,
+      "loss": 0.3915,
+      "step": 3974
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602730079670188e-05,
+      "loss": 0.3401,
+      "step": 3975
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4600072637406453e-05,
+      "loss": 0.3739,
+      "step": 3976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597414783029264e-05,
+      "loss": 0.3738,
+      "step": 3977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594756516776742e-05,
+      "loss": 0.4137,
+      "step": 3978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592097838887034e-05,
+      "loss": 0.3807,
+      "step": 3979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589438749598327e-05,
+      "loss": 0.4078,
+      "step": 3980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586779249148844e-05,
+      "loss": 0.3948,
+      "step": 3981
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584119337776848e-05,
+      "loss": 0.3769,
+      "step": 3982
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581459015720632e-05,
+      "loss": 0.4059,
+      "step": 3983
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578798283218534e-05,
+      "loss": 0.3625,
+      "step": 3984
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576137140508926e-05,
+      "loss": 0.3652,
+      "step": 3985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4573475587830208e-05,
+      "loss": 0.3986,
+      "step": 3986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570813625420836e-05,
+      "loss": 0.415,
+      "step": 3987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568151253519282e-05,
+      "loss": 0.3828,
+      "step": 3988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4565488472364066e-05,
+      "loss": 0.4008,
+      "step": 3989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4562825282193745e-05,
+      "loss": 0.3635,
+      "step": 3990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456016168324691e-05,
+      "loss": 0.3924,
+      "step": 3991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557497675762187e-05,
+      "loss": 0.3644,
+      "step": 3992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554833259978239e-05,
+      "loss": 0.3857,
+      "step": 3993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552168436133766e-05,
+      "loss": 0.321,
+      "step": 3994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4549503204467513e-05,
+      "loss": 0.4336,
+      "step": 3995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546837565218246e-05,
+      "loss": 0.4081,
+      "step": 3996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544171518624778e-05,
+      "loss": 0.4271,
+      "step": 3997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541505064925958e-05,
+      "loss": 0.3737,
+      "step": 3998
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538838204360665e-05,
+      "loss": 0.3851,
+      "step": 3999
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536170937167827e-05,
+      "loss": 0.394,
+      "step": 4000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533503263586393e-05,
+      "loss": 0.3976,
+      "step": 4001
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4530835183855356e-05,
+      "loss": 0.3682,
+      "step": 4002
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452816669821375e-05,
+      "loss": 0.4052,
+      "step": 4003
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525497806900637e-05,
+      "loss": 0.3595,
+      "step": 4004
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522828510155123e-05,
+      "loss": 0.4247,
+      "step": 4005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520158808216336e-05,
+      "loss": 0.4199,
+      "step": 4006
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4517488701323461e-05,
+      "loss": 0.4334,
+      "step": 4007
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514818189715707e-05,
+      "loss": 0.3773,
+      "step": 4008
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512147273632318e-05,
+      "loss": 0.3612,
+      "step": 4009
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4509475953312577e-05,
+      "loss": 0.4004,
+      "step": 4010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4506804228995805e-05,
+      "loss": 0.3672,
+      "step": 4011
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450413210092136e-05,
+      "loss": 0.3851,
+      "step": 4012
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501459569328631e-05,
+      "loss": 0.3429,
+      "step": 4013
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498786634457045e-05,
+      "loss": 0.3799,
+      "step": 4014
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496113296546068e-05,
+      "loss": 0.3709,
+      "step": 4015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493439555835198e-05,
+      "loss": 0.3389,
+      "step": 4016
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490765412563978e-05,
+      "loss": 0.3792,
+      "step": 4017
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488090866971971e-05,
+      "loss": 0.391,
+      "step": 4018
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485415919298793e-05,
+      "loss": 0.3979,
+      "step": 4019
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482740569784084e-05,
+      "loss": 0.3742,
+      "step": 4020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448006481866753e-05,
+      "loss": 0.411,
+      "step": 4021
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477388666188837e-05,
+      "loss": 0.3651,
+      "step": 4022
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474712112587768e-05,
+      "loss": 0.335,
+      "step": 4023
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472035158104106e-05,
+      "loss": 0.366,
+      "step": 4024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469357802977674e-05,
+      "loss": 0.3961,
+      "step": 4025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466680047448338e-05,
+      "loss": 0.3849,
+      "step": 4026
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446400189175599e-05,
+      "loss": 0.4025,
+      "step": 4027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461323336140562e-05,
+      "loss": 0.3956,
+      "step": 4028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458644380842021e-05,
+      "loss": 0.4156,
+      "step": 4029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455965026100374e-05,
+      "loss": 0.3499,
+      "step": 4030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4453285272155655e-05,
+      "loss": 0.3904,
+      "step": 4031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450605119247942e-05,
+      "loss": 0.3678,
+      "step": 4032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4447924567617349e-05,
+      "loss": 0.3839,
+      "step": 4033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4445243617504015e-05,
+      "loss": 0.4166,
+      "step": 4034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442562269148128e-05,
+      "loss": 0.3657,
+      "step": 4035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439880522789905e-05,
+      "loss": 0.3997,
+      "step": 4036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437198378669598e-05,
+      "loss": 0.3519,
+      "step": 4037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4434515837027496e-05,
+      "loss": 0.383,
+      "step": 4038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4431832898103929e-05,
+      "loss": 0.3981,
+      "step": 4039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442914956213925e-05,
+      "loss": 0.3938,
+      "step": 4040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426465829373856e-05,
+      "loss": 0.3871,
+      "step": 4041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423781700048184e-05,
+      "loss": 0.4289,
+      "step": 4042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421097174402696e-05,
+      "loss": 0.3883,
+      "step": 4043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4418412252677897e-05,
+      "loss": 0.3467,
+      "step": 4044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415726935114324e-05,
+      "loss": 0.3619,
+      "step": 4045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441304122195255e-05,
+      "loss": 0.4159,
+      "step": 4046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410355113433187e-05,
+      "loss": 0.4086,
+      "step": 4047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407668609796874e-05,
+      "loss": 0.3478,
+      "step": 4048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4404981711284296e-05,
+      "loss": 0.4087,
+      "step": 4049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402294418136164e-05,
+      "loss": 0.3636,
+      "step": 4050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4399606730593232e-05,
+      "loss": 0.3644,
+      "step": 4051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396918648896283e-05,
+      "loss": 0.3708,
+      "step": 4052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4394230173286138e-05,
+      "loss": 0.41,
+      "step": 4053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4391541304003655e-05,
+      "loss": 0.3771,
+      "step": 4054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388852041289726e-05,
+      "loss": 0.3923,
+      "step": 4055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4386162385385279e-05,
+      "loss": 0.3517,
+      "step": 4056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438347233653127e-05,
+      "loss": 0.4244,
+      "step": 4057
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4380781894968703e-05,
+      "loss": 0.4044,
+      "step": 4058
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4378091060938609e-05,
+      "loss": 0.3722,
+      "step": 4059
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375399834682055e-05,
+      "loss": 0.4028,
+      "step": 4060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372708216440143e-05,
+      "loss": 0.3892,
+      "step": 4061
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370016206454012e-05,
+      "loss": 0.4155,
+      "step": 4062
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4367323804964832e-05,
+      "loss": 0.3702,
+      "step": 4063
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364631012213818e-05,
+      "loss": 0.403,
+      "step": 4064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361937828442207e-05,
+      "loss": 0.3832,
+      "step": 4065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435924425389128e-05,
+      "loss": 0.3835,
+      "step": 4066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356550288802347e-05,
+      "loss": 0.388,
+      "step": 4067
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435385593341676e-05,
+      "loss": 0.3787,
+      "step": 4068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4351161187975904e-05,
+      "loss": 0.3781,
+      "step": 4069
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348466052721186e-05,
+      "loss": 0.3737,
+      "step": 4070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345770527894074e-05,
+      "loss": 0.3995,
+      "step": 4071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343074613736047e-05,
+      "loss": 0.386,
+      "step": 4072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340378310488628e-05,
+      "loss": 0.3896,
+      "step": 4073
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337681618393377e-05,
+      "loss": 0.3101,
+      "step": 4074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4334984537691887e-05,
+      "loss": 0.4137,
+      "step": 4075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332287068625785e-05,
+      "loss": 0.3813,
+      "step": 4076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329589211436733e-05,
+      "loss": 0.3762,
+      "step": 4077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326890966366423e-05,
+      "loss": 0.3969,
+      "step": 4078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324192333656592e-05,
+      "loss": 0.3544,
+      "step": 4079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432149331354901e-05,
+      "loss": 0.4141,
+      "step": 4080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318793906285467e-05,
+      "loss": 0.3915,
+      "step": 4081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316094112107806e-05,
+      "loss": 0.3838,
+      "step": 4082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313393931257897e-05,
+      "loss": 0.414,
+      "step": 4083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310693363977646e-05,
+      "loss": 0.415,
+      "step": 4084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430799241050899e-05,
+      "loss": 0.3906,
+      "step": 4085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305291071093905e-05,
+      "loss": 0.3922,
+      "step": 4086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302589345974398e-05,
+      "loss": 0.4021,
+      "step": 4087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4299887235392516e-05,
+      "loss": 0.3533,
+      "step": 4088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429718473959033e-05,
+      "loss": 0.3819,
+      "step": 4089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294481858809961e-05,
+      "loss": 0.3766,
+      "step": 4090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4291778593293551e-05,
+      "loss": 0.3904,
+      "step": 4091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289074943283284e-05,
+      "loss": 0.3728,
+      "step": 4092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286370909021372e-05,
+      "loss": 0.3885,
+      "step": 4093
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4283666490750068e-05,
+      "loss": 0.3916,
+      "step": 4094
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280961688711657e-05,
+      "loss": 0.3436,
+      "step": 4095
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427825650314846e-05,
+      "loss": 0.3838,
+      "step": 4096
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275550934302822e-05,
+      "loss": 0.3818,
+      "step": 4097
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427284498241714e-05,
+      "loss": 0.4113,
+      "step": 4098
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270138647733831e-05,
+      "loss": 0.4218,
+      "step": 4099
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4267431930495352e-05,
+      "loss": 0.3918,
+      "step": 4100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264724830944198e-05,
+      "loss": 0.3973,
+      "step": 4101
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4262017349322889e-05,
+      "loss": 0.3695,
+      "step": 4102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259309485873986e-05,
+      "loss": 0.3727,
+      "step": 4103
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425660124084008e-05,
+      "loss": 0.3799,
+      "step": 4104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253892614463804e-05,
+      "loss": 0.3986,
+      "step": 4105
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4251183606987813e-05,
+      "loss": 0.3945,
+      "step": 4106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248474218654804e-05,
+      "loss": 0.3826,
+      "step": 4107
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245764449707515e-05,
+      "loss": 0.3839,
+      "step": 4108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4243054300388698e-05,
+      "loss": 0.4073,
+      "step": 4109
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240343770941161e-05,
+      "loss": 0.3688,
+      "step": 4110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237632861607731e-05,
+      "loss": 0.3782,
+      "step": 4111
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4234921572631275e-05,
+      "loss": 0.4003,
+      "step": 4112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232209904254695e-05,
+      "loss": 0.3499,
+      "step": 4113
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229497856720922e-05,
+      "loss": 0.3944,
+      "step": 4114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226785430272926e-05,
+      "loss": 0.4252,
+      "step": 4115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422407262515371e-05,
+      "loss": 0.3725,
+      "step": 4116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221359441606311e-05,
+      "loss": 0.3513,
+      "step": 4117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4218645879873796e-05,
+      "loss": 0.3996,
+      "step": 4118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215931940199269e-05,
+      "loss": 0.3731,
+      "step": 4119
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213217622825868e-05,
+      "loss": 0.3924,
+      "step": 4120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210502927996767e-05,
+      "loss": 0.4148,
+      "step": 4121
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4207787855955168e-05,
+      "loss": 0.3872,
+      "step": 4122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205072406944312e-05,
+      "loss": 0.3814,
+      "step": 4123
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202356581207472e-05,
+      "loss": 0.3738,
+      "step": 4124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199640378987955e-05,
+      "loss": 0.3549,
+      "step": 4125
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.41969238005291e-05,
+      "loss": 0.4127,
+      "step": 4126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419420684607428e-05,
+      "loss": 0.3738,
+      "step": 4127
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419148951586691e-05,
+      "loss": 0.4048,
+      "step": 4128
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4188771810150422e-05,
+      "loss": 0.3953,
+      "step": 4129
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4186053729168295e-05,
+      "loss": 0.3863,
+      "step": 4130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4183335273164035e-05,
+      "loss": 0.3646,
+      "step": 4131
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180616442381191e-05,
+      "loss": 0.4037,
+      "step": 4132
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177897237063336e-05,
+      "loss": 0.3832,
+      "step": 4133
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175177657454075e-05,
+      "loss": 0.3864,
+      "step": 4134
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172457703797058e-05,
+      "loss": 0.3537,
+      "step": 4135
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169737376335955e-05,
+      "loss": 0.408,
+      "step": 4136
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416701667531448e-05,
+      "loss": 0.3571,
+      "step": 4137
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4164295600976375e-05,
+      "loss": 0.3977,
+      "step": 4138
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161574153565414e-05,
+      "loss": 0.3708,
+      "step": 4139
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158852333325415e-05,
+      "loss": 0.3602,
+      "step": 4140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156130140500213e-05,
+      "loss": 0.3651,
+      "step": 4141
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4153407575333695e-05,
+      "loss": 0.4031,
+      "step": 4142
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4150684638069758e-05,
+      "loss": 0.4381,
+      "step": 4143
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414796132895236e-05,
+      "loss": 0.3791,
+      "step": 4144
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4145237648225467e-05,
+      "loss": 0.3782,
+      "step": 4145
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4142513596133095e-05,
+      "loss": 0.381,
+      "step": 4146
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139789172919287e-05,
+      "loss": 0.3669,
+      "step": 4147
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413706437882812e-05,
+      "loss": 0.4173,
+      "step": 4148
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134339214103703e-05,
+      "loss": 0.3786,
+      "step": 4149
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131613678990182e-05,
+      "loss": 0.4007,
+      "step": 4150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128887773731728e-05,
+      "loss": 0.3877,
+      "step": 4151
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126161498572555e-05,
+      "loss": 0.3521,
+      "step": 4152
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123434853756906e-05,
+      "loss": 0.4159,
+      "step": 4153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120707839529053e-05,
+      "loss": 0.3532,
+      "step": 4154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117980456133306e-05,
+      "loss": 0.378,
+      "step": 4155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115252703814015e-05,
+      "loss": 0.4233,
+      "step": 4156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112524582815546e-05,
+      "loss": 0.3939,
+      "step": 4157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410979609338231e-05,
+      "loss": 0.3771,
+      "step": 4158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107067235758747e-05,
+      "loss": 0.3845,
+      "step": 4159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104338010189336e-05,
+      "loss": 0.3986,
+      "step": 4160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101608416918579e-05,
+      "loss": 0.3531,
+      "step": 4161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098878456191017e-05,
+      "loss": 0.4227,
+      "step": 4162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096148128251224e-05,
+      "loss": 0.3647,
+      "step": 4163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093417433343804e-05,
+      "loss": 0.3958,
+      "step": 4164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4090686371713403e-05,
+      "loss": 0.3781,
+      "step": 4165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408795494360468e-05,
+      "loss": 0.4383,
+      "step": 4166
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408522314926235e-05,
+      "loss": 0.3642,
+      "step": 4167
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4082490988931146e-05,
+      "loss": 0.3766,
+      "step": 4168
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079758462855838e-05,
+      "loss": 0.3646,
+      "step": 4169
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077025571281233e-05,
+      "loss": 0.3704,
+      "step": 4170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074292314452159e-05,
+      "loss": 0.3824,
+      "step": 4171
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071558692613489e-05,
+      "loss": 0.3812,
+      "step": 4172
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068824706010124e-05,
+      "loss": 0.4003,
+      "step": 4173
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066090354887001e-05,
+      "loss": 0.3633,
+      "step": 4174
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063355639489076e-05,
+      "loss": 0.3921,
+      "step": 4175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4060620560061358e-05,
+      "loss": 0.4419,
+      "step": 4176
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057885116848873e-05,
+      "loss": 0.4049,
+      "step": 4177
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055149310096691e-05,
+      "loss": 0.4158,
+      "step": 4178
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4052413140049898e-05,
+      "loss": 0.3538,
+      "step": 4179
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049676606953635e-05,
+      "loss": 0.3824,
+      "step": 4180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404693971105306e-05,
+      "loss": 0.409,
+      "step": 4181
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044202452593364e-05,
+      "loss": 0.3991,
+      "step": 4182
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041464831819776e-05,
+      "loss": 0.345,
+      "step": 4183
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4038726848977553e-05,
+      "loss": 0.3663,
+      "step": 4184
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035988504311995e-05,
+      "loss": 0.3881,
+      "step": 4185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4033249798068416e-05,
+      "loss": 0.3686,
+      "step": 4186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030510730492176e-05,
+      "loss": 0.4344,
+      "step": 4187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027771301828663e-05,
+      "loss": 0.3835,
+      "step": 4188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4025031512323301e-05,
+      "loss": 0.3631,
+      "step": 4189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022291362221546e-05,
+      "loss": 0.386,
+      "step": 4190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019550851768873e-05,
+      "loss": 0.3745,
+      "step": 4191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016809981210812e-05,
+      "loss": 0.3932,
+      "step": 4192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4014068750792906e-05,
+      "loss": 0.4063,
+      "step": 4193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4011327160760744e-05,
+      "loss": 0.3307,
+      "step": 4194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.400858521135993e-05,
+      "loss": 0.4235,
+      "step": 4195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005842902836122e-05,
+      "loss": 0.3804,
+      "step": 4196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4003100235434998e-05,
+      "loss": 0.3902,
+      "step": 4197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000357209402262e-05,
+      "loss": 0.409,
+      "step": 4198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997613824983665e-05,
+      "loss": 0.4052,
+      "step": 4199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3994870082424976e-05,
+      "loss": 0.3927,
+      "step": 4200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399212598197201e-05,
+      "loss": 0.3866,
+      "step": 4201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989381523870603e-05,
+      "loss": 0.3538,
+      "step": 4202
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986636708366623e-05,
+      "loss": 0.4041,
+      "step": 4203
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398389153570598e-05,
+      "loss": 0.3456,
+      "step": 4204
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981146006134609e-05,
+      "loss": 0.3922,
+      "step": 4205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3978400119898476e-05,
+      "loss": 0.3846,
+      "step": 4206
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3975653877243577e-05,
+      "loss": 0.3647,
+      "step": 4207
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972907278415952e-05,
+      "loss": 0.3985,
+      "step": 4208
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397016032366166e-05,
+      "loss": 0.3851,
+      "step": 4209
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967413013226796e-05,
+      "loss": 0.379,
+      "step": 4210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964665347357484e-05,
+      "loss": 0.377,
+      "step": 4211
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3961917326299893e-05,
+      "loss": 0.4013,
+      "step": 4212
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959168950300207e-05,
+      "loss": 0.3584,
+      "step": 4213
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956420219604647e-05,
+      "loss": 0.369,
+      "step": 4214
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953671134459473e-05,
+      "loss": 0.3801,
+      "step": 4215
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950921695110966e-05,
+      "loss": 0.376,
+      "step": 4216
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948171901805452e-05,
+      "loss": 0.3506,
+      "step": 4217
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394542175478927e-05,
+      "loss": 0.3723,
+      "step": 4218
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394267125430881e-05,
+      "loss": 0.3844,
+      "step": 4219
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939920400610483e-05,
+      "loss": 0.4116,
+      "step": 4220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3937169193940729e-05,
+      "loss": 0.3949,
+      "step": 4221
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934417634546033e-05,
+      "loss": 0.3959,
+      "step": 4222
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931665722672898e-05,
+      "loss": 0.4039,
+      "step": 4223
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928913458567861e-05,
+      "loss": 0.3326,
+      "step": 4224
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926160842477499e-05,
+      "loss": 0.4071,
+      "step": 4225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3923407874648412e-05,
+      "loss": 0.409,
+      "step": 4226
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920654555327229e-05,
+      "loss": 0.3632,
+      "step": 4227
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917900884760624e-05,
+      "loss": 0.4091,
+      "step": 4228
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915146863195292e-05,
+      "loss": 0.3969,
+      "step": 4229
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912392490877957e-05,
+      "loss": 0.3818,
+      "step": 4230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909637768055385e-05,
+      "loss": 0.401,
+      "step": 4231
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906882694974361e-05,
+      "loss": 0.3746,
+      "step": 4232
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904127271881716e-05,
+      "loss": 0.4147,
+      "step": 4233
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3901371499024297e-05,
+      "loss": 0.3504,
+      "step": 4234
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898615376648992e-05,
+      "loss": 0.3852,
+      "step": 4235
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895858905002715e-05,
+      "loss": 0.3962,
+      "step": 4236
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389310208433242e-05,
+      "loss": 0.3588,
+      "step": 4237
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3890344914885084e-05,
+      "loss": 0.3665,
+      "step": 4238
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3887587396907716e-05,
+      "loss": 0.4039,
+      "step": 4239
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3884829530647355e-05,
+      "loss": 0.389,
+      "step": 4240
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388207131635108e-05,
+      "loss": 0.3655,
+      "step": 4241
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3879312754265994e-05,
+      "loss": 0.3782,
+      "step": 4242
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876553844639229e-05,
+      "loss": 0.3824,
+      "step": 4243
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873794587717955e-05,
+      "loss": 0.3942,
+      "step": 4244
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3871034983749367e-05,
+      "loss": 0.3882,
+      "step": 4245
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3868275032980694e-05,
+      "loss": 0.3469,
+      "step": 4246
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865514735659199e-05,
+      "loss": 0.3385,
+      "step": 4247
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862754092032167e-05,
+      "loss": 0.3758,
+      "step": 4248
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3859993102346927e-05,
+      "loss": 0.3922,
+      "step": 4249
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857231766850828e-05,
+      "loss": 0.3667,
+      "step": 4250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385447008579125e-05,
+      "loss": 0.3865,
+      "step": 4251
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851708059415614e-05,
+      "loss": 0.3798,
+      "step": 4252
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848945687971363e-05,
+      "loss": 0.3595,
+      "step": 4253
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846182971705976e-05,
+      "loss": 0.3942,
+      "step": 4254
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3843419910866953e-05,
+      "loss": 0.4336,
+      "step": 4255
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840656505701842e-05,
+      "loss": 0.3957,
+      "step": 4256
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837892756458206e-05,
+      "loss": 0.4016,
+      "step": 4257
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383512866338365e-05,
+      "loss": 0.4209,
+      "step": 4258
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832364226725798e-05,
+      "loss": 0.3738,
+      "step": 4259
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829599446732318e-05,
+      "loss": 0.4006,
+      "step": 4260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.381,
+      "step": 4261
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3824068857729265e-05,
+      "loss": 0.4059,
+      "step": 4262
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821303049215171e-05,
+      "loss": 0.3733,
+      "step": 4263
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818536898356398e-05,
+      "loss": 0.3806,
+      "step": 4264
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815770405400769e-05,
+      "loss": 0.3674,
+      "step": 4265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.381300357059612e-05,
+      "loss": 0.3712,
+      "step": 4266
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810236394190333e-05,
+      "loss": 0.4172,
+      "step": 4267
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807468876431313e-05,
+      "loss": 0.3768,
+      "step": 4268
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804701017567004e-05,
+      "loss": 0.3652,
+      "step": 4269
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801932817845364e-05,
+      "loss": 0.385,
+      "step": 4270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799164277514397e-05,
+      "loss": 0.4219,
+      "step": 4271
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796395396822134e-05,
+      "loss": 0.3376,
+      "step": 4272
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379362617601663e-05,
+      "loss": 0.386,
+      "step": 4273
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3790856615345982e-05,
+      "loss": 0.376,
+      "step": 4274
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3788086715058305e-05,
+      "loss": 0.3365,
+      "step": 4275
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3785316475401752e-05,
+      "loss": 0.3567,
+      "step": 4276
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782545896624502e-05,
+      "loss": 0.3975,
+      "step": 4277
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3779774978974772e-05,
+      "loss": 0.3759,
+      "step": 4278
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777003722700801e-05,
+      "loss": 0.3808,
+      "step": 4279
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377423212805086e-05,
+      "loss": 0.4018,
+      "step": 4280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3771460195273259e-05,
+      "loss": 0.3306,
+      "step": 4281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768687924616322e-05,
+      "loss": 0.394,
+      "step": 4282
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3765915316328417e-05,
+      "loss": 0.3832,
+      "step": 4283
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763142370657939e-05,
+      "loss": 0.3654,
+      "step": 4284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760369087853312e-05,
+      "loss": 0.3821,
+      "step": 4285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757595468162986e-05,
+      "loss": 0.37,
+      "step": 4286
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3754821511835447e-05,
+      "loss": 0.3682,
+      "step": 4287
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3752047219119214e-05,
+      "loss": 0.3681,
+      "step": 4288
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749272590262823e-05,
+      "loss": 0.3717,
+      "step": 4289
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746497625514858e-05,
+      "loss": 0.3563,
+      "step": 4290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374372232512392e-05,
+      "loss": 0.3869,
+      "step": 4291
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740946689338642e-05,
+      "loss": 0.3871,
+      "step": 4292
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3738170718407689e-05,
+      "loss": 0.3681,
+      "step": 4293
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373539441257976e-05,
+      "loss": 0.3663,
+      "step": 4294
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732617772103575e-05,
+      "loss": 0.4086,
+      "step": 4295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372984079722789e-05,
+      "loss": 0.3722,
+      "step": 4296
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727063488201498e-05,
+      "loss": 0.4136,
+      "step": 4297
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3724285845273201e-05,
+      "loss": 0.3879,
+      "step": 4298
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721507868691852e-05,
+      "loss": 0.3254,
+      "step": 4299
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3718729558706322e-05,
+      "loss": 0.3824,
+      "step": 4300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3715950915565521e-05,
+      "loss": 0.3823,
+      "step": 4301
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3713171939518378e-05,
+      "loss": 0.3681,
+      "step": 4302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710392630813859e-05,
+      "loss": 0.4074,
+      "step": 4303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707612989700953e-05,
+      "loss": 0.3663,
+      "step": 4304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704833016428692e-05,
+      "loss": 0.3669,
+      "step": 4305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702052711246128e-05,
+      "loss": 0.4085,
+      "step": 4306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699272074402339e-05,
+      "loss": 0.3885,
+      "step": 4307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696491106146444e-05,
+      "loss": 0.3888,
+      "step": 4308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369370980672758e-05,
+      "loss": 0.3551,
+      "step": 4309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3690928176394924e-05,
+      "loss": 0.3932,
+      "step": 4310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3688146215397678e-05,
+      "loss": 0.3654,
+      "step": 4311
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685363923985064e-05,
+      "loss": 0.3731,
+      "step": 4312
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682581302406361e-05,
+      "loss": 0.3499,
+      "step": 4313
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679798350910844e-05,
+      "loss": 0.3946,
+      "step": 4314
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677015069747843e-05,
+      "loss": 0.3788,
+      "step": 4315
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36742314591667e-05,
+      "loss": 0.396,
+      "step": 4316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3671447519416803e-05,
+      "loss": 0.3463,
+      "step": 4317
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668663250747554e-05,
+      "loss": 0.3903,
+      "step": 4318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665878653408394e-05,
+      "loss": 0.4119,
+      "step": 4319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366309372764879e-05,
+      "loss": 0.3049,
+      "step": 4320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366030847371824e-05,
+      "loss": 0.3926,
+      "step": 4321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657522891866274e-05,
+      "loss": 0.3416,
+      "step": 4322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365473698234244e-05,
+      "loss": 0.3904,
+      "step": 4323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651950745396328e-05,
+      "loss": 0.3841,
+      "step": 4324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649164181277554e-05,
+      "loss": 0.3859,
+      "step": 4325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646377290235762e-05,
+      "loss": 0.3706,
+      "step": 4326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364359007252062e-05,
+      "loss": 0.411,
+      "step": 4327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640802528381836e-05,
+      "loss": 0.3621,
+      "step": 4328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3638014658069143e-05,
+      "loss": 0.3821,
+      "step": 4329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635226461832296e-05,
+      "loss": 0.4077,
+      "step": 4330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363243793992109e-05,
+      "loss": 0.3848,
+      "step": 4331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362964909258534e-05,
+      "loss": 0.3689,
+      "step": 4332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626859920074902e-05,
+      "loss": 0.3778,
+      "step": 4333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624070422639646e-05,
+      "loss": 0.3977,
+      "step": 4334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3621280600529483e-05,
+      "loss": 0.3871,
+      "step": 4335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618490453994345e-05,
+      "loss": 0.3885,
+      "step": 4336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3615699983284204e-05,
+      "loss": 0.3714,
+      "step": 4337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612909188649046e-05,
+      "loss": 0.3944,
+      "step": 4338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36101180703389e-05,
+      "loss": 0.3884,
+      "step": 4339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607326628603815e-05,
+      "loss": 0.3576,
+      "step": 4340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360453486369387e-05,
+      "loss": 0.4035,
+      "step": 4341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601742775859182e-05,
+      "loss": 0.3724,
+      "step": 4342
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598950365349884e-05,
+      "loss": 0.3466,
+      "step": 4343
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596157632416143e-05,
+      "loss": 0.3889,
+      "step": 4344
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359336457730816e-05,
+      "loss": 0.3848,
+      "step": 4345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590571200276159e-05,
+      "loss": 0.3732,
+      "step": 4346
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3587777501570395e-05,
+      "loss": 0.4554,
+      "step": 4347
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584983481441145e-05,
+      "loss": 0.3765,
+      "step": 4348
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3582189140138731e-05,
+      "loss": 0.3517,
+      "step": 4349
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579394477913486e-05,
+      "loss": 0.3931,
+      "step": 4350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576599495015784e-05,
+      "loss": 0.3937,
+      "step": 4351
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573804191696015e-05,
+      "loss": 0.3733,
+      "step": 4352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571008568204619e-05,
+      "loss": 0.3795,
+      "step": 4353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568212624792044e-05,
+      "loss": 0.4464,
+      "step": 4354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565416361708774e-05,
+      "loss": 0.3702,
+      "step": 4355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3562619779205324e-05,
+      "loss": 0.3545,
+      "step": 4356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559822877532234e-05,
+      "loss": 0.4013,
+      "step": 4357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3557025656940078e-05,
+      "loss": 0.4217,
+      "step": 4358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3554228117679446e-05,
+      "loss": 0.4187,
+      "step": 4359
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551430260000975e-05,
+      "loss": 0.3885,
+      "step": 4360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548632084155315e-05,
+      "loss": 0.3666,
+      "step": 4361
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545833590393152e-05,
+      "loss": 0.4002,
+      "step": 4362
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543034778965199e-05,
+      "loss": 0.3941,
+      "step": 4363
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3540235650122195e-05,
+      "loss": 0.3979,
+      "step": 4364
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537436204114913e-05,
+      "loss": 0.3314,
+      "step": 4365
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534636441194149e-05,
+      "loss": 0.3631,
+      "step": 4366
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531836361610733e-05,
+      "loss": 0.3869,
+      "step": 4367
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529035965615511e-05,
+      "loss": 0.4115,
+      "step": 4368
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526235253459375e-05,
+      "loss": 0.3921,
+      "step": 4369
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523434225393234e-05,
+      "loss": 0.3544,
+      "step": 4370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520632881668028e-05,
+      "loss": 0.3965,
+      "step": 4371
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517831222534725e-05,
+      "loss": 0.4037,
+      "step": 4372
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3515029248244318e-05,
+      "loss": 0.3749,
+      "step": 4373
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512226959047839e-05,
+      "loss": 0.367,
+      "step": 4374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509424355196333e-05,
+      "loss": 0.3891,
+      "step": 4375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506621436940881e-05,
+      "loss": 0.3458,
+      "step": 4376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503818204532597e-05,
+      "loss": 0.3664,
+      "step": 4377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3501014658222617e-05,
+      "loss": 0.3909,
+      "step": 4378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349821079826211e-05,
+      "loss": 0.3903,
+      "step": 4379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3495406624902261e-05,
+      "loss": 0.3586,
+      "step": 4380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492602138394297e-05,
+      "loss": 0.3755,
+      "step": 4381
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489797338989468e-05,
+      "loss": 0.3908,
+      "step": 4382
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348699222693905e-05,
+      "loss": 0.3938,
+      "step": 4383
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484186802494346e-05,
+      "loss": 0.3665,
+      "step": 4384
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3481381065906696e-05,
+      "loss": 0.3952,
+      "step": 4385
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478575017427457e-05,
+      "loss": 0.4368,
+      "step": 4386
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475768657308021e-05,
+      "loss": 0.3453,
+      "step": 4387
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472961985799803e-05,
+      "loss": 0.4093,
+      "step": 4388
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470155003154251e-05,
+      "loss": 0.3749,
+      "step": 4389
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3467347709622839e-05,
+      "loss": 0.4013,
+      "step": 4390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3464540105457063e-05,
+      "loss": 0.3847,
+      "step": 4391
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3461732190908456e-05,
+      "loss": 0.4363,
+      "step": 4392
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3458923966228574e-05,
+      "loss": 0.3773,
+      "step": 4393
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456115431669e-05,
+      "loss": 0.4053,
+      "step": 4394
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453306587481353e-05,
+      "loss": 0.3643,
+      "step": 4395
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450497433917262e-05,
+      "loss": 0.3437,
+      "step": 4396
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447687971228402e-05,
+      "loss": 0.3791,
+      "step": 4397
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344487819966647e-05,
+      "loss": 0.3541,
+      "step": 4398
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442068119483185e-05,
+      "loss": 0.4035,
+      "step": 4399
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439257730930295e-05,
+      "loss": 0.402,
+      "step": 4400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3436447034259583e-05,
+      "loss": 0.3932,
+      "step": 4401
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343363602972286e-05,
+      "loss": 0.356,
+      "step": 4402
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430824717571948e-05,
+      "loss": 0.3372,
+      "step": 4403
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3428013098058719e-05,
+      "loss": 0.3951,
+      "step": 4404
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425201171435051e-05,
+      "loss": 0.3863,
+      "step": 4405
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3422388937952872e-05,
+      "loss": 0.395,
+      "step": 4406
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3419576397864115e-05,
+      "loss": 0.3838,
+      "step": 4407
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3416763551420756e-05,
+      "loss": 0.41,
+      "step": 4408
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413950398874795e-05,
+      "loss": 0.3926,
+      "step": 4409
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411136940478258e-05,
+      "loss": 0.4044,
+      "step": 4410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3408323176483195e-05,
+      "loss": 0.4008,
+      "step": 4411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405509107141687e-05,
+      "loss": 0.4033,
+      "step": 4412
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402694732705846e-05,
+      "loss": 0.4209,
+      "step": 4413
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3399880053427805e-05,
+      "loss": 0.3707,
+      "step": 4414
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397065069559726e-05,
+      "loss": 0.3364,
+      "step": 4415
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.33942497813538e-05,
+      "loss": 0.3617,
+      "step": 4416
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391434189062245e-05,
+      "loss": 0.4118,
+      "step": 4417
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3388618292937305e-05,
+      "loss": 0.3452,
+      "step": 4418
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338580209323125e-05,
+      "loss": 0.413,
+      "step": 4419
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382985590196384e-05,
+      "loss": 0.3918,
+      "step": 4420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3380168784085028e-05,
+      "loss": 0.3686,
+      "step": 4421
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337735167514954e-05,
+      "loss": 0.3868,
+      "step": 4422
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374534263642297e-05,
+      "loss": 0.3779,
+      "step": 4423
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371716549815705e-05,
+      "loss": 0.3876,
+      "step": 4424
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368898533922202e-05,
+      "loss": 0.3992,
+      "step": 4425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366080216214248e-05,
+      "loss": 0.3817,
+      "step": 4426
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3363261596944335e-05,
+      "loss": 0.3932,
+      "step": 4427
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3360442676364975e-05,
+      "loss": 0.3782,
+      "step": 4428
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357623454728711e-05,
+      "loss": 0.3739,
+      "step": 4429
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354803932288116e-05,
+      "loss": 0.3837,
+      "step": 4430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3351984109295787e-05,
+      "loss": 0.3723,
+      "step": 4431
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3349163986004342e-05,
+      "loss": 0.4132,
+      "step": 4432
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346343562666434e-05,
+      "loss": 0.3696,
+      "step": 4433
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334352283953474e-05,
+      "loss": 0.4111,
+      "step": 4434
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340701816861964e-05,
+      "loss": 0.3746,
+      "step": 4435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3337880494900843e-05,
+      "loss": 0.4007,
+      "step": 4436
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335058873904128e-05,
+      "loss": 0.3909,
+      "step": 4437
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333223695412461e-05,
+      "loss": 0.3608,
+      "step": 4438
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329414735815092e-05,
+      "loss": 0.3397,
+      "step": 4439
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326592219228415e-05,
+      "loss": 0.3783,
+      "step": 4440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332376940461745e-05,
+      "loss": 0.3807,
+      "step": 4441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320946292235082e-05,
+      "loss": 0.4317,
+      "step": 4442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318122882334235e-05,
+      "loss": 0.4022,
+      "step": 4443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315299175167847e-05,
+      "loss": 0.3854,
+      "step": 4444
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312475170988894e-05,
+      "loss": 0.3758,
+      "step": 4445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309650870050372e-05,
+      "loss": 0.3706,
+      "step": 4446
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3306826272605312e-05,
+      "loss": 0.3878,
+      "step": 4447
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3304001378906754e-05,
+      "loss": 0.3919,
+      "step": 4448
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3301176189207785e-05,
+      "loss": 0.391,
+      "step": 4449
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298350703761506e-05,
+      "loss": 0.4128,
+      "step": 4450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329552492282105e-05,
+      "loss": 0.3883,
+      "step": 4451
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292698846639573e-05,
+      "loss": 0.3941,
+      "step": 4452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289872475470257e-05,
+      "loss": 0.3747,
+      "step": 4453
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3287045809566317e-05,
+      "loss": 0.3597,
+      "step": 4454
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3284218849180985e-05,
+      "loss": 0.3622,
+      "step": 4455
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281391594567525e-05,
+      "loss": 0.3784,
+      "step": 4456
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327856404597923e-05,
+      "loss": 0.4073,
+      "step": 4457
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275736203669412e-05,
+      "loss": 0.3659,
+      "step": 4458
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272908067891412e-05,
+      "loss": 0.3953,
+      "step": 4459
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270079638898603e-05,
+      "loss": 0.3761,
+      "step": 4460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267250916944374e-05,
+      "loss": 0.3962,
+      "step": 4461
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264421902282152e-05,
+      "loss": 0.4334,
+      "step": 4462
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326159259516538e-05,
+      "loss": 0.4051,
+      "step": 4463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258762995847532e-05,
+      "loss": 0.3744,
+      "step": 4464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3255933104582106e-05,
+      "loss": 0.3715,
+      "step": 4465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253102921622632e-05,
+      "loss": 0.3905,
+      "step": 4466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250272447222657e-05,
+      "loss": 0.4126,
+      "step": 4467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3247441681635761e-05,
+      "loss": 0.3743,
+      "step": 4468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244610625115546e-05,
+      "loss": 0.4432,
+      "step": 4469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324177927791565e-05,
+      "loss": 0.4057,
+      "step": 4470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323894764028972e-05,
+      "loss": 0.3432,
+      "step": 4471
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236115712491438e-05,
+      "loss": 0.3861,
+      "step": 4472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233283494774516e-05,
+      "loss": 0.3559,
+      "step": 4473
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323045098739269e-05,
+      "loss": 0.3745,
+      "step": 4474
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3227618190599714e-05,
+      "loss": 0.3934,
+      "step": 4475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3224785104649379e-05,
+      "loss": 0.3984,
+      "step": 4476
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221951729795492e-05,
+      "loss": 0.3467,
+      "step": 4477
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219118066291896e-05,
+      "loss": 0.4128,
+      "step": 4478
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216284114392456e-05,
+      "loss": 0.4156,
+      "step": 4479
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321344987435105e-05,
+      "loss": 0.3798,
+      "step": 4480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3210615346421603e-05,
+      "loss": 0.354,
+      "step": 4481
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207780530858055e-05,
+      "loss": 0.3895,
+      "step": 4482
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204945427914371e-05,
+      "loss": 0.368,
+      "step": 4483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202110037844544e-05,
+      "loss": 0.3733,
+      "step": 4484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319927436090259e-05,
+      "loss": 0.3843,
+      "step": 4485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319643839734256e-05,
+      "loss": 0.4021,
+      "step": 4486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193602147418516e-05,
+      "loss": 0.3183,
+      "step": 4487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3190765611384558e-05,
+      "loss": 0.3588,
+      "step": 4488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.31879287894948e-05,
+      "loss": 0.367,
+      "step": 4489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185091682003398e-05,
+      "loss": 0.3687,
+      "step": 4490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3182254289164517e-05,
+      "loss": 0.3589,
+      "step": 4491
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179416611232361e-05,
+      "loss": 0.3989,
+      "step": 4492
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176578648461145e-05,
+      "loss": 0.3605,
+      "step": 4493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173740401105124e-05,
+      "loss": 0.3624,
+      "step": 4494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3170901869418573e-05,
+      "loss": 0.3644,
+      "step": 4495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3168063053655785e-05,
+      "loss": 0.391,
+      "step": 4496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165223954071093e-05,
+      "loss": 0.4285,
+      "step": 4497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3162384570918842e-05,
+      "loss": 0.3901,
+      "step": 4498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315954490445341e-05,
+      "loss": 0.4054,
+      "step": 4499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3156704954929198e-05,
+      "loss": 0.3453,
+      "step": 4500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315386472260063e-05,
+      "loss": 0.3955,
+      "step": 4501
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3151024207722168e-05,
+      "loss": 0.3462,
+      "step": 4502
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3148183410548278e-05,
+      "loss": 0.4467,
+      "step": 4503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.314534233133347e-05,
+      "loss": 0.3454,
+      "step": 4504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142500970332264e-05,
+      "loss": 0.3642,
+      "step": 4505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139659327799221e-05,
+      "loss": 0.3686,
+      "step": 4506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136817403988918e-05,
+      "loss": 0.3969,
+      "step": 4507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133975199155953e-05,
+      "loss": 0.3897,
+      "step": 4508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3131132713554962e-05,
+      "loss": 0.3934,
+      "step": 4509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128289947440593e-05,
+      "loss": 0.3719,
+      "step": 4510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125446901067533e-05,
+      "loss": 0.3523,
+      "step": 4511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122603574690478e-05,
+      "loss": 0.3645,
+      "step": 4512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311975996856416e-05,
+      "loss": 0.3673,
+      "step": 4513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116916082943332e-05,
+      "loss": 0.3935,
+      "step": 4514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3114071918082778e-05,
+      "loss": 0.367,
+      "step": 4515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111227474237297e-05,
+      "loss": 0.4078,
+      "step": 4516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108382751661722e-05,
+      "loss": 0.344,
+      "step": 4517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105537750610906e-05,
+      "loss": 0.3905,
+      "step": 4518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102692471339729e-05,
+      "loss": 0.384,
+      "step": 4519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099846914103093e-05,
+      "loss": 0.3809,
+      "step": 4520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3097001079155928e-05,
+      "loss": 0.393,
+      "step": 4521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3094154966753192e-05,
+      "loss": 0.3783,
+      "step": 4522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3091308577149858e-05,
+      "loss": 0.3969,
+      "step": 4523
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088461910600929e-05,
+      "loss": 0.3245,
+      "step": 4524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308561496736144e-05,
+      "loss": 0.3939,
+      "step": 4525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082767747686441e-05,
+      "loss": 0.4205,
+      "step": 4526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079920251831011e-05,
+      "loss": 0.3788,
+      "step": 4527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3077072480050248e-05,
+      "loss": 0.4058,
+      "step": 4528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3074224432599286e-05,
+      "loss": 0.3708,
+      "step": 4529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3071376109733273e-05,
+      "loss": 0.4047,
+      "step": 4530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306852751170739e-05,
+      "loss": 0.3746,
+      "step": 4531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3065678638776833e-05,
+      "loss": 0.4013,
+      "step": 4532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062829491196831e-05,
+      "loss": 0.3461,
+      "step": 4533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059980069222635e-05,
+      "loss": 0.4071,
+      "step": 4534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057130373109519e-05,
+      "loss": 0.3695,
+      "step": 4535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3054280403112787e-05,
+      "loss": 0.3628,
+      "step": 4536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3051430159487756e-05,
+      "loss": 0.4112,
+      "step": 4537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048579642489784e-05,
+      "loss": 0.3746,
+      "step": 4538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045728852374238e-05,
+      "loss": 0.3633,
+      "step": 4539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3042877789396515e-05,
+      "loss": 0.3661,
+      "step": 4540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040026453812042e-05,
+      "loss": 0.3993,
+      "step": 4541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037174845876266e-05,
+      "loss": 0.4051,
+      "step": 4542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034322965844655e-05,
+      "loss": 0.4042,
+      "step": 4543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3031470813972704e-05,
+      "loss": 0.3505,
+      "step": 4544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3028618390515935e-05,
+      "loss": 0.3741,
+      "step": 4545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025765695729892e-05,
+      "loss": 0.3776,
+      "step": 4546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3022912729870145e-05,
+      "loss": 0.4072,
+      "step": 4547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020059493192283e-05,
+      "loss": 0.3573,
+      "step": 4548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017205985951926e-05,
+      "loss": 0.388,
+      "step": 4549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014352208404712e-05,
+      "loss": 0.3528,
+      "step": 4550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301149816080631e-05,
+      "loss": 0.3948,
+      "step": 4551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008643843412414e-05,
+      "loss": 0.3857,
+      "step": 4552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005789256478725e-05,
+      "loss": 0.3508,
+      "step": 4553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3002934400260994e-05,
+      "loss": 0.426,
+      "step": 4554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000079275014976e-05,
+      "loss": 0.3668,
+      "step": 4555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299722388099646e-05,
+      "loss": 0.3768,
+      "step": 4556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994368218461255e-05,
+      "loss": 0.363,
+      "step": 4557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991512287665196e-05,
+      "loss": 0.3844,
+      "step": 4558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988656088864142e-05,
+      "loss": 0.3616,
+      "step": 4559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985799622313976e-05,
+      "loss": 0.3702,
+      "step": 4560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.29829428882706e-05,
+      "loss": 0.3684,
+      "step": 4561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298008588698995e-05,
+      "loss": 0.3665,
+      "step": 4562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977228618727981e-05,
+      "loss": 0.3917,
+      "step": 4563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2974371083740667e-05,
+      "loss": 0.3776,
+      "step": 4564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297151328228401e-05,
+      "loss": 0.3732,
+      "step": 4565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968655214614036e-05,
+      "loss": 0.379,
+      "step": 4566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965796880986798e-05,
+      "loss": 0.3697,
+      "step": 4567
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2962938281658372e-05,
+      "loss": 0.423,
+      "step": 4568
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.296007941688485e-05,
+      "loss": 0.4178,
+      "step": 4569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957220286922355e-05,
+      "loss": 0.4218,
+      "step": 4570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2954360892027029e-05,
+      "loss": 0.3701,
+      "step": 4571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951501232455049e-05,
+      "loss": 0.4036,
+      "step": 4572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948641308462598e-05,
+      "loss": 0.4021,
+      "step": 4573
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945781120305897e-05,
+      "loss": 0.371,
+      "step": 4574
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294292066824119e-05,
+      "loss": 0.3879,
+      "step": 4575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940059952524732e-05,
+      "loss": 0.3816,
+      "step": 4576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2937198973412816e-05,
+      "loss": 0.3666,
+      "step": 4577
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934337731161753e-05,
+      "loss": 0.3738,
+      "step": 4578
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2931476226027875e-05,
+      "loss": 0.3477,
+      "step": 4579
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928614458267537e-05,
+      "loss": 0.3867,
+      "step": 4580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925752428137126e-05,
+      "loss": 0.4095,
+      "step": 4581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922890135893042e-05,
+      "loss": 0.3899,
+      "step": 4582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920027581791716e-05,
+      "loss": 0.3961,
+      "step": 4583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2917164766089605e-05,
+      "loss": 0.4065,
+      "step": 4584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914301689043173e-05,
+      "loss": 0.3915,
+      "step": 4585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2911438350908927e-05,
+      "loss": 0.3672,
+      "step": 4586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908574751943386e-05,
+      "loss": 0.3837,
+      "step": 4587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905710892403096e-05,
+      "loss": 0.3983,
+      "step": 4588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.3733,
+      "step": 4589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899982392624566e-05,
+      "loss": 0.3947,
+      "step": 4590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897117752899537e-05,
+      "loss": 0.4177,
+      "step": 4591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2894252853626172e-05,
+      "loss": 0.3847,
+      "step": 4592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891387695061134e-05,
+      "loss": 0.3984,
+      "step": 4593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888522277461111e-05,
+      "loss": 0.3769,
+      "step": 4594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288565660108281e-05,
+      "loss": 0.4058,
+      "step": 4595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288279066618296e-05,
+      "loss": 0.383,
+      "step": 4596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.287992447301832e-05,
+      "loss": 0.3953,
+      "step": 4597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877058021845663e-05,
+      "loss": 0.3963,
+      "step": 4598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874191312921794e-05,
+      "loss": 0.3923,
+      "step": 4599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2871324346503537e-05,
+      "loss": 0.3912,
+      "step": 4600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868457122847737e-05,
+      "loss": 0.3766,
+      "step": 4601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2865589642211264e-05,
+      "loss": 0.3869,
+      "step": 4602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862721904851012e-05,
+      "loss": 0.3956,
+      "step": 4603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859853911023898e-05,
+      "loss": 0.4078,
+      "step": 4604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856985660986861e-05,
+      "loss": 0.3878,
+      "step": 4605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285411715499686e-05,
+      "loss": 0.393,
+      "step": 4606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851248393310885e-05,
+      "loss": 0.3835,
+      "step": 4607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284837937618594e-05,
+      "loss": 0.3865,
+      "step": 4608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2845510103879057e-05,
+      "loss": 0.3577,
+      "step": 4609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842640576647292e-05,
+      "loss": 0.3907,
+      "step": 4610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839770794747721e-05,
+      "loss": 0.3722,
+      "step": 4611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283690075843744e-05,
+      "loss": 0.3847,
+      "step": 4612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2834030467973572e-05,
+      "loss": 0.3584,
+      "step": 4613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831159923613265e-05,
+      "loss": 0.3648,
+      "step": 4614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828289125613684e-05,
+      "loss": 0.3785,
+      "step": 4615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2825418074232023e-05,
+      "loss": 0.3865,
+      "step": 4616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822546769725487e-05,
+      "loss": 0.3928,
+      "step": 4617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819675212351321e-05,
+      "loss": 0.3737,
+      "step": 4618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816803402366778e-05,
+      "loss": 0.3478,
+      "step": 4619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813931340029141e-05,
+      "loss": 0.3903,
+      "step": 4620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2811059025595714e-05,
+      "loss": 0.3461,
+      "step": 4621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808186459323816e-05,
+      "loss": 0.4217,
+      "step": 4622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2805313641470809e-05,
+      "loss": 0.3781,
+      "step": 4623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802440572294057e-05,
+      "loss": 0.3636,
+      "step": 4624
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2799567252050955e-05,
+      "loss": 0.3481,
+      "step": 4625
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2796693680998912e-05,
+      "loss": 0.3683,
+      "step": 4626
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2793819859395383e-05,
+      "loss": 0.3985,
+      "step": 4627
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790945787497813e-05,
+      "loss": 0.3831,
+      "step": 4628
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788071465563693e-05,
+      "loss": 0.3842,
+      "step": 4629
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785196893850532e-05,
+      "loss": 0.352,
+      "step": 4630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782322072615853e-05,
+      "loss": 0.4494,
+      "step": 4631
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779447002117212e-05,
+      "loss": 0.3788,
+      "step": 4632
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2776571682612176e-05,
+      "loss": 0.3897,
+      "step": 4633
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2773696114358344e-05,
+      "loss": 0.3967,
+      "step": 4634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2770820297613331e-05,
+      "loss": 0.4137,
+      "step": 4635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767944232634784e-05,
+      "loss": 0.4215,
+      "step": 4636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765067919680357e-05,
+      "loss": 0.4008,
+      "step": 4637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2762191359007739e-05,
+      "loss": 0.3514,
+      "step": 4638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759314550874633e-05,
+      "loss": 0.3632,
+      "step": 4639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756437495538774e-05,
+      "loss": 0.443,
+      "step": 4640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275356019325791e-05,
+      "loss": 0.3826,
+      "step": 4641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275068264428981e-05,
+      "loss": 0.3962,
+      "step": 4642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2747804848892275e-05,
+      "loss": 0.3838,
+      "step": 4643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274492680732312e-05,
+      "loss": 0.388,
+      "step": 4644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742048519840182e-05,
+      "loss": 0.3957,
+      "step": 4645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2739169986701328e-05,
+      "loss": 0.37,
+      "step": 4646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736291208164438e-05,
+      "loss": 0.3699,
+      "step": 4647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733412184487417e-05,
+      "loss": 0.3733,
+      "step": 4648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.273053291592819e-05,
+      "loss": 0.3743,
+      "step": 4649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272765340274471e-05,
+      "loss": 0.4031,
+      "step": 4650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2724773645194949e-05,
+      "loss": 0.3714,
+      "step": 4651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2721893643536899e-05,
+      "loss": 0.36,
+      "step": 4652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719013398028573e-05,
+      "loss": 0.3597,
+      "step": 4653
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2716132908928006e-05,
+      "loss": 0.3678,
+      "step": 4654
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713252176493263e-05,
+      "loss": 0.3822,
+      "step": 4655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271037120098242e-05,
+      "loss": 0.3766,
+      "step": 4656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707489982653578e-05,
+      "loss": 0.3854,
+      "step": 4657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704608521764862e-05,
+      "loss": 0.3919,
+      "step": 4658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701726818574424e-05,
+      "loss": 0.389,
+      "step": 4659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698844873340423e-05,
+      "loss": 0.4067,
+      "step": 4660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695962686321052e-05,
+      "loss": 0.3569,
+      "step": 4661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693080257774519e-05,
+      "loss": 0.3566,
+      "step": 4662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269019758795906e-05,
+      "loss": 0.3599,
+      "step": 4663
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2687314677132927e-05,
+      "loss": 0.4059,
+      "step": 4664
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684431525554394e-05,
+      "loss": 0.4078,
+      "step": 4665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2681548133481762e-05,
+      "loss": 0.4051,
+      "step": 4666
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678664501173346e-05,
+      "loss": 0.4078,
+      "step": 4667
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.267578062888749e-05,
+      "loss": 0.3934,
+      "step": 4668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672896516882551e-05,
+      "loss": 0.3569,
+      "step": 4669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670012165416914e-05,
+      "loss": 0.3844,
+      "step": 4670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2667127574748985e-05,
+      "loss": 0.3981,
+      "step": 4671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664242745137189e-05,
+      "loss": 0.3879,
+      "step": 4672
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661357676839975e-05,
+      "loss": 0.3527,
+      "step": 4673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265847237011581e-05,
+      "loss": 0.3553,
+      "step": 4674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2655586825223183e-05,
+      "loss": 0.4048,
+      "step": 4675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265270104242061e-05,
+      "loss": 0.3627,
+      "step": 4676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264981502196662e-05,
+      "loss": 0.4338,
+      "step": 4677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646928764119766e-05,
+      "loss": 0.3272,
+      "step": 4678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644042269138628e-05,
+      "loss": 0.3798,
+      "step": 4679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26411555372818e-05,
+      "loss": 0.4001,
+      "step": 4680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26382685688079e-05,
+      "loss": 0.4084,
+      "step": 4681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635381363975566e-05,
+      "loss": 0.3703,
+      "step": 4682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263249392304346e-05,
+      "loss": 0.3919,
+      "step": 4683
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629606246270266e-05,
+      "loss": 0.4074,
+      "step": 4684
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262671833391468e-05,
+      "loss": 0.4035,
+      "step": 4685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623830186235431e-05,
+      "loss": 0.4193,
+      "step": 4686
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2620941803491258e-05,
+      "loss": 0.4183,
+      "step": 4687
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2618053185940934e-05,
+      "loss": 0.3412,
+      "step": 4688
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615164333843244e-05,
+      "loss": 0.3949,
+      "step": 4689
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612275247456989e-05,
+      "loss": 0.3861,
+      "step": 4690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609385927041008e-05,
+      "loss": 0.3846,
+      "step": 4691
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606496372854142e-05,
+      "loss": 0.3317,
+      "step": 4692
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260360658515527e-05,
+      "loss": 0.3689,
+      "step": 4693
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600716564203274e-05,
+      "loss": 0.3713,
+      "step": 4694
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597826310257077e-05,
+      "loss": 0.3773,
+      "step": 4695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2594935823575604e-05,
+      "loss": 0.3664,
+      "step": 4696
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592045104417811e-05,
+      "loss": 0.3681,
+      "step": 4697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2589154153042677e-05,
+      "loss": 0.4051,
+      "step": 4698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586262969709196e-05,
+      "loss": 0.363,
+      "step": 4699
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583371554676383e-05,
+      "loss": 0.3769,
+      "step": 4700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580479908203277e-05,
+      "loss": 0.3829,
+      "step": 4701
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577588030548933e-05,
+      "loss": 0.3525,
+      "step": 4702
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2574695921972435e-05,
+      "loss": 0.4182,
+      "step": 4703
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257180358273288e-05,
+      "loss": 0.4169,
+      "step": 4704
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256891101308939e-05,
+      "loss": 0.3734,
+      "step": 4705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566018213301101e-05,
+      "loss": 0.4049,
+      "step": 4706
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563125183627178e-05,
+      "loss": 0.4451,
+      "step": 4707
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2560231924326805e-05,
+      "loss": 0.3756,
+      "step": 4708
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255733843565918e-05,
+      "loss": 0.361,
+      "step": 4709
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255444471788353e-05,
+      "loss": 0.3893,
+      "step": 4710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2551550771259098e-05,
+      "loss": 0.3451,
+      "step": 4711
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548656596045147e-05,
+      "loss": 0.3481,
+      "step": 4712
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2545762192500964e-05,
+      "loss": 0.38,
+      "step": 4713
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542867560885849e-05,
+      "loss": 0.4137,
+      "step": 4714
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539972701459133e-05,
+      "loss": 0.3827,
+      "step": 4715
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2537077614480163e-05,
+      "loss": 0.4105,
+      "step": 4716
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2534182300208299e-05,
+      "loss": 0.3785,
+      "step": 4717
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2531286758902931e-05,
+      "loss": 0.3802,
+      "step": 4718
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252839099082347e-05,
+      "loss": 0.3656,
+      "step": 4719
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252549499622934e-05,
+      "loss": 0.3945,
+      "step": 4720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522598775379985e-05,
+      "loss": 0.3717,
+      "step": 4721
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519702328534878e-05,
+      "loss": 0.3856,
+      "step": 4722
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2516805655953504e-05,
+      "loss": 0.3379,
+      "step": 4723
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2513908757895375e-05,
+      "loss": 0.3634,
+      "step": 4724
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511011634620018e-05,
+      "loss": 0.4116,
+      "step": 4725
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2508114286386979e-05,
+      "loss": 0.3664,
+      "step": 4726
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2505216713455835e-05,
+      "loss": 0.4071,
+      "step": 4727
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502318916086166e-05,
+      "loss": 0.4,
+      "step": 4728
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499420894537586e-05,
+      "loss": 0.4005,
+      "step": 4729
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2496522649069723e-05,
+      "loss": 0.3833,
+      "step": 4730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493624179942225e-05,
+      "loss": 0.3817,
+      "step": 4731
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490725487414768e-05,
+      "loss": 0.3781,
+      "step": 4732
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487826571747033e-05,
+      "loss": 0.3836,
+      "step": 4733
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484927433198733e-05,
+      "loss": 0.3614,
+      "step": 4734
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2482028072029596e-05,
+      "loss": 0.3502,
+      "step": 4735
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247912848849938e-05,
+      "loss": 0.3883,
+      "step": 4736
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247622868286784e-05,
+      "loss": 0.3678,
+      "step": 4737
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2473328655394772e-05,
+      "loss": 0.3596,
+      "step": 4738
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470428406339985e-05,
+      "loss": 0.3746,
+      "step": 4739
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246752793596331e-05,
+      "loss": 0.3819,
+      "step": 4740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2464627244524595e-05,
+      "loss": 0.3611,
+      "step": 4741
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461726332283704e-05,
+      "loss": 0.3805,
+      "step": 4742
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458825199500528e-05,
+      "loss": 0.3589,
+      "step": 4743
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455923846434976e-05,
+      "loss": 0.3819,
+      "step": 4744
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453022273346976e-05,
+      "loss": 0.3431,
+      "step": 4745
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2450120480496474e-05,
+      "loss": 0.3973,
+      "step": 4746
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447218468143439e-05,
+      "loss": 0.4131,
+      "step": 4747
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444316236547855e-05,
+      "loss": 0.3352,
+      "step": 4748
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441413785969732e-05,
+      "loss": 0.3759,
+      "step": 4749
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438511116669091e-05,
+      "loss": 0.3835,
+      "step": 4750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435608228905982e-05,
+      "loss": 0.3867,
+      "step": 4751
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2432705122940472e-05,
+      "loss": 0.3724,
+      "step": 4752
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242980179903264e-05,
+      "loss": 0.4002,
+      "step": 4753
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426898257442594e-05,
+      "loss": 0.3469,
+      "step": 4754
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423994498430458e-05,
+      "loss": 0.3685,
+      "step": 4755
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421090522256375e-05,
+      "loss": 0.3765,
+      "step": 4756
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418186329180506e-05,
+      "loss": 0.3941,
+      "step": 4757
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415281919463034e-05,
+      "loss": 0.3959,
+      "step": 4758
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241237729336416e-05,
+      "loss": 0.3835,
+      "step": 4759
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409472451144107e-05,
+      "loss": 0.3787,
+      "step": 4760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406567393063116e-05,
+      "loss": 0.3785,
+      "step": 4761
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2403662119381442e-05,
+      "loss": 0.3591,
+      "step": 4762
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2400756630359368e-05,
+      "loss": 0.3778,
+      "step": 4763
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397850926257193e-05,
+      "loss": 0.3761,
+      "step": 4764
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394945007335233e-05,
+      "loss": 0.4257,
+      "step": 4765
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2392038873853822e-05,
+      "loss": 0.3778,
+      "step": 4766
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238913252607332e-05,
+      "loss": 0.3744,
+      "step": 4767
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2386225964254105e-05,
+      "loss": 0.3661,
+      "step": 4768
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383319188656563e-05,
+      "loss": 0.4385,
+      "step": 4769
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380412199541112e-05,
+      "loss": 0.4408,
+      "step": 4770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377504997168186e-05,
+      "loss": 0.4042,
+      "step": 4771
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2374597581798232e-05,
+      "loss": 0.387,
+      "step": 4772
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371689953691733e-05,
+      "loss": 0.3745,
+      "step": 4773
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368782113109163e-05,
+      "loss": 0.3911,
+      "step": 4774
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236587406031104e-05,
+      "loss": 0.3878,
+      "step": 4775
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236296579555789e-05,
+      "loss": 0.3914,
+      "step": 4776
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360057319110264e-05,
+      "loss": 0.3938,
+      "step": 4777
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357148631228719e-05,
+      "loss": 0.3986,
+      "step": 4778
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2354239732173844e-05,
+      "loss": 0.3566,
+      "step": 4779
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2351330622206251e-05,
+      "loss": 0.388,
+      "step": 4780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2348421301586551e-05,
+      "loss": 0.3783,
+      "step": 4781
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345511770575389e-05,
+      "loss": 0.3477,
+      "step": 4782
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342602029433427e-05,
+      "loss": 0.3535,
+      "step": 4783
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339692078421347e-05,
+      "loss": 0.3887,
+      "step": 4784
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336781917799843e-05,
+      "loss": 0.3729,
+      "step": 4785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2333871547829632e-05,
+      "loss": 0.3882,
+      "step": 4786
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233096096877145e-05,
+      "loss": 0.3802,
+      "step": 4787
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232805018088605e-05,
+      "loss": 0.4097,
+      "step": 4788
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325139184434211e-05,
+      "loss": 0.394,
+      "step": 4789
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322227979676717e-05,
+      "loss": 0.3244,
+      "step": 4790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319316566874382e-05,
+      "loss": 0.4254,
+      "step": 4791
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2316404946288035e-05,
+      "loss": 0.3666,
+      "step": 4792
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313493118178524e-05,
+      "loss": 0.393,
+      "step": 4793
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310581082806713e-05,
+      "loss": 0.4057,
+      "step": 4794
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307668840433486e-05,
+      "loss": 0.396,
+      "step": 4795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304756391319754e-05,
+      "loss": 0.3912,
+      "step": 4796
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.230184373572643e-05,
+      "loss": 0.3819,
+      "step": 4797
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2298930873914463e-05,
+      "loss": 0.4052,
+      "step": 4798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22960178061448e-05,
+      "loss": 0.3678,
+      "step": 4799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293104532678428e-05,
+      "loss": 0.3798,
+      "step": 4800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229019105377634e-05,
+      "loss": 0.3903,
+      "step": 4801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287277369699548e-05,
+      "loss": 0.3372,
+      "step": 4802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2284363480709084e-05,
+      "loss": 0.3951,
+      "step": 4803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281449387066006e-05,
+      "loss": 0.3491,
+      "step": 4804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278535089031377e-05,
+      "loss": 0.356,
+      "step": 4805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275620586866286e-05,
+      "loss": 0.3953,
+      "step": 4806
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2272705880831836e-05,
+      "loss": 0.3531,
+      "step": 4807
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269790971189155e-05,
+      "loss": 0.3834,
+      "step": 4808
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266875858199385e-05,
+      "loss": 0.3478,
+      "step": 4809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2263960542123685e-05,
+      "loss": 0.3924,
+      "step": 4810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226104502322323e-05,
+      "loss": 0.3813,
+      "step": 4811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2258129301759224e-05,
+      "loss": 0.3641,
+      "step": 4812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225521337799288e-05,
+      "loss": 0.405,
+      "step": 4813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252297252185432e-05,
+      "loss": 0.3971,
+      "step": 4814
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2249380924598123e-05,
+      "loss": 0.39,
+      "step": 4815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246464395492233e-05,
+      "loss": 0.3563,
+      "step": 4816
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2243547665129043e-05,
+      "loss": 0.3984,
+      "step": 4817
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240630733769861e-05,
+      "loss": 0.3644,
+      "step": 4818
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237713601676009e-05,
+      "loss": 0.3852,
+      "step": 4819
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234796269108831e-05,
+      "loss": 0.3617,
+      "step": 4820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2231878736329684e-05,
+      "loss": 0.4099,
+      "step": 4821
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228961003599944e-05,
+      "loss": 0.3574,
+      "step": 4822
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2226043071181009e-05,
+      "loss": 0.3592,
+      "step": 4823
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222312493933429e-05,
+      "loss": 0.4032,
+      "step": 4824
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2220206608321223e-05,
+      "loss": 0.3664,
+      "step": 4825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221728807840325e-05,
+      "loss": 0.3832,
+      "step": 4826
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214369349841837e-05,
+      "loss": 0.3752,
+      "step": 4827
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211450422898476e-05,
+      "loss": 0.4172,
+      "step": 4828
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208531297834665e-05,
+      "loss": 0.3805,
+      "step": 4829
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205611974911924e-05,
+      "loss": 0.4083,
+      "step": 4830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2202692454391787e-05,
+      "loss": 0.4186,
+      "step": 4831
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199772736535817e-05,
+      "loss": 0.394,
+      "step": 4832
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196852821605583e-05,
+      "loss": 0.4107,
+      "step": 4833
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2193932709862674e-05,
+      "loss": 0.3882,
+      "step": 4834
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191012401568698e-05,
+      "loss": 0.3933,
+      "step": 4835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2188091896985285e-05,
+      "loss": 0.3411,
+      "step": 4836
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218517119637408e-05,
+      "loss": 0.3472,
+      "step": 4837
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2182250299996735e-05,
+      "loss": 0.3645,
+      "step": 4838
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2179329208114938e-05,
+      "loss": 0.349,
+      "step": 4839
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176407920990379e-05,
+      "loss": 0.3551,
+      "step": 4840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2173486438884776e-05,
+      "loss": 0.4036,
+      "step": 4841
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170564762059857e-05,
+      "loss": 0.3607,
+      "step": 4842
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2167642890777371e-05,
+      "loss": 0.3826,
+      "step": 4843
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164720825299085e-05,
+      "loss": 0.3509,
+      "step": 4844
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2161798565886782e-05,
+      "loss": 0.4146,
+      "step": 4845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158876112802265e-05,
+      "loss": 0.3979,
+      "step": 4846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155953466307345e-05,
+      "loss": 0.37,
+      "step": 4847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2153030626663866e-05,
+      "loss": 0.3514,
+      "step": 4848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2150107594133675e-05,
+      "loss": 0.3975,
+      "step": 4849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147184368978649e-05,
+      "loss": 0.3381,
+      "step": 4850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2144260951460662e-05,
+      "loss": 0.361,
+      "step": 4851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.214133734184163e-05,
+      "loss": 0.386,
+      "step": 4852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138413540383474e-05,
+      "loss": 0.4092,
+      "step": 4853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135489547348128e-05,
+      "loss": 0.3798,
+      "step": 4854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213256536299755e-05,
+      "loss": 0.3901,
+      "step": 4855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2129640987593714e-05,
+      "loss": 0.3869,
+      "step": 4856
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212671642139861e-05,
+      "loss": 0.3772,
+      "step": 4857
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123791664674244e-05,
+      "loss": 0.3854,
+      "step": 4858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120866717682642e-05,
+      "loss": 0.3519,
+      "step": 4859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117941580685846e-05,
+      "loss": 0.3752,
+      "step": 4860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2115016253945913e-05,
+      "loss": 0.4344,
+      "step": 4861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2112090737724922e-05,
+      "loss": 0.3349,
+      "step": 4862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2109165032284957e-05,
+      "loss": 0.357,
+      "step": 4863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2106239137888139e-05,
+      "loss": 0.3972,
+      "step": 4864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103313054796586e-05,
+      "loss": 0.3853,
+      "step": 4865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100386783272444e-05,
+      "loss": 0.3475,
+      "step": 4866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097460323577871e-05,
+      "loss": 0.3291,
+      "step": 4867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2094533675975048e-05,
+      "loss": 0.3717,
+      "step": 4868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209160684072617e-05,
+      "loss": 0.3988,
+      "step": 4869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208867981809344e-05,
+      "loss": 0.3384,
+      "step": 4870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2085752608339092e-05,
+      "loss": 0.3804,
+      "step": 4871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208282521172537e-05,
+      "loss": 0.3523,
+      "step": 4872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079897628514531e-05,
+      "loss": 0.3926,
+      "step": 4873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2076969858968854e-05,
+      "loss": 0.3645,
+      "step": 4874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074041903350637e-05,
+      "loss": 0.3905,
+      "step": 4875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2071113761922187e-05,
+      "loss": 0.3877,
+      "step": 4876
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2068185434945834e-05,
+      "loss": 0.3635,
+      "step": 4877
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206525692268392e-05,
+      "loss": 0.394,
+      "step": 4878
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206232822539881e-05,
+      "loss": 0.373,
+      "step": 4879
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059399343352875e-05,
+      "loss": 0.3847,
+      "step": 4880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056470276808516e-05,
+      "loss": 0.4007,
+      "step": 4881
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2053541026028141e-05,
+      "loss": 0.3479,
+      "step": 4882
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050611591274172e-05,
+      "loss": 0.3468,
+      "step": 4883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2047681972809062e-05,
+      "loss": 0.4092,
+      "step": 4884
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044752170895264e-05,
+      "loss": 0.3725,
+      "step": 4885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041822185795256e-05,
+      "loss": 0.3454,
+      "step": 4886
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2038892017771533e-05,
+      "loss": 0.3321,
+      "step": 4887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035961667086603e-05,
+      "loss": 0.3868,
+      "step": 4888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203303113400299e-05,
+      "loss": 0.3903,
+      "step": 4889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030100418783237e-05,
+      "loss": 0.3623,
+      "step": 4890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027169521689904e-05,
+      "loss": 0.4108,
+      "step": 4891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024238442985563e-05,
+      "loss": 0.396,
+      "step": 4892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021307182932806e-05,
+      "loss": 0.4039,
+      "step": 4893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018375741794243e-05,
+      "loss": 0.3899,
+      "step": 4894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2015444119832491e-05,
+      "loss": 0.3454,
+      "step": 4895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012512317310194e-05,
+      "loss": 0.3868,
+      "step": 4896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2009580334490007e-05,
+      "loss": 0.383,
+      "step": 4897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006648171634602e-05,
+      "loss": 0.3894,
+      "step": 4898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2003715829006664e-05,
+      "loss": 0.3871,
+      "step": 4899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2000783306868904e-05,
+      "loss": 0.3781,
+      "step": 4900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997850605484034e-05,
+      "loss": 0.3894,
+      "step": 4901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994917725114795e-05,
+      "loss": 0.3835,
+      "step": 4902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991984666023937e-05,
+      "loss": 0.3995,
+      "step": 4903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198905142847423e-05,
+      "loss": 0.3861,
+      "step": 4904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986118012728459e-05,
+      "loss": 0.3955,
+      "step": 4905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198318441904942e-05,
+      "loss": 0.353,
+      "step": 4906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980250647699934e-05,
+      "loss": 0.3594,
+      "step": 4907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977316698942831e-05,
+      "loss": 0.4356,
+      "step": 4908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197438257304096e-05,
+      "loss": 0.3534,
+      "step": 4909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971448270257185e-05,
+      "loss": 0.4184,
+      "step": 4910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196851379085438e-05,
+      "loss": 0.4131,
+      "step": 4911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965579135095447e-05,
+      "loss": 0.3596,
+      "step": 4912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1962644303243296e-05,
+      "loss": 0.3428,
+      "step": 4913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959709295560856e-05,
+      "loss": 0.3868,
+      "step": 4914
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956774112311064e-05,
+      "loss": 0.3438,
+      "step": 4915
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195383875375688e-05,
+      "loss": 0.4109,
+      "step": 4916
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.3785,
+      "step": 4917
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1947967511787262e-05,
+      "loss": 0.3531,
+      "step": 4918
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1945031628897816e-05,
+      "loss": 0.3644,
+      "step": 4919
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942095571755974e-05,
+      "loss": 0.3394,
+      "step": 4920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939159340624771e-05,
+      "loss": 0.3865,
+      "step": 4921
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1936222935767256e-05,
+      "loss": 0.3837,
+      "step": 4922
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19332863574465e-05,
+      "loss": 0.3399,
+      "step": 4923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1930349605925584e-05,
+      "loss": 0.425,
+      "step": 4924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1927412681467613e-05,
+      "loss": 0.37,
+      "step": 4925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1924475584335695e-05,
+      "loss": 0.4086,
+      "step": 4926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1921538314792963e-05,
+      "loss": 0.3868,
+      "step": 4927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191860087310256e-05,
+      "loss": 0.3932,
+      "step": 4928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191566325952765e-05,
+      "loss": 0.3857,
+      "step": 4929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191272547433141e-05,
+      "loss": 0.3519,
+      "step": 4930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909787517777028e-05,
+      "loss": 0.3851,
+      "step": 4931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906849390127712e-05,
+      "loss": 0.3304,
+      "step": 4932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903911091646684e-05,
+      "loss": 0.3757,
+      "step": 4933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900972622597184e-05,
+      "loss": 0.4053,
+      "step": 4934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898033983242468e-05,
+      "loss": 0.3716,
+      "step": 4935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895095173845795e-05,
+      "loss": 0.4249,
+      "step": 4936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892156194670454e-05,
+      "loss": 0.3897,
+      "step": 4937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1889217045979745e-05,
+      "loss": 0.3854,
+      "step": 4938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886277728036977e-05,
+      "loss": 0.3827,
+      "step": 4939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883338241105481e-05,
+      "loss": 0.3168,
+      "step": 4940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880398585448607e-05,
+      "loss": 0.3716,
+      "step": 4941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877458761329708e-05,
+      "loss": 0.3607,
+      "step": 4942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874518769012155e-05,
+      "loss": 0.3918,
+      "step": 4943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871578608759345e-05,
+      "loss": 0.3743,
+      "step": 4944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868638280834679e-05,
+      "loss": 0.3713,
+      "step": 4945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865697785501578e-05,
+      "loss": 0.3599,
+      "step": 4946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862757123023474e-05,
+      "loss": 0.3625,
+      "step": 4947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859816293663817e-05,
+      "loss": 0.3906,
+      "step": 4948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856875297686072e-05,
+      "loss": 0.3835,
+      "step": 4949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1853934135353717e-05,
+      "loss": 0.3603,
+      "step": 4950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1850992806930252e-05,
+      "loss": 0.376,
+      "step": 4951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848051312679179e-05,
+      "loss": 0.409,
+      "step": 4952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1845109652864026e-05,
+      "loss": 0.4002,
+      "step": 4953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1842167827748328e-05,
+      "loss": 0.4186,
+      "step": 4954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839225837595643e-05,
+      "loss": 0.3349,
+      "step": 4955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836283682669537e-05,
+      "loss": 0.3964,
+      "step": 4956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1833341363233594e-05,
+      "loss": 0.3278,
+      "step": 4957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830398879551412e-05,
+      "loss": 0.3708,
+      "step": 4958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18274562318866e-05,
+      "loss": 0.3843,
+      "step": 4959
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824513420502792e-05,
+      "loss": 0.3886,
+      "step": 4960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821570445663624e-05,
+      "loss": 0.3727,
+      "step": 4961
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1818627307632757e-05,
+      "loss": 0.385,
+      "step": 4962
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181568400667386e-05,
+      "loss": 0.3629,
+      "step": 4963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812740543050616e-05,
+      "loss": 0.3741,
+      "step": 4964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180979691702673e-05,
+      "loss": 0.4039,
+      "step": 4965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806853128865916e-05,
+      "loss": 0.3607,
+      "step": 4966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803909178831903e-05,
+      "loss": 0.3874,
+      "step": 4967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180096506718843e-05,
+      "loss": 0.3549,
+      "step": 4968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798020794199268e-05,
+      "loss": 0.3667,
+      "step": 4969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795076360128178e-05,
+      "loss": 0.3263,
+      "step": 4970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792131765238951e-05,
+      "loss": 0.3518,
+      "step": 4971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789187009795389e-05,
+      "loss": 0.3794,
+      "step": 4972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786242094061309e-05,
+      "loss": 0.3544,
+      "step": 4973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178329701830054e-05,
+      "loss": 0.4142,
+      "step": 4974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780351782776929e-05,
+      "loss": 0.3635,
+      "step": 4975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177740638775433e-05,
+      "loss": 0.3722,
+      "step": 4976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1774460833496623e-05,
+      "loss": 0.3853,
+      "step": 4977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771515120267696e-05,
+      "loss": 0.3863,
+      "step": 4978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1768569248331444e-05,
+      "loss": 0.3584,
+      "step": 4979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765623217951785e-05,
+      "loss": 0.3907,
+      "step": 4980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762677029392654e-05,
+      "loss": 0.3434,
+      "step": 4981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759730682917994e-05,
+      "loss": 0.3748,
+      "step": 4982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175678417879176e-05,
+      "loss": 0.3662,
+      "step": 4983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753837517277924e-05,
+      "loss": 0.3478,
+      "step": 4984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1750890698640482e-05,
+      "loss": 0.4057,
+      "step": 4985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747943723143427e-05,
+      "loss": 0.3484,
+      "step": 4986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1744996591050777e-05,
+      "loss": 0.3496,
+      "step": 4987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1742049302626555e-05,
+      "loss": 0.4086,
+      "step": 4988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739101858134813e-05,
+      "loss": 0.367,
+      "step": 4989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736154257839604e-05,
+      "loss": 0.3697,
+      "step": 4990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733206502004999e-05,
+      "loss": 0.3938,
+      "step": 4991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1730258590895084e-05,
+      "loss": 0.4,
+      "step": 4992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1727310524773956e-05,
+      "loss": 0.3503,
+      "step": 4993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172436230390573e-05,
+      "loss": 0.3998,
+      "step": 4994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721413928554529e-05,
+      "loss": 0.4267,
+      "step": 4995
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1718465398984496e-05,
+      "loss": 0.3647,
+      "step": 4996
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1715516715459784e-05,
+      "loss": 0.373,
+      "step": 4997
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712567878244565e-05,
+      "loss": 0.3349,
+      "step": 4998
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709618887603013e-05,
+      "loss": 0.3797,
+      "step": 4999
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170666974379933e-05,
+      "loss": 0.3666,
+      "step": 5000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703720447097726e-05,
+      "loss": 0.3408,
+      "step": 5001
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700770997762419e-05,
+      "loss": 0.3844,
+      "step": 5002
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697821396057652e-05,
+      "loss": 0.3744,
+      "step": 5003
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1694871642247665e-05,
+      "loss": 0.3689,
+      "step": 5004
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691921736596732e-05,
+      "loss": 0.3511,
+      "step": 5005
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688971679369125e-05,
+      "loss": 0.401,
+      "step": 5006
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168602147082914e-05,
+      "loss": 0.4025,
+      "step": 5007
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1683071111241075e-05,
+      "loss": 0.3867,
+      "step": 5008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1680120600869254e-05,
+      "loss": 0.3579,
+      "step": 5009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677169939978009e-05,
+      "loss": 0.3193,
+      "step": 5010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674219128831678e-05,
+      "loss": 0.3689,
+      "step": 5011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671268167694624e-05,
+      "loss": 0.4038,
+      "step": 5012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668317056831223e-05,
+      "loss": 0.4105,
+      "step": 5013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1665365796505856e-05,
+      "loss": 0.3719,
+      "step": 5014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1662414386982921e-05,
+      "loss": 0.3665,
+      "step": 5015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1659462828526833e-05,
+      "loss": 0.389,
+      "step": 5016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1656511121402016e-05,
+      "loss": 0.3529,
+      "step": 5017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165355926587291e-05,
+      "loss": 0.3685,
+      "step": 5018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650607262203967e-05,
+      "loss": 0.3636,
+      "step": 5019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647655110659651e-05,
+      "loss": 0.3417,
+      "step": 5020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644702811504443e-05,
+      "loss": 0.4134,
+      "step": 5021
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1641750365002834e-05,
+      "loss": 0.391,
+      "step": 5022
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163879777141933e-05,
+      "loss": 0.3765,
+      "step": 5023
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1635845031018444e-05,
+      "loss": 0.3763,
+      "step": 5024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632892144064714e-05,
+      "loss": 0.348,
+      "step": 5025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629939110822686e-05,
+      "loss": 0.3829,
+      "step": 5026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1626985931556911e-05,
+      "loss": 0.3695,
+      "step": 5027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1624032606531964e-05,
+      "loss": 0.3644,
+      "step": 5028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621079136012426e-05,
+      "loss": 0.3797,
+      "step": 5029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1618125520262898e-05,
+      "loss": 0.3677,
+      "step": 5030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615171759547985e-05,
+      "loss": 0.3715,
+      "step": 5031
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1612217854132313e-05,
+      "loss": 0.4119,
+      "step": 5032
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609263804280519e-05,
+      "loss": 0.378,
+      "step": 5033
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606309610257249e-05,
+      "loss": 0.3113,
+      "step": 5034
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603355272327167e-05,
+      "loss": 0.3851,
+      "step": 5035
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1600400790754945e-05,
+      "loss": 0.3643,
+      "step": 5036
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597446165805272e-05,
+      "loss": 0.4039,
+      "step": 5037
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159449139774285e-05,
+      "loss": 0.3672,
+      "step": 5038
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159153648683239e-05,
+      "loss": 0.3741,
+      "step": 5039
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588581433338614e-05,
+      "loss": 0.358,
+      "step": 5040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585626237526269e-05,
+      "loss": 0.4109,
+      "step": 5041
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582670899660105e-05,
+      "loss": 0.3639,
+      "step": 5042
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157971542000488e-05,
+      "loss": 0.3422,
+      "step": 5043
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576759798825375e-05,
+      "loss": 0.4001,
+      "step": 5044
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573804036386377e-05,
+      "loss": 0.3246,
+      "step": 5045
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1570848132952694e-05,
+      "loss": 0.3615,
+      "step": 5046
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567892088789137e-05,
+      "loss": 0.3565,
+      "step": 5047
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156493590416053e-05,
+      "loss": 0.3562,
+      "step": 5048
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156197957933172e-05,
+      "loss": 0.3658,
+      "step": 5049
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1559023114567554e-05,
+      "loss": 0.4026,
+      "step": 5050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1556066510132904e-05,
+      "loss": 0.4193,
+      "step": 5051
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553109766292638e-05,
+      "loss": 0.4348,
+      "step": 5052
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550152883311653e-05,
+      "loss": 0.4095,
+      "step": 5053
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154719586145485e-05,
+      "loss": 0.3703,
+      "step": 5054
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544238700987145e-05,
+      "loss": 0.3753,
+      "step": 5055
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1541281402173461e-05,
+      "loss": 0.3797,
+      "step": 5056
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1538323965278744e-05,
+      "loss": 0.3669,
+      "step": 5057
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535366390567944e-05,
+      "loss": 0.4078,
+      "step": 5058
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532408678306024e-05,
+      "loss": 0.3734,
+      "step": 5059
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152945082875796e-05,
+      "loss": 0.4035,
+      "step": 5060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526492842188746e-05,
+      "loss": 0.4115,
+      "step": 5061
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523534718863381e-05,
+      "loss": 0.3457,
+      "step": 5062
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1520576459046877e-05,
+      "loss": 0.3562,
+      "step": 5063
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1517618063004261e-05,
+      "loss": 0.3637,
+      "step": 5064
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1514659531000572e-05,
+      "loss": 0.402,
+      "step": 5065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511700863300859e-05,
+      "loss": 0.3886,
+      "step": 5066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1508742060170187e-05,
+      "loss": 0.3804,
+      "step": 5067
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1505783121873626e-05,
+      "loss": 0.366,
+      "step": 5068
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150282404867627e-05,
+      "loss": 0.3858,
+      "step": 5069
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149986484084321e-05,
+      "loss": 0.3434,
+      "step": 5070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496905498639562e-05,
+      "loss": 0.3841,
+      "step": 5071
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493946022330444e-05,
+      "loss": 0.3707,
+      "step": 5072
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490986412180995e-05,
+      "loss": 0.3486,
+      "step": 5073
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488026668456368e-05,
+      "loss": 0.3758,
+      "step": 5074
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485066791421709e-05,
+      "loss": 0.3679,
+      "step": 5075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482106781342194e-05,
+      "loss": 0.4155,
+      "step": 5076
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147914663848301e-05,
+      "loss": 0.3574,
+      "step": 5077
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147618636310935e-05,
+      "loss": 0.3648,
+      "step": 5078
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473225955486416e-05,
+      "loss": 0.4039,
+      "step": 5079
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470265415879428e-05,
+      "loss": 0.4054,
+      "step": 5080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467304744553618e-05,
+      "loss": 0.3425,
+      "step": 5081
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464343941774227e-05,
+      "loss": 0.3473,
+      "step": 5082
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1461383007806515e-05,
+      "loss": 0.3707,
+      "step": 5083
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458421942915736e-05,
+      "loss": 0.4149,
+      "step": 5084
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455460747367173e-05,
+      "loss": 0.368,
+      "step": 5085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452499421426118e-05,
+      "loss": 0.3333,
+      "step": 5086
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449537965357869e-05,
+      "loss": 0.4032,
+      "step": 5087
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446576379427734e-05,
+      "loss": 0.3533,
+      "step": 5088
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443614663901044e-05,
+      "loss": 0.3812,
+      "step": 5089
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144065281904313e-05,
+      "loss": 0.3693,
+      "step": 5090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143769084511934e-05,
+      "loss": 0.4044,
+      "step": 5091
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434728742395034e-05,
+      "loss": 0.3748,
+      "step": 5092
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1431766511135581e-05,
+      "loss": 0.3335,
+      "step": 5093
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1428804151606366e-05,
+      "loss": 0.3923,
+      "step": 5094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1425841664072776e-05,
+      "loss": 0.3536,
+      "step": 5095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422879048800222e-05,
+      "loss": 0.4034,
+      "step": 5096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419916306054115e-05,
+      "loss": 0.3601,
+      "step": 5097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416953436099884e-05,
+      "loss": 0.4318,
+      "step": 5098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413990439202974e-05,
+      "loss": 0.3896,
+      "step": 5099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411027315628827e-05,
+      "loss": 0.3492,
+      "step": 5100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1408064065642907e-05,
+      "loss": 0.3743,
+      "step": 5101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140510068951069e-05,
+      "loss": 0.3771,
+      "step": 5102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140213718749766e-05,
+      "loss": 0.3746,
+      "step": 5103
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1399173559869307e-05,
+      "loss": 0.375,
+      "step": 5104
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396209806891144e-05,
+      "loss": 0.3393,
+      "step": 5105
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393245928828684e-05,
+      "loss": 0.3922,
+      "step": 5106
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390281925947461e-05,
+      "loss": 0.3615,
+      "step": 5107
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1387317798513016e-05,
+      "loss": 0.3943,
+      "step": 5108
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384353546790891e-05,
+      "loss": 0.3431,
+      "step": 5109
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381389171046665e-05,
+      "loss": 0.3497,
+      "step": 5110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378424671545898e-05,
+      "loss": 0.3858,
+      "step": 5111
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137546004855418e-05,
+      "loss": 0.3511,
+      "step": 5112
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1372495302337107e-05,
+      "loss": 0.3785,
+      "step": 5113
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1369530433160288e-05,
+      "loss": 0.3325,
+      "step": 5114
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366565441289337e-05,
+      "loss": 0.3898,
+      "step": 5115
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363600326989886e-05,
+      "loss": 0.3917,
+      "step": 5116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360635090527571e-05,
+      "loss": 0.3575,
+      "step": 5117
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357669732168051e-05,
+      "loss": 0.3933,
+      "step": 5118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354704252176983e-05,
+      "loss": 0.411,
+      "step": 5119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351738650820034e-05,
+      "loss": 0.4,
+      "step": 5120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1348772928362898e-05,
+      "loss": 0.3557,
+      "step": 5121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1345807085071263e-05,
+      "loss": 0.378,
+      "step": 5122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342841121210836e-05,
+      "loss": 0.3794,
+      "step": 5123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339875037047339e-05,
+      "loss": 0.3603,
+      "step": 5124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336908832846485e-05,
+      "loss": 0.3981,
+      "step": 5125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333942508874025e-05,
+      "loss": 0.3475,
+      "step": 5126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1330976065395702e-05,
+      "loss": 0.4218,
+      "step": 5127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328009502677275e-05,
+      "loss": 0.4026,
+      "step": 5128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325042820984515e-05,
+      "loss": 0.3993,
+      "step": 5129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322076020583204e-05,
+      "loss": 0.3929,
+      "step": 5130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319109101739127e-05,
+      "loss": 0.3589,
+      "step": 5131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1316142064718093e-05,
+      "loss": 0.3558,
+      "step": 5132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131317490978591e-05,
+      "loss": 0.3581,
+      "step": 5133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1310207637208398e-05,
+      "loss": 0.4096,
+      "step": 5134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307240247251401e-05,
+      "loss": 0.3913,
+      "step": 5135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130427274018075e-05,
+      "loss": 0.3732,
+      "step": 5136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130130511626231e-05,
+      "loss": 0.3832,
+      "step": 5137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1298337375761938e-05,
+      "loss": 0.3598,
+      "step": 5138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295369518945517e-05,
+      "loss": 0.3929,
+      "step": 5139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1292401546078926e-05,
+      "loss": 0.3494,
+      "step": 5140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128943345742806e-05,
+      "loss": 0.352,
+      "step": 5141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286465253258837e-05,
+      "loss": 0.353,
+      "step": 5142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283496933837158e-05,
+      "loss": 0.3417,
+      "step": 5143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280528499428964e-05,
+      "loss": 0.3792,
+      "step": 5144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277559950300183e-05,
+      "loss": 0.3568,
+      "step": 5145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274591286716769e-05,
+      "loss": 0.3939,
+      "step": 5146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271622508944677e-05,
+      "loss": 0.4102,
+      "step": 5147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268653617249874e-05,
+      "loss": 0.4053,
+      "step": 5148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265684611898342e-05,
+      "loss": 0.3482,
+      "step": 5149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1262715493156067e-05,
+      "loss": 0.3378,
+      "step": 5150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125974626128905e-05,
+      "loss": 0.386,
+      "step": 5151
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256776916563296e-05,
+      "loss": 0.3228,
+      "step": 5152
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1253807459244828e-05,
+      "loss": 0.3562,
+      "step": 5153
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250837889599674e-05,
+      "loss": 0.4091,
+      "step": 5154
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1247868207893871e-05,
+      "loss": 0.3871,
+      "step": 5155
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244898414393472e-05,
+      "loss": 0.3455,
+      "step": 5156
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241928509364533e-05,
+      "loss": 0.3296,
+      "step": 5157
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238958493073127e-05,
+      "loss": 0.3425,
+      "step": 5158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1235988365785326e-05,
+      "loss": 0.3776,
+      "step": 5159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123301812776723e-05,
+      "loss": 0.3653,
+      "step": 5160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230047779284926e-05,
+      "loss": 0.4226,
+      "step": 5161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227077320604534e-05,
+      "loss": 0.3468,
+      "step": 5162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224106751992164e-05,
+      "loss": 0.36,
+      "step": 5163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122113607371395e-05,
+      "loss": 0.3901,
+      "step": 5164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1218165286036026e-05,
+      "loss": 0.3378,
+      "step": 5165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1215194389224546e-05,
+      "loss": 0.36,
+      "step": 5166
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212223383545669e-05,
+      "loss": 0.3731,
+      "step": 5167
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1209252269265553e-05,
+      "loss": 0.3725,
+      "step": 5168
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206281046650385e-05,
+      "loss": 0.4088,
+      "step": 5169
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203309715966348e-05,
+      "loss": 0.3613,
+      "step": 5170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200338277479643e-05,
+      "loss": 0.3514,
+      "step": 5171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119736673145647e-05,
+      "loss": 0.3739,
+      "step": 5172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194395078163047e-05,
+      "loss": 0.3787,
+      "step": 5173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1191423317865606e-05,
+      "loss": 0.3818,
+      "step": 5174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188451450830375e-05,
+      "loss": 0.3422,
+      "step": 5175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1185479477323607e-05,
+      "loss": 0.3753,
+      "step": 5176
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1182507397611546e-05,
+      "loss": 0.3765,
+      "step": 5177
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179535211960465e-05,
+      "loss": 0.3828,
+      "step": 5178
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1176562920636633e-05,
+      "loss": 0.3763,
+      "step": 5179
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1173590523906334e-05,
+      "loss": 0.3562,
+      "step": 5180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1170618022035862e-05,
+      "loss": 0.3516,
+      "step": 5181
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167645415291517e-05,
+      "loss": 0.3314,
+      "step": 5182
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164672703939614e-05,
+      "loss": 0.3772,
+      "step": 5183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161699888246466e-05,
+      "loss": 0.3363,
+      "step": 5184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115872696847841e-05,
+      "loss": 0.3521,
+      "step": 5185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115575394490178e-05,
+      "loss": 0.3107,
+      "step": 5186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152780817782934e-05,
+      "loss": 0.3471,
+      "step": 5187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114980758738822e-05,
+      "loss": 0.3718,
+      "step": 5188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1146834253984008e-05,
+      "loss": 0.3821,
+      "step": 5189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143860817836674e-05,
+      "loss": 0.384,
+      "step": 5190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140887279212605e-05,
+      "loss": 0.3545,
+      "step": 5191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1137913638378198e-05,
+      "loss": 0.3548,
+      "step": 5192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134939895599847e-05,
+      "loss": 0.4009,
+      "step": 5193
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131966051143981e-05,
+      "loss": 0.3864,
+      "step": 5194
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128992105277008e-05,
+      "loss": 0.3515,
+      "step": 5195
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1126018058265366e-05,
+      "loss": 0.4038,
+      "step": 5196
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123043910375495e-05,
+      "loss": 0.366,
+      "step": 5197
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1120069661873843e-05,
+      "loss": 0.3403,
+      "step": 5198
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117095313026869e-05,
+      "loss": 0.366,
+      "step": 5199
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111412086410104e-05,
+      "loss": 0.3878,
+      "step": 5200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111146315362831e-05,
+      "loss": 0.3482,
+      "step": 5201
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1108171667078728e-05,
+      "loss": 0.3843,
+      "step": 5202
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1105196919515231e-05,
+      "loss": 0.3545,
+      "step": 5203
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102222072938832e-05,
+      "loss": 0.3561,
+      "step": 5204
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099247127616051e-05,
+      "loss": 0.403,
+      "step": 5205
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096272083813406e-05,
+      "loss": 0.3777,
+      "step": 5206
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109329694179743e-05,
+      "loss": 0.3233,
+      "step": 5207
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1090321701834656e-05,
+      "loss": 0.3687,
+      "step": 5208
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1087346364191632e-05,
+      "loss": 0.3289,
+      "step": 5209
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108437092913492e-05,
+      "loss": 0.3724,
+      "step": 5210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1081395396931079e-05,
+      "loss": 0.3605,
+      "step": 5211
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1078419767846684e-05,
+      "loss": 0.3616,
+      "step": 5212
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1075444042148313e-05,
+      "loss": 0.3619,
+      "step": 5213
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072468220102564e-05,
+      "loss": 0.359,
+      "step": 5214
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069492301976032e-05,
+      "loss": 0.365,
+      "step": 5215
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1066516288035328e-05,
+      "loss": 0.3745,
+      "step": 5216
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1063540178547064e-05,
+      "loss": 0.3621,
+      "step": 5217
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060563973777864e-05,
+      "loss": 0.3735,
+      "step": 5218
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1057587673994372e-05,
+      "loss": 0.3239,
+      "step": 5219
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105461127946322e-05,
+      "loss": 0.3574,
+      "step": 5220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105163479045106e-05,
+      "loss": 0.394,
+      "step": 5221
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048658207224554e-05,
+      "loss": 0.3324,
+      "step": 5222
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045681530050367e-05,
+      "loss": 0.324,
+      "step": 5223
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042704759195181e-05,
+      "loss": 0.3788,
+      "step": 5224
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1039727894925672e-05,
+      "loss": 0.3877,
+      "step": 5225
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1036750937508535e-05,
+      "loss": 0.3475,
+      "step": 5226
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033773887210475e-05,
+      "loss": 0.3824,
+      "step": 5227
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.10307967442982e-05,
+      "loss": 0.3551,
+      "step": 5228
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1027819509038422e-05,
+      "loss": 0.3611,
+      "step": 5229
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1024842181697872e-05,
+      "loss": 0.3575,
+      "step": 5230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1021864762543288e-05,
+      "loss": 0.4045,
+      "step": 5231
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1018887251841406e-05,
+      "loss": 0.3794,
+      "step": 5232
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015909649858977e-05,
+      "loss": 0.3528,
+      "step": 5233
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101293195686276e-05,
+      "loss": 0.3714,
+      "step": 5234
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009954173119528e-05,
+      "loss": 0.396,
+      "step": 5235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1006976298896049e-05,
+      "loss": 0.3911,
+      "step": 5236
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1003998334459107e-05,
+      "loss": 0.3671,
+      "step": 5237
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001020280075496e-05,
+      "loss": 0.3301,
+      "step": 5238
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998042136012014e-05,
+      "loss": 0.3554,
+      "step": 5239
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099506390253547e-05,
+      "loss": 0.3914,
+      "step": 5240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0992085579912678e-05,
+      "loss": 0.3951,
+      "step": 5241
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989107168410457e-05,
+      "loss": 0.3961,
+      "step": 5242
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986128668295646e-05,
+      "loss": 0.4043,
+      "step": 5243
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983150079835081e-05,
+      "loss": 0.3595,
+      "step": 5244
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.3609,
+      "step": 5245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977192638944081e-05,
+      "loss": 0.3765,
+      "step": 5246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.097421378704737e-05,
+      "loss": 0.409,
+      "step": 5247
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971234847872335e-05,
+      "loss": 0.3831,
+      "step": 5248
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0968255821685864e-05,
+      "loss": 0.3719,
+      "step": 5249
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0965276708754838e-05,
+      "loss": 0.3884,
+      "step": 5250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962297509346156e-05,
+      "loss": 0.3857,
+      "step": 5251
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959318223726714e-05,
+      "loss": 0.4115,
+      "step": 5252
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956338852163424e-05,
+      "loss": 0.4049,
+      "step": 5253
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0953359394923205e-05,
+      "loss": 0.3824,
+      "step": 5254
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0950379852272978e-05,
+      "loss": 0.4125,
+      "step": 5255
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947400224479684e-05,
+      "loss": 0.3522,
+      "step": 5256
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944420511810254e-05,
+      "loss": 0.32,
+      "step": 5257
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0941440714531639e-05,
+      "loss": 0.3835,
+      "step": 5258
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0938460832910795e-05,
+      "loss": 0.4188,
+      "step": 5259
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0935480867214687e-05,
+      "loss": 0.3421,
+      "step": 5260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932500817710284e-05,
+      "loss": 0.3875,
+      "step": 5261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929520684664559e-05,
+      "loss": 0.39,
+      "step": 5262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0926540468344507e-05,
+      "loss": 0.4059,
+      "step": 5263
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0923560169017114e-05,
+      "loss": 0.3729,
+      "step": 5264
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0920579786949381e-05,
+      "loss": 0.3752,
+      "step": 5265
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917599322408318e-05,
+      "loss": 0.3684,
+      "step": 5266
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0914618775660943e-05,
+      "loss": 0.4237,
+      "step": 5267
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911638146974273e-05,
+      "loss": 0.4291,
+      "step": 5268
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908657436615342e-05,
+      "loss": 0.3759,
+      "step": 5269
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0905676644851183e-05,
+      "loss": 0.3692,
+      "step": 5270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0902695771948843e-05,
+      "loss": 0.3429,
+      "step": 5271
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899714818175378e-05,
+      "loss": 0.4043,
+      "step": 5272
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896733783797839e-05,
+      "loss": 0.3913,
+      "step": 5273
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.08937526690833e-05,
+      "loss": 0.4057,
+      "step": 5274
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0890771474298826e-05,
+      "loss": 0.3506,
+      "step": 5275
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887790199711508e-05,
+      "loss": 0.4303,
+      "step": 5276
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884808845588424e-05,
+      "loss": 0.3567,
+      "step": 5277
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881827412196673e-05,
+      "loss": 0.3953,
+      "step": 5278
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087884589980336e-05,
+      "loss": 0.337,
+      "step": 5279
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087586430867559e-05,
+      "loss": 0.3912,
+      "step": 5280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872882639080482e-05,
+      "loss": 0.3423,
+      "step": 5281
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0869900891285154e-05,
+      "loss": 0.3554,
+      "step": 5282
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866919065556745e-05,
+      "loss": 0.4204,
+      "step": 5283
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863937162162383e-05,
+      "loss": 0.377,
+      "step": 5284
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860955181369219e-05,
+      "loss": 0.4001,
+      "step": 5285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857973123444401e-05,
+      "loss": 0.385,
+      "step": 5286
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854990988655085e-05,
+      "loss": 0.3366,
+      "step": 5287
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852008777268441e-05,
+      "loss": 0.3941,
+      "step": 5288
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849026489551637e-05,
+      "loss": 0.3845,
+      "step": 5289
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846044125771853e-05,
+      "loss": 0.3632,
+      "step": 5290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0843061686196274e-05,
+      "loss": 0.3964,
+      "step": 5291
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0840079171092093e-05,
+      "loss": 0.3668,
+      "step": 5292
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837096580726508e-05,
+      "loss": 0.3874,
+      "step": 5293
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834113915366721e-05,
+      "loss": 0.3906,
+      "step": 5294
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831131175279952e-05,
+      "loss": 0.3788,
+      "step": 5295
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828148360733414e-05,
+      "loss": 0.3899,
+      "step": 5296
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082516547199434e-05,
+      "loss": 0.4004,
+      "step": 5297
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822182509329951e-05,
+      "loss": 0.3814,
+      "step": 5298
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0819199473007498e-05,
+      "loss": 0.3806,
+      "step": 5299
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816216363294219e-05,
+      "loss": 0.3125,
+      "step": 5300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813233180457368e-05,
+      "loss": 0.376,
+      "step": 5301
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810249924764205e-05,
+      "loss": 0.354,
+      "step": 5302
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807266596481993e-05,
+      "loss": 0.3903,
+      "step": 5303
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804283195878007e-05,
+      "loss": 0.3729,
+      "step": 5304
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801299723219526e-05,
+      "loss": 0.3624,
+      "step": 5305
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0798316178773827e-05,
+      "loss": 0.3385,
+      "step": 5306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079533256280821e-05,
+      "loss": 0.3813,
+      "step": 5307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792348875589968e-05,
+      "loss": 0.3541,
+      "step": 5308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0789365117386408e-05,
+      "loss": 0.3369,
+      "step": 5309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786381288464836e-05,
+      "loss": 0.3887,
+      "step": 5310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0783397389092571e-05,
+      "loss": 0.3428,
+      "step": 5311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780413419536935e-05,
+      "loss": 0.3665,
+      "step": 5312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777429380065263e-05,
+      "loss": 0.348,
+      "step": 5313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077444527094488e-05,
+      "loss": 0.4185,
+      "step": 5314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771461092443138e-05,
+      "loss": 0.3829,
+      "step": 5315
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0768476844827378e-05,
+      "loss": 0.3789,
+      "step": 5316
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076549252836496e-05,
+      "loss": 0.3782,
+      "step": 5317
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762508143323236e-05,
+      "loss": 0.374,
+      "step": 5318
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075952368996958e-05,
+      "loss": 0.3737,
+      "step": 5319
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0756539168571362e-05,
+      "loss": 0.3528,
+      "step": 5320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0753554579395961e-05,
+      "loss": 0.3956,
+      "step": 5321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.075056992271076e-05,
+      "loss": 0.3545,
+      "step": 5322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747585198783155e-05,
+      "loss": 0.3652,
+      "step": 5323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744600407880537e-05,
+      "loss": 0.361,
+      "step": 5324
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074161555027031e-05,
+      "loss": 0.3079,
+      "step": 5325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0738630626219885e-05,
+      "loss": 0.3623,
+      "step": 5326
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735645635996676e-05,
+      "loss": 0.389,
+      "step": 5327
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0732660579868102e-05,
+      "loss": 0.3887,
+      "step": 5328
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729675458101594e-05,
+      "loss": 0.4262,
+      "step": 5329
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726690270964575e-05,
+      "loss": 0.3671,
+      "step": 5330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723705018724496e-05,
+      "loss": 0.3561,
+      "step": 5331
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720719701648792e-05,
+      "loss": 0.3538,
+      "step": 5332
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717734320004918e-05,
+      "loss": 0.3944,
+      "step": 5333
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714748874060324e-05,
+      "loss": 0.3621,
+      "step": 5334
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071176336408248e-05,
+      "loss": 0.336,
+      "step": 5335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0708777790338845e-05,
+      "loss": 0.3693,
+      "step": 5336
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705792153096894e-05,
+      "loss": 0.387,
+      "step": 5337
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702806452624109e-05,
+      "loss": 0.3863,
+      "step": 5338
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699820689187971e-05,
+      "loss": 0.3868,
+      "step": 5339
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0696834863055973e-05,
+      "loss": 0.3623,
+      "step": 5340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0693848974495609e-05,
+      "loss": 0.3129,
+      "step": 5341
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069086302377438e-05,
+      "loss": 0.3523,
+      "step": 5342
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687877011159791e-05,
+      "loss": 0.3532,
+      "step": 5343
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0684890936919356e-05,
+      "loss": 0.3656,
+      "step": 5344
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681904801320598e-05,
+      "loss": 0.4034,
+      "step": 5345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678918604631028e-05,
+      "loss": 0.375,
+      "step": 5346
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675932347118188e-05,
+      "loss": 0.3616,
+      "step": 5347
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672946029049603e-05,
+      "loss": 0.3662,
+      "step": 5348
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0669959650692818e-05,
+      "loss": 0.377,
+      "step": 5349
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666973212315372e-05,
+      "loss": 0.3975,
+      "step": 5350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0663986714184825e-05,
+      "loss": 0.3623,
+      "step": 5351
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661000156568721e-05,
+      "loss": 0.3484,
+      "step": 5352
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658013539734631e-05,
+      "loss": 0.4063,
+      "step": 5353
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0655026863950118e-05,
+      "loss": 0.3598,
+      "step": 5354
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065204012948275e-05,
+      "loss": 0.3515,
+      "step": 5355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0649053336600113e-05,
+      "loss": 0.4031,
+      "step": 5356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646066485569779e-05,
+      "loss": 0.3652,
+      "step": 5357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643079576659342e-05,
+      "loss": 0.3314,
+      "step": 5358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064009261013639e-05,
+      "loss": 0.3295,
+      "step": 5359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637105586268524e-05,
+      "loss": 0.3641,
+      "step": 5360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0634118505323349e-05,
+      "loss": 0.3606,
+      "step": 5361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631131367568467e-05,
+      "loss": 0.3535,
+      "step": 5362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628144173271495e-05,
+      "loss": 0.3431,
+      "step": 5363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0625156922700051e-05,
+      "loss": 0.3701,
+      "step": 5364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062216961612176e-05,
+      "loss": 0.3417,
+      "step": 5365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619182253804241e-05,
+      "loss": 0.4104,
+      "step": 5366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0616194836015141e-05,
+      "loss": 0.326,
+      "step": 5367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613207363022086e-05,
+      "loss": 0.3985,
+      "step": 5368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0610219835092728e-05,
+      "loss": 0.3845,
+      "step": 5369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607232252494707e-05,
+      "loss": 0.3917,
+      "step": 5370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604244615495684e-05,
+      "loss": 0.4211,
+      "step": 5371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601256924363314e-05,
+      "loss": 0.3304,
+      "step": 5372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598269179365257e-05,
+      "loss": 0.3902,
+      "step": 5373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595281380769182e-05,
+      "loss": 0.3509,
+      "step": 5374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059229352884276e-05,
+      "loss": 0.3561,
+      "step": 5375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589305623853675e-05,
+      "loss": 0.4103,
+      "step": 5376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0586317666069598e-05,
+      "loss": 0.3766,
+      "step": 5377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0583329655758224e-05,
+      "loss": 0.3465,
+      "step": 5378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058034159318724e-05,
+      "loss": 0.3531,
+      "step": 5379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0577353478624342e-05,
+      "loss": 0.3372,
+      "step": 5380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574365312337235e-05,
+      "loss": 0.3248,
+      "step": 5381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0571377094593616e-05,
+      "loss": 0.3798,
+      "step": 5382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0568388825661204e-05,
+      "loss": 0.3834,
+      "step": 5383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565400505807709e-05,
+      "loss": 0.3627,
+      "step": 5384
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562412135300849e-05,
+      "loss": 0.3462,
+      "step": 5385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055942371440835e-05,
+      "loss": 0.3861,
+      "step": 5386
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0556435243397936e-05,
+      "loss": 0.4051,
+      "step": 5387
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553446722537346e-05,
+      "loss": 0.3847,
+      "step": 5388
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550458152094312e-05,
+      "loss": 0.3815,
+      "step": 5389
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547469532336576e-05,
+      "loss": 0.3535,
+      "step": 5390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0544480863531884e-05,
+      "loss": 0.3677,
+      "step": 5391
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541492145947992e-05,
+      "loss": 0.3601,
+      "step": 5392
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538503379852644e-05,
+      "loss": 0.3724,
+      "step": 5393
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0535514565513606e-05,
+      "loss": 0.3652,
+      "step": 5394
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053252570319864e-05,
+      "loss": 0.3629,
+      "step": 5395
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529536793175514e-05,
+      "loss": 0.3849,
+      "step": 5396
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526547835712e-05,
+      "loss": 0.4141,
+      "step": 5397
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523558831075874e-05,
+      "loss": 0.3816,
+      "step": 5398
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052056977953491e-05,
+      "loss": 0.4207,
+      "step": 5399
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0517580681356902e-05,
+      "loss": 0.3968,
+      "step": 5400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0514591536809635e-05,
+      "loss": 0.367,
+      "step": 5401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511602346160903e-05,
+      "loss": 0.3536,
+      "step": 5402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05086131096785e-05,
+      "loss": 0.405,
+      "step": 5403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050562382763023e-05,
+      "loss": 0.4114,
+      "step": 5404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502634500283898e-05,
+      "loss": 0.404,
+      "step": 5405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049964512790731e-05,
+      "loss": 0.3652,
+      "step": 5406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496655710768283e-05,
+      "loss": 0.3779,
+      "step": 5407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0493666249134636e-05,
+      "loss": 0.3877,
+      "step": 5408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490676743274181e-05,
+      "loss": 0.3725,
+      "step": 5409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0487687193454753e-05,
+      "loss": 0.3763,
+      "step": 5410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484697599944178e-05,
+      "loss": 0.3457,
+      "step": 5411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481707963010287e-05,
+      "loss": 0.4094,
+      "step": 5412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0478718282920922e-05,
+      "loss": 0.3052,
+      "step": 5413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475728559943919e-05,
+      "loss": 0.3831,
+      "step": 5414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472738794347121e-05,
+      "loss": 0.3553,
+      "step": 5415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469748986398382e-05,
+      "loss": 0.3521,
+      "step": 5416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0466759136365556e-05,
+      "loss": 0.4183,
+      "step": 5417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046376924451649e-05,
+      "loss": 0.3634,
+      "step": 5418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046077931111905e-05,
+      "loss": 0.3631,
+      "step": 5419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.04577893364411e-05,
+      "loss": 0.3617,
+      "step": 5420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454799320750505e-05,
+      "loss": 0.3509,
+      "step": 5421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451809264315139e-05,
+      "loss": 0.3778,
+      "step": 5422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0448819167402871e-05,
+      "loss": 0.335,
+      "step": 5423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0445829030281584e-05,
+      "loss": 0.3744,
+      "step": 5424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044283885321916e-05,
+      "loss": 0.343,
+      "step": 5425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0439848636483478e-05,
+      "loss": 0.3382,
+      "step": 5426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436858380342434e-05,
+      "loss": 0.3681,
+      "step": 5427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0433868085063918e-05,
+      "loss": 0.365,
+      "step": 5428
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0430877750915828e-05,
+      "loss": 0.3449,
+      "step": 5429
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042788737816606e-05,
+      "loss": 0.4072,
+      "step": 5430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424896967082518e-05,
+      "loss": 0.3902,
+      "step": 5431
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042190651793311e-05,
+      "loss": 0.4243,
+      "step": 5432
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418916030985744e-05,
+      "loss": 0.3699,
+      "step": 5433
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415925506508333e-05,
+      "loss": 0.3662,
+      "step": 5434
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041293494476879e-05,
+      "loss": 0.3665,
+      "step": 5435
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0409944346035047e-05,
+      "loss": 0.36,
+      "step": 5436
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406953710575015e-05,
+      "loss": 0.3854,
+      "step": 5437
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0403963038656623e-05,
+      "loss": 0.3968,
+      "step": 5438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400972330547804e-05,
+      "loss": 0.3669,
+      "step": 5439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039798158651649e-05,
+      "loss": 0.3585,
+      "step": 5440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394990806830616e-05,
+      "loss": 0.3655,
+      "step": 5441
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039199999175812e-05,
+      "loss": 0.3607,
+      "step": 5442
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0389009141566949e-05,
+      "loss": 0.3696,
+      "step": 5443
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038601825652504e-05,
+      "loss": 0.3752,
+      "step": 5444
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383027336900356e-05,
+      "loss": 0.358,
+      "step": 5445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0380036382960837e-05,
+      "loss": 0.3979,
+      "step": 5446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037704539497444e-05,
+      "loss": 0.4256,
+      "step": 5447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0374054373209126e-05,
+      "loss": 0.3658,
+      "step": 5448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371063317932856e-05,
+      "loss": 0.3838,
+      "step": 5449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036807222941359e-05,
+      "loss": 0.4053,
+      "step": 5450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365081107919297e-05,
+      "loss": 0.3646,
+      "step": 5451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362089953717952e-05,
+      "loss": 0.3673,
+      "step": 5452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035909876707752e-05,
+      "loss": 0.3714,
+      "step": 5453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0356107548265985e-05,
+      "loss": 0.3767,
+      "step": 5454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0353116297551315e-05,
+      "loss": 0.3289,
+      "step": 5455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350125015201502e-05,
+      "loss": 0.3661,
+      "step": 5456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347133701484526e-05,
+      "loss": 0.3571,
+      "step": 5457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0344142356668372e-05,
+      "loss": 0.3944,
+      "step": 5458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341150981021034e-05,
+      "loss": 0.3407,
+      "step": 5459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0338159574810502e-05,
+      "loss": 0.3956,
+      "step": 5460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335168138304776e-05,
+      "loss": 0.3835,
+      "step": 5461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332176671771847e-05,
+      "loss": 0.3627,
+      "step": 5462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.032918517547972e-05,
+      "loss": 0.3721,
+      "step": 5463
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326193649696399e-05,
+      "loss": 0.3644,
+      "step": 5464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0323202094689893e-05,
+      "loss": 0.3427,
+      "step": 5465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320210510728201e-05,
+      "loss": 0.4092,
+      "step": 5466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0317218898079341e-05,
+      "loss": 0.3625,
+      "step": 5467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0314227257011334e-05,
+      "loss": 0.403,
+      "step": 5468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0311235587792183e-05,
+      "loss": 0.3654,
+      "step": 5469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308243890689915e-05,
+      "loss": 0.3672,
+      "step": 5470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305252165972548e-05,
+      "loss": 0.3894,
+      "step": 5471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302260413908112e-05,
+      "loss": 0.3331,
+      "step": 5472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299268634764626e-05,
+      "loss": 0.3987,
+      "step": 5473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029627682881012e-05,
+      "loss": 0.3855,
+      "step": 5474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293284996312631e-05,
+      "loss": 0.3783,
+      "step": 5475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029029313754019e-05,
+      "loss": 0.3732,
+      "step": 5476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287301252760833e-05,
+      "loss": 0.331,
+      "step": 5477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284309342242595e-05,
+      "loss": 0.3871,
+      "step": 5478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.028131740625352e-05,
+      "loss": 0.3583,
+      "step": 5479
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027832544506165e-05,
+      "loss": 0.3832,
+      "step": 5480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275333458935031e-05,
+      "loss": 0.3925,
+      "step": 5481
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272341448141709e-05,
+      "loss": 0.3683,
+      "step": 5482
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0269349412949732e-05,
+      "loss": 0.3825,
+      "step": 5483
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266357353627161e-05,
+      "loss": 0.3483,
+      "step": 5484
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0263365270442038e-05,
+      "loss": 0.3701,
+      "step": 5485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260373163662429e-05,
+      "loss": 0.3839,
+      "step": 5486
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257381033556382e-05,
+      "loss": 0.4019,
+      "step": 5487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254388880391968e-05,
+      "loss": 0.3908,
+      "step": 5488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251396704437241e-05,
+      "loss": 0.3975,
+      "step": 5489
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248404505960275e-05,
+      "loss": 0.3738,
+      "step": 5490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245412285229124e-05,
+      "loss": 0.3933,
+      "step": 5491
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242420042511865e-05,
+      "loss": 0.3803,
+      "step": 5492
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0239427778076567e-05,
+      "loss": 0.3959,
+      "step": 5493
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0236435492191301e-05,
+      "loss": 0.35,
+      "step": 5494
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0233443185124143e-05,
+      "loss": 0.411,
+      "step": 5495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230450857143168e-05,
+      "loss": 0.3373,
+      "step": 5496
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227458508516455e-05,
+      "loss": 0.3781,
+      "step": 5497
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0224466139512082e-05,
+      "loss": 0.4176,
+      "step": 5498
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0221473750398133e-05,
+      "loss": 0.3891,
+      "step": 5499
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.021848134144269e-05,
+      "loss": 0.3671,
+      "step": 5500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215488912913838e-05,
+      "loss": 0.3764,
+      "step": 5501
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0212496465079667e-05,
+      "loss": 0.3647,
+      "step": 5502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020950399820826e-05,
+      "loss": 0.3327,
+      "step": 5503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206511512567717e-05,
+      "loss": 0.3658,
+      "step": 5504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020351900842612e-05,
+      "loss": 0.3501,
+      "step": 5505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0200526486051569e-05,
+      "loss": 0.3656,
+      "step": 5506
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197533945712154e-05,
+      "loss": 0.3301,
+      "step": 5507
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019454138767598e-05,
+      "loss": 0.4109,
+      "step": 5508
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191548812211143e-05,
+      "loss": 0.3589,
+      "step": 5509
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188556219585738e-05,
+      "loss": 0.36,
+      "step": 5510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0185563610067872e-05,
+      "loss": 0.3985,
+      "step": 5511
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182570983925648e-05,
+      "loss": 0.3607,
+      "step": 5512
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017957834142717e-05,
+      "loss": 0.3632,
+      "step": 5513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176585682840543e-05,
+      "loss": 0.4074,
+      "step": 5514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0173593008433876e-05,
+      "loss": 0.3701,
+      "step": 5515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170600318475276e-05,
+      "loss": 0.3883,
+      "step": 5516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167607613232856e-05,
+      "loss": 0.3645,
+      "step": 5517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016461489297473e-05,
+      "loss": 0.38,
+      "step": 5518
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161622157969005e-05,
+      "loss": 0.3666,
+      "step": 5519
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01586294084838e-05,
+      "loss": 0.3614,
+      "step": 5520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155636644787231e-05,
+      "loss": 0.3899,
+      "step": 5521
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152643867147414e-05,
+      "loss": 0.371,
+      "step": 5522
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014965107583246e-05,
+      "loss": 0.3427,
+      "step": 5523
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146658271110504e-05,
+      "loss": 0.4128,
+      "step": 5524
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143665453249656e-05,
+      "loss": 0.3652,
+      "step": 5525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014067262251804e-05,
+      "loss": 0.3603,
+      "step": 5526
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013767977918378e-05,
+      "loss": 0.3779,
+      "step": 5527
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134686923514998e-05,
+      "loss": 0.3619,
+      "step": 5528
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0131694055779825e-05,
+      "loss": 0.344,
+      "step": 5529
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128701176246383e-05,
+      "loss": 0.3595,
+      "step": 5530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125708285182799e-05,
+      "loss": 0.3803,
+      "step": 5531
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01227153828572e-05,
+      "loss": 0.4047,
+      "step": 5532
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119722469537721e-05,
+      "loss": 0.3306,
+      "step": 5533
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011672954549249e-05,
+      "loss": 0.4124,
+      "step": 5534
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113736610989637e-05,
+      "loss": 0.3292,
+      "step": 5535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0110743666297296e-05,
+      "loss": 0.3943,
+      "step": 5536
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.01077507116836e-05,
+      "loss": 0.3456,
+      "step": 5537
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104757747416685e-05,
+      "loss": 0.4118,
+      "step": 5538
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101764773764677e-05,
+      "loss": 0.354,
+      "step": 5539
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098771790995727e-05,
+      "loss": 0.3549,
+      "step": 5540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009577879937796e-05,
+      "loss": 0.361,
+      "step": 5541
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009278579917952e-05,
+      "loss": 0.3449,
+      "step": 5542
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089792790668538e-05,
+      "loss": 0.3383,
+      "step": 5543
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008679977411316e-05,
+      "loss": 0.3926,
+      "step": 5544
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083806749781527e-05,
+      "loss": 0.3596,
+      "step": 5545
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008081371794177e-05,
+      "loss": 0.3928,
+      "step": 5546
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007782067886204e-05,
+      "loss": 0.4007,
+      "step": 5547
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074827632810474e-05,
+      "loss": 0.3474,
+      "step": 5548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071834580055213e-05,
+      "loss": 0.3841,
+      "step": 5549
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068841520864405e-05,
+      "loss": 0.3691,
+      "step": 5550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065848455506191e-05,
+      "loss": 0.3838,
+      "step": 5551
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062855384248712e-05,
+      "loss": 0.3502,
+      "step": 5552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059862307360118e-05,
+      "loss": 0.389,
+      "step": 5553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056869225108552e-05,
+      "loss": 0.4242,
+      "step": 5554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053876137762157e-05,
+      "loss": 0.3252,
+      "step": 5555
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050883045589084e-05,
+      "loss": 0.3619,
+      "step": 5556
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047889948857477e-05,
+      "loss": 0.3558,
+      "step": 5557
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044896847835479e-05,
+      "loss": 0.4035,
+      "step": 5558
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041903742791246e-05,
+      "loss": 0.3964,
+      "step": 5559
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038910633992918e-05,
+      "loss": 0.3838,
+      "step": 5560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003591752170865e-05,
+      "loss": 0.4088,
+      "step": 5561
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032924406206582e-05,
+      "loss": 0.4102,
+      "step": 5562
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029931287754868e-05,
+      "loss": 0.357,
+      "step": 5563
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0026938166621656e-05,
+      "loss": 0.3841,
+      "step": 5564
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023945043075094e-05,
+      "loss": 0.3768,
+      "step": 5565
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020951917383337e-05,
+      "loss": 0.3859,
+      "step": 5566
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017958789814526e-05,
+      "loss": 0.3532,
+      "step": 5567
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014965660636812e-05,
+      "loss": 0.347,
+      "step": 5568
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011972530118352e-05,
+      "loss": 0.3783,
+      "step": 5569
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0008979398527292e-05,
+      "loss": 0.3477,
+      "step": 5570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005986266131778e-05,
+      "loss": 0.3456,
+      "step": 5571
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0002993133199966e-05,
+      "loss": 0.3826,
+      "step": 5572
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.3886,
+      "step": 5573
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997006866800037e-06,
+      "loss": 0.3816,
+      "step": 5574
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994013733868224e-06,
+      "loss": 0.3483,
+      "step": 5575
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.991020601472713e-06,
+      "loss": 0.3895,
+      "step": 5576
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.988027469881652e-06,
+      "loss": 0.3506,
+      "step": 5577
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98503433936319e-06,
+      "loss": 0.3776,
+      "step": 5578
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.982041210185479e-06,
+      "loss": 0.3397,
+      "step": 5579
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97904808261667e-06,
+      "loss": 0.3529,
+      "step": 5580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976054956924908e-06,
+      "loss": 0.3749,
+      "step": 5581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973061833378347e-06,
+      "loss": 0.3692,
+      "step": 5582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970068712245135e-06,
+      "loss": 0.3418,
+      "step": 5583
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96707559379342e-06,
+      "loss": 0.3913,
+      "step": 5584
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.964082478291358e-06,
+      "loss": 0.3661,
+      "step": 5585
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961089366007085e-06,
+      "loss": 0.3868,
+      "step": 5586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958096257208759e-06,
+      "loss": 0.3777,
+      "step": 5587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955103152164521e-06,
+      "loss": 0.3571,
+      "step": 5588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95211005114253e-06,
+      "loss": 0.3888,
+      "step": 5589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94911695441092e-06,
+      "loss": 0.3591,
+      "step": 5590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946123862237847e-06,
+      "loss": 0.3762,
+      "step": 5591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943130774891451e-06,
+      "loss": 0.394,
+      "step": 5592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940137692639882e-06,
+      "loss": 0.3831,
+      "step": 5593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937144615751293e-06,
+      "loss": 0.3853,
+      "step": 5594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934151544493812e-06,
+      "loss": 0.3484,
+      "step": 5595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.931158479135597e-06,
+      "loss": 0.3952,
+      "step": 5596
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.928165419944788e-06,
+      "loss": 0.4029,
+      "step": 5597
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925172367189528e-06,
+      "loss": 0.403,
+      "step": 5598
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.922179321137962e-06,
+      "loss": 0.3775,
+      "step": 5599
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919186282058232e-06,
+      "loss": 0.35,
+      "step": 5600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916193250218477e-06,
+      "loss": 0.3767,
+      "step": 5601
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91320022588684e-06,
+      "loss": 0.3685,
+      "step": 5602
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910207209331465e-06,
+      "loss": 0.3216,
+      "step": 5603
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.907214200820486e-06,
+      "loss": 0.3994,
+      "step": 5604
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904221200622042e-06,
+      "loss": 0.3578,
+      "step": 5605
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901228209004276e-06,
+      "loss": 0.3504,
+      "step": 5606
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898235226235323e-06,
+      "loss": 0.3612,
+      "step": 5607
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895242252583322e-06,
+      "loss": 0.3809,
+      "step": 5608
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.892249288316404e-06,
+      "loss": 0.3789,
+      "step": 5609
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.889256333702707e-06,
+      "loss": 0.3461,
+      "step": 5610
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886263389010365e-06,
+      "loss": 0.3532,
+      "step": 5611
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883270454507514e-06,
+      "loss": 0.3588,
+      "step": 5612
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880277530462282e-06,
+      "loss": 0.3633,
+      "step": 5613
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877284617142802e-06,
+      "loss": 0.3673,
+      "step": 5614
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874291714817206e-06,
+      "loss": 0.3325,
+      "step": 5615
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871298823753622e-06,
+      "loss": 0.3821,
+      "step": 5616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868305944220178e-06,
+      "loss": 0.3569,
+      "step": 5617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865313076485004e-06,
+      "loss": 0.3453,
+      "step": 5618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862320220816223e-06,
+      "loss": 0.3554,
+      "step": 5619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.85932737748196e-06,
+      "loss": 0.3585,
+      "step": 5620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856334546750349e-06,
+      "loss": 0.3627,
+      "step": 5621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853341728889501e-06,
+      "loss": 0.3718,
+      "step": 5622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.850348924167541e-06,
+      "loss": 0.3556,
+      "step": 5623
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84735613285259e-06,
+      "loss": 0.3627,
+      "step": 5624
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84436335521277e-06,
+      "loss": 0.3993,
+      "step": 5625
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841370591516206e-06,
+      "loss": 0.3775,
+      "step": 5626
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.838377842031e-06,
+      "loss": 0.3457,
+      "step": 5627
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.835385107025272e-06,
+      "loss": 0.4164,
+      "step": 5628
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832392386767145e-06,
+      "loss": 0.3682,
+      "step": 5629
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829399681524729e-06,
+      "loss": 0.4026,
+      "step": 5630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826406991566127e-06,
+      "loss": 0.3457,
+      "step": 5631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823414317159459e-06,
+      "loss": 0.4262,
+      "step": 5632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.820421658572832e-06,
+      "loss": 0.3475,
+      "step": 5633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817429016074352e-06,
+      "loss": 0.3969,
+      "step": 5634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81443638993213e-06,
+      "loss": 0.3587,
+      "step": 5635
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811443780414263e-06,
+      "loss": 0.3711,
+      "step": 5636
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80845118778886e-06,
+      "loss": 0.3803,
+      "step": 5637
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80545861232402e-06,
+      "loss": 0.3756,
+      "step": 5638
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802466054287846e-06,
+      "loss": 0.3711,
+      "step": 5639
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.799473513948435e-06,
+      "loss": 0.3496,
+      "step": 5640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.796480991573884e-06,
+      "loss": 0.404,
+      "step": 5641
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793488487432288e-06,
+      "loss": 0.3666,
+      "step": 5642
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.79049600179174e-06,
+      "loss": 0.3796,
+      "step": 5643
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.787503534920338e-06,
+      "loss": 0.3532,
+      "step": 5644
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.784511087086165e-06,
+      "loss": 0.3349,
+      "step": 5645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.781518658557314e-06,
+      "loss": 0.3597,
+      "step": 5646
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77852624960187e-06,
+      "loss": 0.3616,
+      "step": 5647
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77553386048792e-06,
+      "loss": 0.3915,
+      "step": 5648
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77254149148355e-06,
+      "loss": 0.3791,
+      "step": 5649
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769549142856836e-06,
+      "loss": 0.3624,
+      "step": 5650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76655681487586e-06,
+      "loss": 0.365,
+      "step": 5651
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7635645078087e-06,
+      "loss": 0.3717,
+      "step": 5652
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.760572221923436e-06,
+      "loss": 0.3959,
+      "step": 5653
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757579957488138e-06,
+      "loss": 0.4003,
+      "step": 5654
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75458771477088e-06,
+      "loss": 0.3706,
+      "step": 5655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751595494039728e-06,
+      "loss": 0.4112,
+      "step": 5656
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748603295562757e-06,
+      "loss": 0.3683,
+      "step": 5657
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.745611119608036e-06,
+      "loss": 0.3941,
+      "step": 5658
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742618966443622e-06,
+      "loss": 0.3338,
+      "step": 5659
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739626836337574e-06,
+      "loss": 0.3324,
+      "step": 5660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.736634729557962e-06,
+      "loss": 0.3484,
+      "step": 5661
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733642646372845e-06,
+      "loss": 0.4121,
+      "step": 5662
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73065058705027e-06,
+      "loss": 0.3621,
+      "step": 5663
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727658551858293e-06,
+      "loss": 0.3799,
+      "step": 5664
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724666541064972e-06,
+      "loss": 0.3069,
+      "step": 5665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.721674554938351e-06,
+      "loss": 0.3269,
+      "step": 5666
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718682593746484e-06,
+      "loss": 0.3589,
+      "step": 5667
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715690657757408e-06,
+      "loss": 0.4174,
+      "step": 5668
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.712698747239172e-06,
+      "loss": 0.3863,
+      "step": 5669
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709706862459812e-06,
+      "loss": 0.3988,
+      "step": 5670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70671500368737e-06,
+      "loss": 0.373,
+      "step": 5671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703723171189881e-06,
+      "loss": 0.3663,
+      "step": 5672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700731365235377e-06,
+      "loss": 0.3457,
+      "step": 5673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697739586091892e-06,
+      "loss": 0.3843,
+      "step": 5674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694747834027453e-06,
+      "loss": 0.3275,
+      "step": 5675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691756109310089e-06,
+      "loss": 0.3785,
+      "step": 5676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68876441220782e-06,
+      "loss": 0.4174,
+      "step": 5677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685772742988671e-06,
+      "loss": 0.3645,
+      "step": 5678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682781101920659e-06,
+      "loss": 0.3711,
+      "step": 5679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.6797894892718e-06,
+      "loss": 0.395,
+      "step": 5680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.676797905310114e-06,
+      "loss": 0.3634,
+      "step": 5681
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.673806350303603e-06,
+      "loss": 0.3809,
+      "step": 5682
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670814824520283e-06,
+      "loss": 0.3421,
+      "step": 5683
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667823328228156e-06,
+      "loss": 0.346,
+      "step": 5684
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664831861695229e-06,
+      "loss": 0.3669,
+      "step": 5685
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.6618404251895e-06,
+      "loss": 0.3676,
+      "step": 5686
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658849018978968e-06,
+      "loss": 0.4247,
+      "step": 5687
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65585764333163e-06,
+      "loss": 0.3945,
+      "step": 5688
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652866298515475e-06,
+      "loss": 0.3398,
+      "step": 5689
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649874984798502e-06,
+      "loss": 0.394,
+      "step": 5690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646883702448688e-06,
+      "loss": 0.3321,
+      "step": 5691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.643892451734018e-06,
+      "loss": 0.3536,
+      "step": 5692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640901232922479e-06,
+      "loss": 0.3796,
+      "step": 5693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637910046282053e-06,
+      "loss": 0.378,
+      "step": 5694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634918892080706e-06,
+      "loss": 0.3677,
+      "step": 5695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631927770586412e-06,
+      "loss": 0.3675,
+      "step": 5696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628936682067147e-06,
+      "loss": 0.3634,
+      "step": 5697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.625945626790876e-06,
+      "loss": 0.3631,
+      "step": 5698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622954605025565e-06,
+      "loss": 0.3691,
+      "step": 5699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619963617039167e-06,
+      "loss": 0.3949,
+      "step": 5700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616972663099648e-06,
+      "loss": 0.3521,
+      "step": 5701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.613981743474958e-06,
+      "loss": 0.3502,
+      "step": 5702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610990858433056e-06,
+      "loss": 0.3486,
+      "step": 5703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608000008241881e-06,
+      "loss": 0.4009,
+      "step": 5704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605009193169387e-06,
+      "loss": 0.3476,
+      "step": 5705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602018413483512e-06,
+      "loss": 0.3846,
+      "step": 5706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599027669452196e-06,
+      "loss": 0.3711,
+      "step": 5707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59603696134338e-06,
+      "loss": 0.404,
+      "step": 5708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593046289424989e-06,
+      "loss": 0.4014,
+      "step": 5709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590055653964957e-06,
+      "loss": 0.3863,
+      "step": 5710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58706505523121e-06,
+      "loss": 0.3637,
+      "step": 5711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584074493491674e-06,
+      "loss": 0.3946,
+      "step": 5712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.581083969014261e-06,
+      "loss": 0.3527,
+      "step": 5713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578093482066894e-06,
+      "loss": 0.3401,
+      "step": 5714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575103032917485e-06,
+      "loss": 0.3516,
+      "step": 5715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572112621833941e-06,
+      "loss": 0.3487,
+      "step": 5716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.569122249084177e-06,
+      "loss": 0.3477,
+      "step": 5717
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566131914936085e-06,
+      "loss": 0.3629,
+      "step": 5718
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.563141619657568e-06,
+      "loss": 0.3647,
+      "step": 5719
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.560151363516525e-06,
+      "loss": 0.38,
+      "step": 5720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557161146780842e-06,
+      "loss": 0.3486,
+      "step": 5721
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.554170969718421e-06,
+      "loss": 0.3695,
+      "step": 5722
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551180832597132e-06,
+      "loss": 0.3754,
+      "step": 5723
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.548190735684866e-06,
+      "loss": 0.3697,
+      "step": 5724
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545200679249496e-06,
+      "loss": 0.3929,
+      "step": 5725
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542210663558905e-06,
+      "loss": 0.3554,
+      "step": 5726
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.539220688880954e-06,
+      "loss": 0.3421,
+      "step": 5727
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536230755483513e-06,
+      "loss": 0.3436,
+      "step": 5728
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533240863634447e-06,
+      "loss": 0.4068,
+      "step": 5729
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530251013601618e-06,
+      "loss": 0.3719,
+      "step": 5730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527261205652884e-06,
+      "loss": 0.3617,
+      "step": 5731
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524271440056085e-06,
+      "loss": 0.3428,
+      "step": 5732
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521281717079082e-06,
+      "loss": 0.4179,
+      "step": 5733
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.518292036989713e-06,
+      "loss": 0.3678,
+      "step": 5734
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.515302400055828e-06,
+      "loss": 0.3566,
+      "step": 5735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51231280654525e-06,
+      "loss": 0.4074,
+      "step": 5736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50932325672582e-06,
+      "loss": 0.3679,
+      "step": 5737
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50633375086537e-06,
+      "loss": 0.391,
+      "step": 5738
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503344289231717e-06,
+      "loss": 0.3637,
+      "step": 5739
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500354872092694e-06,
+      "loss": 0.3561,
+      "step": 5740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.497365499716106e-06,
+      "loss": 0.3582,
+      "step": 5741
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494376172369771e-06,
+      "loss": 0.3645,
+      "step": 5742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491386890321501e-06,
+      "loss": 0.3911,
+      "step": 5743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488397653839102e-06,
+      "loss": 0.3788,
+      "step": 5744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485408463190368e-06,
+      "loss": 0.3879,
+      "step": 5745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.4824193186431e-06,
+      "loss": 0.3416,
+      "step": 5746
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479430220465092e-06,
+      "loss": 0.3614,
+      "step": 5747
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.476441168924131e-06,
+      "loss": 0.3807,
+      "step": 5748
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.473452164288005e-06,
+      "loss": 0.3982,
+      "step": 5749
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470463206824489e-06,
+      "loss": 0.4017,
+      "step": 5750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467474296801362e-06,
+      "loss": 0.3726,
+      "step": 5751
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.464485434486395e-06,
+      "loss": 0.3723,
+      "step": 5752
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46149662014736e-06,
+      "loss": 0.3766,
+      "step": 5753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458507854052013e-06,
+      "loss": 0.4323,
+      "step": 5754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.455519136468117e-06,
+      "loss": 0.3557,
+      "step": 5755
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.452530467663426e-06,
+      "loss": 0.3422,
+      "step": 5756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449541847905688e-06,
+      "loss": 0.3779,
+      "step": 5757
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.446553277462657e-06,
+      "loss": 0.3951,
+      "step": 5758
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.443564756602065e-06,
+      "loss": 0.3465,
+      "step": 5759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440576285591654e-06,
+      "loss": 0.3717,
+      "step": 5760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437587864699151e-06,
+      "loss": 0.416,
+      "step": 5761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434599494192291e-06,
+      "loss": 0.3274,
+      "step": 5762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.4316111743388e-06,
+      "loss": 0.3551,
+      "step": 5763
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428622905406385e-06,
+      "loss": 0.3896,
+      "step": 5764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425634687662768e-06,
+      "loss": 0.3406,
+      "step": 5765
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.422646521375658e-06,
+      "loss": 0.3476,
+      "step": 5766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.419658406812766e-06,
+      "loss": 0.3601,
+      "step": 5767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.41667034424178e-06,
+      "loss": 0.3663,
+      "step": 5768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.413682333930403e-06,
+      "loss": 0.3607,
+      "step": 5769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410694376146327e-06,
+      "loss": 0.3522,
+      "step": 5770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407706471157238e-06,
+      "loss": 0.3721,
+      "step": 5771
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40471861923082e-06,
+      "loss": 0.3391,
+      "step": 5772
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401730820634746e-06,
+      "loss": 0.3555,
+      "step": 5773
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.398743075636688e-06,
+      "loss": 0.3956,
+      "step": 5774
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395755384504316e-06,
+      "loss": 0.39,
+      "step": 5775
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392767747505294e-06,
+      "loss": 0.3805,
+      "step": 5776
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389780164907276e-06,
+      "loss": 0.3644,
+      "step": 5777
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386792636977915e-06,
+      "loss": 0.3392,
+      "step": 5778
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.383805163984862e-06,
+      "loss": 0.4111,
+      "step": 5779
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38081774619576e-06,
+      "loss": 0.379,
+      "step": 5780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.377830383878246e-06,
+      "loss": 0.385,
+      "step": 5781
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374843077299952e-06,
+      "loss": 0.4021,
+      "step": 5782
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.371855826728508e-06,
+      "loss": 0.376,
+      "step": 5783
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368868632431534e-06,
+      "loss": 0.3634,
+      "step": 5784
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.365881494676656e-06,
+      "loss": 0.3849,
+      "step": 5785
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362894413731478e-06,
+      "loss": 0.4003,
+      "step": 5786
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359907389863613e-06,
+      "loss": 0.4407,
+      "step": 5787
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356920423340662e-06,
+      "loss": 0.3856,
+      "step": 5788
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353933514430225e-06,
+      "loss": 0.3585,
+      "step": 5789
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.350946663399894e-06,
+      "loss": 0.364,
+      "step": 5790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347959870517253e-06,
+      "loss": 0.3942,
+      "step": 5791
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.344973136049887e-06,
+      "loss": 0.3581,
+      "step": 5792
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.34198646026537e-06,
+      "loss": 0.3319,
+      "step": 5793
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338999843431282e-06,
+      "loss": 0.3888,
+      "step": 5794
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33601328581518e-06,
+      "loss": 0.401,
+      "step": 5795
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33302678768463e-06,
+      "loss": 0.3745,
+      "step": 5796
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.330040349307185e-06,
+      "loss": 0.3503,
+      "step": 5797
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.327053970950397e-06,
+      "loss": 0.4029,
+      "step": 5798
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324067652881817e-06,
+      "loss": 0.3926,
+      "step": 5799
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321081395368973e-06,
+      "loss": 0.3626,
+      "step": 5800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318095198679405e-06,
+      "loss": 0.3647,
+      "step": 5801
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.315109063080642e-06,
+      "loss": 0.4051,
+      "step": 5802
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312122988840209e-06,
+      "loss": 0.3717,
+      "step": 5803
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309136976225623e-06,
+      "loss": 0.3514,
+      "step": 5804
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.306151025504393e-06,
+      "loss": 0.46,
+      "step": 5805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303165136944028e-06,
+      "loss": 0.353,
+      "step": 5806
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30017931081203e-06,
+      "loss": 0.3663,
+      "step": 5807
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297193547375895e-06,
+      "loss": 0.3331,
+      "step": 5808
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29420784690311e-06,
+      "loss": 0.4071,
+      "step": 5809
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29122220966116e-06,
+      "loss": 0.3909,
+      "step": 5810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288236635917524e-06,
+      "loss": 0.3736,
+      "step": 5811
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285251125939678e-06,
+      "loss": 0.3515,
+      "step": 5812
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282265679995087e-06,
+      "loss": 0.3648,
+      "step": 5813
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279280298351211e-06,
+      "loss": 0.3281,
+      "step": 5814
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276294981275506e-06,
+      "loss": 0.3779,
+      "step": 5815
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273309729035426e-06,
+      "loss": 0.3742,
+      "step": 5816
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270324541898412e-06,
+      "loss": 0.3678,
+      "step": 5817
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.267339420131901e-06,
+      "loss": 0.4086,
+      "step": 5818
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264354364003327e-06,
+      "loss": 0.3527,
+      "step": 5819
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261369373780118e-06,
+      "loss": 0.3617,
+      "step": 5820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258384449729691e-06,
+      "loss": 0.3835,
+      "step": 5821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255399592119468e-06,
+      "loss": 0.3773,
+      "step": 5822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.252414801216848e-06,
+      "loss": 0.3584,
+      "step": 5823
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249430077289241e-06,
+      "loss": 0.3615,
+      "step": 5824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24644542060404e-06,
+      "loss": 0.3495,
+      "step": 5825
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.243460831428641e-06,
+      "loss": 0.3932,
+      "step": 5826
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.240476310030423e-06,
+      "loss": 0.3782,
+      "step": 5827
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.237491856676768e-06,
+      "loss": 0.364,
+      "step": 5828
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234507471635043e-06,
+      "loss": 0.3487,
+      "step": 5829
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231523155172622e-06,
+      "loss": 0.3589,
+      "step": 5830
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228538907556867e-06,
+      "loss": 0.3822,
+      "step": 5831
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225554729055123e-06,
+      "loss": 0.3487,
+      "step": 5832
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22257061993474e-06,
+      "loss": 0.3467,
+      "step": 5833
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219586580463065e-06,
+      "loss": 0.3693,
+      "step": 5834
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.216602610907434e-06,
+      "loss": 0.3738,
+      "step": 5835
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21361871153517e-06,
+      "loss": 0.3752,
+      "step": 5836
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210634882613595e-06,
+      "loss": 0.3705,
+      "step": 5837
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207651124410034e-06,
+      "loss": 0.3817,
+      "step": 5838
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204667437191792e-06,
+      "loss": 0.3539,
+      "step": 5839
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.201683821226176e-06,
+      "loss": 0.3596,
+      "step": 5840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19870027678048e-06,
+      "loss": 0.3628,
+      "step": 5841
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195716804121995e-06,
+      "loss": 0.4129,
+      "step": 5842
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192733403518008e-06,
+      "loss": 0.354,
+      "step": 5843
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189750075235797e-06,
+      "loss": 0.3492,
+      "step": 5844
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.186766819542635e-06,
+      "loss": 0.3905,
+      "step": 5845
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183783636705785e-06,
+      "loss": 0.3332,
+      "step": 5846
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180800526992506e-06,
+      "loss": 0.3809,
+      "step": 5847
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177817490670049e-06,
+      "loss": 0.3954,
+      "step": 5848
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174834528005666e-06,
+      "loss": 0.3639,
+      "step": 5849
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171851639266587e-06,
+      "loss": 0.3813,
+      "step": 5850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16886882472005e-06,
+      "loss": 0.408,
+      "step": 5851
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16588608463328e-06,
+      "loss": 0.3755,
+      "step": 5852
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162903419273496e-06,
+      "loss": 0.4047,
+      "step": 5853
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.159920828907913e-06,
+      "loss": 0.346,
+      "step": 5854
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.156938313803729e-06,
+      "loss": 0.3805,
+      "step": 5855
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15395587422815e-06,
+      "loss": 0.3635,
+      "step": 5856
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150973510448365e-06,
+      "loss": 0.3587,
+      "step": 5857
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147991222731562e-06,
+      "loss": 0.3562,
+      "step": 5858
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145009011344918e-06,
+      "loss": 0.3357,
+      "step": 5859
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142026876555602e-06,
+      "loss": 0.3595,
+      "step": 5860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.139044818630784e-06,
+      "loss": 0.3788,
+      "step": 5861
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.136062837837617e-06,
+      "loss": 0.4071,
+      "step": 5862
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13308093444326e-06,
+      "loss": 0.377,
+      "step": 5863
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130099108714849e-06,
+      "loss": 0.3386,
+      "step": 5864
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12711736091952e-06,
+      "loss": 0.3709,
+      "step": 5865
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12413569132441e-06,
+      "loss": 0.3978,
+      "step": 5866
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121154100196645e-06,
+      "loss": 0.3957,
+      "step": 5867
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11817258780333e-06,
+      "loss": 0.3758,
+      "step": 5868
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115191154411578e-06,
+      "loss": 0.3991,
+      "step": 5869
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112209800288496e-06,
+      "loss": 0.3482,
+      "step": 5870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109228525701174e-06,
+      "loss": 0.3607,
+      "step": 5871
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106247330916707e-06,
+      "loss": 0.3453,
+      "step": 5872
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.103266216202163e-06,
+      "loss": 0.3697,
+      "step": 5873
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100285181824623e-06,
+      "loss": 0.368,
+      "step": 5874
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.097304228051157e-06,
+      "loss": 0.3367,
+      "step": 5875
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.094323355148822e-06,
+      "loss": 0.3609,
+      "step": 5876
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091342563384661e-06,
+      "loss": 0.4015,
+      "step": 5877
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088361853025728e-06,
+      "loss": 0.3795,
+      "step": 5878
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.085381224339059e-06,
+      "loss": 0.3727,
+      "step": 5879
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.082400677591682e-06,
+      "loss": 0.3808,
+      "step": 5880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079420213050622e-06,
+      "loss": 0.3828,
+      "step": 5881
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.076439830982891e-06,
+      "loss": 0.3628,
+      "step": 5882
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.073459531655498e-06,
+      "loss": 0.3612,
+      "step": 5883
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070479315335443e-06,
+      "loss": 0.3285,
+      "step": 5884
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067499182289721e-06,
+      "loss": 0.3971,
+      "step": 5885
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064519132785318e-06,
+      "loss": 0.366,
+      "step": 5886
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061539167089209e-06,
+      "loss": 0.4096,
+      "step": 5887
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058559285468365e-06,
+      "loss": 0.3775,
+      "step": 5888
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05557948818975e-06,
+      "loss": 0.4182,
+      "step": 5889
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052599775520321e-06,
+      "loss": 0.3987,
+      "step": 5890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049620147727024e-06,
+      "loss": 0.3551,
+      "step": 5891
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046640605076799e-06,
+      "loss": 0.3758,
+      "step": 5892
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043661147836578e-06,
+      "loss": 0.4034,
+      "step": 5893
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.040681776273286e-06,
+      "loss": 0.3914,
+      "step": 5894
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037702490653849e-06,
+      "loss": 0.3804,
+      "step": 5895
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.034723291245163e-06,
+      "loss": 0.4048,
+      "step": 5896
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03174417831414e-06,
+      "loss": 0.376,
+      "step": 5897
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028765152127664e-06,
+      "loss": 0.3519,
+      "step": 5898
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.025786212952636e-06,
+      "loss": 0.4319,
+      "step": 5899
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022807361055922e-06,
+      "loss": 0.3782,
+      "step": 5900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.4127,
+      "step": 5901
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01684992016492e-06,
+      "loss": 0.3881,
+      "step": 5902
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013871331704354e-06,
+      "loss": 0.3715,
+      "step": 5903
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010892831589548e-06,
+      "loss": 0.3437,
+      "step": 5904
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007914420087326e-06,
+      "loss": 0.3524,
+      "step": 5905
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004936097464533e-06,
+      "loss": 0.3792,
+      "step": 5906
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001957863987987e-06,
+      "loss": 0.3947,
+      "step": 5907
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998979719924509e-06,
+      "loss": 0.3844,
+      "step": 5908
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.996001665540896e-06,
+      "loss": 0.3644,
+      "step": 5909
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993023701103954e-06,
+      "loss": 0.3419,
+      "step": 5910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990045826880475e-06,
+      "loss": 0.3801,
+      "step": 5911
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98706804313724e-06,
+      "loss": 0.3493,
+      "step": 5912
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984090350141027e-06,
+      "loss": 0.366,
+      "step": 5913
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.981112748158599e-06,
+      "loss": 0.3719,
+      "step": 5914
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.978135237456714e-06,
+      "loss": 0.3896,
+      "step": 5915
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.975157818302127e-06,
+      "loss": 0.3365,
+      "step": 5916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972180490961581e-06,
+      "loss": 0.3572,
+      "step": 5917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969203255701805e-06,
+      "loss": 0.3577,
+      "step": 5918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966226112789528e-06,
+      "loss": 0.3446,
+      "step": 5919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.963249062491467e-06,
+      "loss": 0.4009,
+      "step": 5920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.960272105074332e-06,
+      "loss": 0.3714,
+      "step": 5921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957295240804824e-06,
+      "loss": 0.3773,
+      "step": 5922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954318469949635e-06,
+      "loss": 0.3692,
+      "step": 5923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95134179277545e-06,
+      "loss": 0.3378,
+      "step": 5924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948365209548943e-06,
+      "loss": 0.3408,
+      "step": 5925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945388720536782e-06,
+      "loss": 0.3571,
+      "step": 5926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.942412326005635e-06,
+      "loss": 0.3731,
+      "step": 5927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.939436026222137e-06,
+      "loss": 0.3669,
+      "step": 5928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.936459821452941e-06,
+      "loss": 0.3906,
+      "step": 5929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.933483711964674e-06,
+      "loss": 0.423,
+      "step": 5930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930507698023973e-06,
+      "loss": 0.3773,
+      "step": 5931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92753177989744e-06,
+      "loss": 0.3563,
+      "step": 5932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92455595785169e-06,
+      "loss": 0.33,
+      "step": 5933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92158023215332e-06,
+      "loss": 0.3625,
+      "step": 5934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.918604603068923e-06,
+      "loss": 0.355,
+      "step": 5935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915629070865086e-06,
+      "loss": 0.3407,
+      "step": 5936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912653635808372e-06,
+      "loss": 0.3565,
+      "step": 5937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.909678298165347e-06,
+      "loss": 0.3712,
+      "step": 5938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.906703058202572e-06,
+      "loss": 0.3979,
+      "step": 5939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903727916186599e-06,
+      "loss": 0.3458,
+      "step": 5940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90075287238395e-06,
+      "loss": 0.3724,
+      "step": 5941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89777792706117e-06,
+      "loss": 0.3683,
+      "step": 5942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894803080484772e-06,
+      "loss": 0.4205,
+      "step": 5943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891828332921272e-06,
+      "loss": 0.3869,
+      "step": 5944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888853684637172e-06,
+      "loss": 0.3619,
+      "step": 5945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.885879135898964e-06,
+      "loss": 0.3858,
+      "step": 5946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882904686973135e-06,
+      "loss": 0.3569,
+      "step": 5947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879930338126159e-06,
+      "loss": 0.3623,
+      "step": 5948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876956089624508e-06,
+      "loss": 0.3939,
+      "step": 5949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873981941734636e-06,
+      "loss": 0.3472,
+      "step": 5950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871007894722995e-06,
+      "loss": 0.3614,
+      "step": 5951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868033948856024e-06,
+      "loss": 0.3277,
+      "step": 5952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865060104400153e-06,
+      "loss": 0.3648,
+      "step": 5953
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.862086361621809e-06,
+      "loss": 0.3907,
+      "step": 5954
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8591127207874e-06,
+      "loss": 0.3754,
+      "step": 5955
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85613918216333e-06,
+      "loss": 0.3618,
+      "step": 5956
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853165746015997e-06,
+      "loss": 0.3451,
+      "step": 5957
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850192412611787e-06,
+      "loss": 0.3237,
+      "step": 5958
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.847219182217071e-06,
+      "loss": 0.38,
+      "step": 5959
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844246055098221e-06,
+      "loss": 0.3762,
+      "step": 5960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841273031521593e-06,
+      "loss": 0.3547,
+      "step": 5961
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838300111753536e-06,
+      "loss": 0.3712,
+      "step": 5962
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835327296060392e-06,
+      "loss": 0.3889,
+      "step": 5963
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832354584708486e-06,
+      "loss": 0.3535,
+      "step": 5964
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.829381977964141e-06,
+      "loss": 0.3909,
+      "step": 5965
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826409476093666e-06,
+      "loss": 0.3171,
+      "step": 5966
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.823437079363367e-06,
+      "loss": 0.3614,
+      "step": 5967
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82046478803954e-06,
+      "loss": 0.3612,
+      "step": 5968
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817492602388458e-06,
+      "loss": 0.3925,
+      "step": 5969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.814520522676396e-06,
+      "loss": 0.3728,
+      "step": 5970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811548549169625e-06,
+      "loss": 0.3653,
+      "step": 5971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.8085766821344e-06,
+      "loss": 0.3494,
+      "step": 5972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805604921836956e-06,
+      "loss": 0.3955,
+      "step": 5973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.802633268543534e-06,
+      "loss": 0.3185,
+      "step": 5974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79966172252036e-06,
+      "loss": 0.3732,
+      "step": 5975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796690284033653e-06,
+      "loss": 0.3755,
+      "step": 5976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793718953349619e-06,
+      "loss": 0.3771,
+      "step": 5977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.790747730734448e-06,
+      "loss": 0.4058,
+      "step": 5978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787776616454335e-06,
+      "loss": 0.3593,
+      "step": 5979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784805610775453e-06,
+      "loss": 0.3484,
+      "step": 5980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.781834713963976e-06,
+      "loss": 0.3669,
+      "step": 5981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778863926286055e-06,
+      "loss": 0.3045,
+      "step": 5982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77589324800784e-06,
+      "loss": 0.3532,
+      "step": 5983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77292267939547e-06,
+      "loss": 0.3713,
+      "step": 5984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769952220715076e-06,
+      "loss": 0.3402,
+      "step": 5985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.766981872232776e-06,
+      "loss": 0.3619,
+      "step": 5986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764011634214677e-06,
+      "loss": 0.3844,
+      "step": 5987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761041506926878e-06,
+      "loss": 0.3512,
+      "step": 5988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758071490635468e-06,
+      "loss": 0.3295,
+      "step": 5989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755101585606533e-06,
+      "loss": 0.3836,
+      "step": 5990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.752131792106132e-06,
+      "loss": 0.3386,
+      "step": 5991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74916211040033e-06,
+      "loss": 0.3599,
+      "step": 5992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.746192540755175e-06,
+      "loss": 0.3556,
+      "step": 5993
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743223083436705e-06,
+      "loss": 0.3323,
+      "step": 5994
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740253738710955e-06,
+      "loss": 0.356,
+      "step": 5995
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.737284506843936e-06,
+      "loss": 0.3674,
+      "step": 5996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73431538810166e-06,
+      "loss": 0.3444,
+      "step": 5997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731346382750128e-06,
+      "loss": 0.3416,
+      "step": 5998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728377491055328e-06,
+      "loss": 0.3607,
+      "step": 5999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725408713283234e-06,
+      "loss": 0.4057,
+      "step": 6000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72244004969982e-06,
+      "loss": 0.367,
+      "step": 6001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719471500571038e-06,
+      "loss": 0.3952,
+      "step": 6002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716503066162841e-06,
+      "loss": 0.4211,
+      "step": 6003
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71353474674117e-06,
+      "loss": 0.3861,
+      "step": 6004
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710566542571943e-06,
+      "loss": 0.3917,
+      "step": 6005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.707598453921078e-06,
+      "loss": 0.3853,
+      "step": 6006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.704630481054485e-06,
+      "loss": 0.34,
+      "step": 6007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.701662624238062e-06,
+      "loss": 0.3818,
+      "step": 6008
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698694883737696e-06,
+      "loss": 0.3792,
+      "step": 6009
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695727259819251e-06,
+      "loss": 0.3607,
+      "step": 6010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692759752748602e-06,
+      "loss": 0.4117,
+      "step": 6011
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.689792362791602e-06,
+      "loss": 0.3656,
+      "step": 6012
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686825090214094e-06,
+      "loss": 0.3826,
+      "step": 6013
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68385793528191e-06,
+      "loss": 0.3753,
+      "step": 6014
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680890898260876e-06,
+      "loss": 0.3668,
+      "step": 6015
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677923979416799e-06,
+      "loss": 0.3847,
+      "step": 6016
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674957179015487e-06,
+      "loss": 0.406,
+      "step": 6017
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671990497322727e-06,
+      "loss": 0.4058,
+      "step": 6018
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6690239346043e-06,
+      "loss": 0.356,
+      "step": 6019
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.666057491125977e-06,
+      "loss": 0.411,
+      "step": 6020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663091167153516e-06,
+      "loss": 0.3677,
+      "step": 6021
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660124962952668e-06,
+      "loss": 0.3583,
+      "step": 6022
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657158878789165e-06,
+      "loss": 0.3454,
+      "step": 6023
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.654192914928739e-06,
+      "loss": 0.3385,
+      "step": 6024
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651227071637105e-06,
+      "loss": 0.3791,
+      "step": 6025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648261349179968e-06,
+      "loss": 0.4092,
+      "step": 6026
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645295747823024e-06,
+      "loss": 0.3568,
+      "step": 6027
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642330267831952e-06,
+      "loss": 0.3497,
+      "step": 6028
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63936490947243e-06,
+      "loss": 0.3611,
+      "step": 6029
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636399673010118e-06,
+      "loss": 0.3697,
+      "step": 6030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633434558710668e-06,
+      "loss": 0.3495,
+      "step": 6031
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630469566839717e-06,
+      "loss": 0.3614,
+      "step": 6032
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627504697662896e-06,
+      "loss": 0.3659,
+      "step": 6033
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.624539951445823e-06,
+      "loss": 0.3715,
+      "step": 6034
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621575328454102e-06,
+      "loss": 0.3981,
+      "step": 6035
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61861082895334e-06,
+      "loss": 0.3801,
+      "step": 6036
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61564645320911e-06,
+      "loss": 0.3932,
+      "step": 6037
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.612682201486987e-06,
+      "loss": 0.3356,
+      "step": 6038
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60971807405254e-06,
+      "loss": 0.4147,
+      "step": 6039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.606754071171321e-06,
+      "loss": 0.3868,
+      "step": 6040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.603790193108863e-06,
+      "loss": 0.3231,
+      "step": 6041
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.600826440130696e-06,
+      "loss": 0.3593,
+      "step": 6042
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.597862812502345e-06,
+      "loss": 0.362,
+      "step": 6043
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59489931048931e-06,
+      "loss": 0.3806,
+      "step": 6044
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591935934357098e-06,
+      "loss": 0.4034,
+      "step": 6045
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588972684371176e-06,
+      "loss": 0.3323,
+      "step": 6046
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586009560797028e-06,
+      "loss": 0.3331,
+      "step": 6047
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583046563900114e-06,
+      "loss": 0.3658,
+      "step": 6048
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580083693945887e-06,
+      "loss": 0.3647,
+      "step": 6049
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577120951199783e-06,
+      "loss": 0.3827,
+      "step": 6050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574158335927226e-06,
+      "loss": 0.3292,
+      "step": 6051
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571195848393638e-06,
+      "loss": 0.354,
+      "step": 6052
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56823348886442e-06,
+      "loss": 0.3597,
+      "step": 6053
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56527125760497e-06,
+      "loss": 0.35,
+      "step": 6054
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562309154880663e-06,
+      "loss": 0.3346,
+      "step": 6055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559347180956874e-06,
+      "loss": 0.3788,
+      "step": 6056
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55638533609896e-06,
+      "loss": 0.4074,
+      "step": 6057
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553423620572268e-06,
+      "loss": 0.3702,
+      "step": 6058
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550462034642138e-06,
+      "loss": 0.3554,
+      "step": 6059
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547500578573887e-06,
+      "loss": 0.3845,
+      "step": 6060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544539252632829e-06,
+      "loss": 0.3573,
+      "step": 6061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541578057084267e-06,
+      "loss": 0.4025,
+      "step": 6062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538616992193492e-06,
+      "loss": 0.3757,
+      "step": 6063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535656058225775e-06,
+      "loss": 0.4041,
+      "step": 6064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532695255446384e-06,
+      "loss": 0.3417,
+      "step": 6065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.529734584120575e-06,
+      "loss": 0.3605,
+      "step": 6066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526774044513585e-06,
+      "loss": 0.3483,
+      "step": 6067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523813636890656e-06,
+      "loss": 0.3387,
+      "step": 6068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520853361516992e-06,
+      "loss": 0.3882,
+      "step": 6069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517893218657807e-06,
+      "loss": 0.3756,
+      "step": 6070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514933208578293e-06,
+      "loss": 0.3963,
+      "step": 6071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511973331543638e-06,
+      "loss": 0.3944,
+      "step": 6072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509013587819007e-06,
+      "loss": 0.411,
+      "step": 6073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50605397766956e-06,
+      "loss": 0.3723,
+      "step": 6074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50309450136044e-06,
+      "loss": 0.3702,
+      "step": 6075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.500135159156792e-06,
+      "loss": 0.3612,
+      "step": 6076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497175951323737e-06,
+      "loss": 0.3415,
+      "step": 6077
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.494216878126376e-06,
+      "loss": 0.3466,
+      "step": 6078
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.491257939829815e-06,
+      "loss": 0.3604,
+      "step": 6079
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488299136699143e-06,
+      "loss": 0.3928,
+      "step": 6080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.485340468999435e-06,
+      "loss": 0.3779,
+      "step": 6081
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482381936995742e-06,
+      "loss": 0.3849,
+      "step": 6082
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.479423540953126e-06,
+      "loss": 0.3931,
+      "step": 6083
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476465281136622e-06,
+      "loss": 0.3681,
+      "step": 6084
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.473507157811254e-06,
+      "loss": 0.3849,
+      "step": 6085
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.470549171242042e-06,
+      "loss": 0.3454,
+      "step": 6086
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46759132169398e-06,
+      "loss": 0.3583,
+      "step": 6087
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46463360943206e-06,
+      "loss": 0.3657,
+      "step": 6088
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461676034721258e-06,
+      "loss": 0.4321,
+      "step": 6089
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45871859782654e-06,
+      "loss": 0.3502,
+      "step": 6090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455761299012859e-06,
+      "loss": 0.3637,
+      "step": 6091
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452804138545153e-06,
+      "loss": 0.3308,
+      "step": 6092
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44984711668835e-06,
+      "loss": 0.3252,
+      "step": 6093
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446890233707364e-06,
+      "loss": 0.3775,
+      "step": 6094
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.443933489867103e-06,
+      "loss": 0.3675,
+      "step": 6095
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.440976885432448e-06,
+      "loss": 0.3835,
+      "step": 6096
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438020420668282e-06,
+      "loss": 0.3632,
+      "step": 6097
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.435064095839472e-06,
+      "loss": 0.3968,
+      "step": 6098
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432107911210868e-06,
+      "loss": 0.3886,
+      "step": 6099
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42915186704731e-06,
+      "loss": 0.348,
+      "step": 6100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426195963613626e-06,
+      "loss": 0.349,
+      "step": 6101
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42324020117463e-06,
+      "loss": 0.4225,
+      "step": 6102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420284579995122e-06,
+      "loss": 0.3512,
+      "step": 6103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417329100339902e-06,
+      "loss": 0.3777,
+      "step": 6104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414373762473733e-06,
+      "loss": 0.3766,
+      "step": 6105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411418566661387e-06,
+      "loss": 0.3944,
+      "step": 6106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408463513167613e-06,
+      "loss": 0.4288,
+      "step": 6107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.40550860225715e-06,
+      "loss": 0.3832,
+      "step": 6108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402553834194733e-06,
+      "loss": 0.4151,
+      "step": 6109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39959920924506e-06,
+      "loss": 0.3912,
+      "step": 6110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396644727672836e-06,
+      "loss": 0.3868,
+      "step": 6111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393690389742753e-06,
+      "loss": 0.3643,
+      "step": 6112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390736195719486e-06,
+      "loss": 0.3785,
+      "step": 6113
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387782145867688e-06,
+      "loss": 0.3486,
+      "step": 6114
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384828240452018e-06,
+      "loss": 0.3547,
+      "step": 6115
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.381874479737105e-06,
+      "loss": 0.387,
+      "step": 6116
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378920863987576e-06,
+      "loss": 0.3657,
+      "step": 6117
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.375967393468041e-06,
+      "loss": 0.374,
+      "step": 6118
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373014068443092e-06,
+      "loss": 0.3343,
+      "step": 6119
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370060889177317e-06,
+      "loss": 0.3502,
+      "step": 6120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367107855935286e-06,
+      "loss": 0.3659,
+      "step": 6121
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36415496898156e-06,
+      "loss": 0.3727,
+      "step": 6122
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361202228580675e-06,
+      "loss": 0.3864,
+      "step": 6123
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.358249634997168e-06,
+      "loss": 0.3843,
+      "step": 6124
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355297188495559e-06,
+      "loss": 0.3579,
+      "step": 6125
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35234488934035e-06,
+      "loss": 0.3904,
+      "step": 6126
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349392737796038e-06,
+      "loss": 0.3876,
+      "step": 6127
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346440734127094e-06,
+      "loss": 0.3816,
+      "step": 6128
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343488878597987e-06,
+      "loss": 0.3685,
+      "step": 6129
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340537171473169e-06,
+      "loss": 0.3511,
+      "step": 6130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.337585613017082e-06,
+      "loss": 0.3882,
+      "step": 6131
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33463420349415e-06,
+      "loss": 0.3334,
+      "step": 6132
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33168294316878e-06,
+      "loss": 0.3566,
+      "step": 6133
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328731832305377e-06,
+      "loss": 0.3589,
+      "step": 6134
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325780871168325e-06,
+      "loss": 0.3337,
+      "step": 6135
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.322830060021998e-06,
+      "loss": 0.3905,
+      "step": 6136
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31987939913075e-06,
+      "loss": 0.3692,
+      "step": 6137
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316928888758927e-06,
+      "loss": 0.3832,
+      "step": 6138
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.313978529170863e-06,
+      "loss": 0.3831,
+      "step": 6139
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311028320630873e-06,
+      "loss": 0.351,
+      "step": 6140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308078263403271e-06,
+      "loss": 0.3657,
+      "step": 6141
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305128357752339e-06,
+      "loss": 0.3733,
+      "step": 6142
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302178603942353e-06,
+      "loss": 0.3168,
+      "step": 6143
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299229002237581e-06,
+      "loss": 0.3834,
+      "step": 6144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296279552902279e-06,
+      "loss": 0.3426,
+      "step": 6145
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293330256200673e-06,
+      "loss": 0.3737,
+      "step": 6146
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290381112396989e-06,
+      "loss": 0.3317,
+      "step": 6147
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287432121755439e-06,
+      "loss": 0.4245,
+      "step": 6148
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.284483284540217e-06,
+      "loss": 0.3573,
+      "step": 6149
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.281534601015507e-06,
+      "loss": 0.3769,
+      "step": 6150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278586071445475e-06,
+      "loss": 0.3413,
+      "step": 6151
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.275637696094273e-06,
+      "loss": 0.3585,
+      "step": 6152
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272689475226046e-06,
+      "loss": 0.3599,
+      "step": 6153
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269741409104918e-06,
+      "loss": 0.4162,
+      "step": 6154
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266793497995003e-06,
+      "loss": 0.3492,
+      "step": 6155
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263845742160397e-06,
+      "loss": 0.3502,
+      "step": 6156
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260898141865188e-06,
+      "loss": 0.376,
+      "step": 6157
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.257950697373445e-06,
+      "loss": 0.3789,
+      "step": 6158
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25500340894923e-06,
+      "loss": 0.3428,
+      "step": 6159
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252056276856578e-06,
+      "loss": 0.3582,
+      "step": 6160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249109301359522e-06,
+      "loss": 0.357,
+      "step": 6161
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246162482722076e-06,
+      "loss": 0.4238,
+      "step": 6162
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243215821208246e-06,
+      "loss": 0.4039,
+      "step": 6163
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240269317082012e-06,
+      "loss": 0.3439,
+      "step": 6164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237322970607348e-06,
+      "loss": 0.3554,
+      "step": 6165
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.234376782048217e-06,
+      "loss": 0.4003,
+      "step": 6166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23143075166856e-06,
+      "loss": 0.3575,
+      "step": 6167
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22848487973231e-06,
+      "loss": 0.3695,
+      "step": 6168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225539166503378e-06,
+      "loss": 0.3498,
+      "step": 6169
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222593612245673e-06,
+      "loss": 0.3836,
+      "step": 6170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219648217223075e-06,
+      "loss": 0.3491,
+      "step": 6171
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21670298169946e-06,
+      "loss": 0.3473,
+      "step": 6172
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213757905938696e-06,
+      "loss": 0.3326,
+      "step": 6173
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210812990204614e-06,
+      "loss": 0.3816,
+      "step": 6174
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207868234761052e-06,
+      "loss": 0.3582,
+      "step": 6175
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204923639871824e-06,
+      "loss": 0.3518,
+      "step": 6176
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.201979205800739e-06,
+      "loss": 0.4198,
+      "step": 6177
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.199034932811571e-06,
+      "loss": 0.3863,
+      "step": 6178
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.1960908211681e-06,
+      "loss": 0.3909,
+      "step": 6179
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193146871134086e-06,
+      "loss": 0.3859,
+      "step": 6180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190203082973272e-06,
+      "loss": 0.3477,
+      "step": 6181
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187259456949389e-06,
+      "loss": 0.3541,
+      "step": 6182
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.184315993326144e-06,
+      "loss": 0.378,
+      "step": 6183
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181372692367246e-06,
+      "loss": 0.4156,
+      "step": 6184
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178429554336376e-06,
+      "loss": 0.3857,
+      "step": 6185
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.175486579497211e-06,
+      "loss": 0.3295,
+      "step": 6186
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.172543768113401e-06,
+      "loss": 0.368,
+      "step": 6187
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169601120448592e-06,
+      "loss": 0.3393,
+      "step": 6188
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.166658636766408e-06,
+      "loss": 0.3977,
+      "step": 6189
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.163716317330465e-06,
+      "loss": 0.3974,
+      "step": 6190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16077416240436e-06,
+      "loss": 0.3479,
+      "step": 6191
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.157832172251674e-06,
+      "loss": 0.3825,
+      "step": 6192
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154890347135977e-06,
+      "loss": 0.3365,
+      "step": 6193
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.151948687320823e-06,
+      "loss": 0.362,
+      "step": 6194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149007193069753e-06,
+      "loss": 0.37,
+      "step": 6195
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146065864646285e-06,
+      "loss": 0.332,
+      "step": 6196
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143124702313932e-06,
+      "loss": 0.3702,
+      "step": 6197
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140183706336186e-06,
+      "loss": 0.3699,
+      "step": 6198
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13724287697653e-06,
+      "loss": 0.3594,
+      "step": 6199
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134302214498427e-06,
+      "loss": 0.377,
+      "step": 6200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.131361719165324e-06,
+      "loss": 0.3555,
+      "step": 6201
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128421391240657e-06,
+      "loss": 0.3821,
+      "step": 6202
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125481230987847e-06,
+      "loss": 0.3367,
+      "step": 6203
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122541238670299e-06,
+      "loss": 0.3319,
+      "step": 6204
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119601414551396e-06,
+      "loss": 0.3457,
+      "step": 6205
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11666175889452e-06,
+      "loss": 0.3525,
+      "step": 6206
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113722271963026e-06,
+      "loss": 0.3731,
+      "step": 6207
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.110782954020257e-06,
+      "loss": 0.4132,
+      "step": 6208
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.107843805329551e-06,
+      "loss": 0.3539,
+      "step": 6209
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10490482615421e-06,
+      "loss": 0.3447,
+      "step": 6210
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101966016757536e-06,
+      "loss": 0.3817,
+      "step": 6211
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099027377402815e-06,
+      "loss": 0.3593,
+      "step": 6212
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096088908353316e-06,
+      "loss": 0.3574,
+      "step": 6213
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093150609872293e-06,
+      "loss": 0.3926,
+      "step": 6214
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090212482222976e-06,
+      "loss": 0.3703,
+      "step": 6215
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087274525668593e-06,
+      "loss": 0.3416,
+      "step": 6216
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084336740472352e-06,
+      "loss": 0.3456,
+      "step": 6217
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.081399126897444e-06,
+      "loss": 0.3724,
+      "step": 6218
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07846168520704e-06,
+      "loss": 0.3424,
+      "step": 6219
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075524415664307e-06,
+      "loss": 0.3792,
+      "step": 6220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07258731853239e-06,
+      "loss": 0.3269,
+      "step": 6221
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.069650394074415e-06,
+      "loss": 0.3505,
+      "step": 6222
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.066713642553502e-06,
+      "loss": 0.3344,
+      "step": 6223
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063777064232746e-06,
+      "loss": 0.3712,
+      "step": 6224
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060840659375232e-06,
+      "loss": 0.3443,
+      "step": 6225
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057904428244026e-06,
+      "loss": 0.3787,
+      "step": 6226
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.054968371102186e-06,
+      "loss": 0.3991,
+      "step": 6227
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.052032488212743e-06,
+      "loss": 0.3911,
+      "step": 6228
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.396,
+      "step": 6229
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046161246243121e-06,
+      "loss": 0.3995,
+      "step": 6230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04322588768894e-06,
+      "loss": 0.3806,
+      "step": 6231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04029070443915e-06,
+      "loss": 0.3709,
+      "step": 6232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037355696756708e-06,
+      "loss": 0.3742,
+      "step": 6233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034420864904555e-06,
+      "loss": 0.4159,
+      "step": 6234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031486209145622e-06,
+      "loss": 0.3802,
+      "step": 6235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028551729742822e-06,
+      "loss": 0.371,
+      "step": 6236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.025617426959046e-06,
+      "loss": 0.4082,
+      "step": 6237
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022683301057172e-06,
+      "loss": 0.3999,
+      "step": 6238
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019749352300067e-06,
+      "loss": 0.3581,
+      "step": 6239
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01681558095058e-06,
+      "loss": 0.3458,
+      "step": 6240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013881987271547e-06,
+      "loss": 0.3758,
+      "step": 6241
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.010948571525774e-06,
+      "loss": 0.3912,
+      "step": 6242
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008015333976066e-06,
+      "loss": 0.3429,
+      "step": 6243
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005082274885207e-06,
+      "loss": 0.3772,
+      "step": 6244
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002149394515973e-06,
+      "loss": 0.3819,
+      "step": 6245
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.999216693131103e-06,
+      "loss": 0.3811,
+      "step": 6246
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99628417099334e-06,
+      "loss": 0.3959,
+      "step": 6247
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.9933518283654e-06,
+      "loss": 0.3464,
+      "step": 6248
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.990419665509995e-06,
+      "loss": 0.3353,
+      "step": 6249
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987487682689811e-06,
+      "loss": 0.3557,
+      "step": 6250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98455588016751e-06,
+      "loss": 0.3686,
+      "step": 6251
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98162425820576e-06,
+      "loss": 0.3247,
+      "step": 6252
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978692817067193e-06,
+      "loss": 0.374,
+      "step": 6253
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975761557014437e-06,
+      "loss": 0.3396,
+      "step": 6254
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972830478310098e-06,
+      "loss": 0.3635,
+      "step": 6255
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969899581216765e-06,
+      "loss": 0.3818,
+      "step": 6256
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.966968865997013e-06,
+      "loss": 0.325,
+      "step": 6257
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964038332913399e-06,
+      "loss": 0.3926,
+      "step": 6258
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.96110798222847e-06,
+      "loss": 0.3632,
+      "step": 6259
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.958177814204745e-06,
+      "loss": 0.3709,
+      "step": 6260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95524782910474e-06,
+      "loss": 0.3494,
+      "step": 6261
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952318027190941e-06,
+      "loss": 0.3282,
+      "step": 6262
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94938840872583e-06,
+      "loss": 0.4189,
+      "step": 6263
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946458973971864e-06,
+      "loss": 0.3873,
+      "step": 6264
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943529723191488e-06,
+      "loss": 0.4103,
+      "step": 6265
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940600656647127e-06,
+      "loss": 0.3651,
+      "step": 6266
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.937671774601194e-06,
+      "loss": 0.387,
+      "step": 6267
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934743077316085e-06,
+      "loss": 0.379,
+      "step": 6268
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93181456505417e-06,
+      "loss": 0.3772,
+      "step": 6269
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.928886238077817e-06,
+      "loss": 0.3685,
+      "step": 6270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925958096649366e-06,
+      "loss": 0.4267,
+      "step": 6271
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923030141031147e-06,
+      "loss": 0.3662,
+      "step": 6272
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.920102371485474e-06,
+      "loss": 0.3605,
+      "step": 6273
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917174788274636e-06,
+      "loss": 0.3299,
+      "step": 6274
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.914247391660911e-06,
+      "loss": 0.3127,
+      "step": 6275
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91132018190656e-06,
+      "loss": 0.3709,
+      "step": 6276
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.908393159273835e-06,
+      "loss": 0.3998,
+      "step": 6277
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.905466324024954e-06,
+      "loss": 0.3673,
+      "step": 6278
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902539676422132e-06,
+      "loss": 0.3148,
+      "step": 6279
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899613216727557e-06,
+      "loss": 0.3372,
+      "step": 6280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896686945203415e-06,
+      "loss": 0.3909,
+      "step": 6281
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.893760862111866e-06,
+      "loss": 0.3159,
+      "step": 6282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.890834967715046e-06,
+      "loss": 0.3926,
+      "step": 6283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887909262275081e-06,
+      "loss": 0.3392,
+      "step": 6284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.884983746054087e-06,
+      "loss": 0.3803,
+      "step": 6285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882058419314157e-06,
+      "loss": 0.3357,
+      "step": 6286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87913328231736e-06,
+      "loss": 0.3503,
+      "step": 6287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876208335325757e-06,
+      "loss": 0.3869,
+      "step": 6288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873283578601391e-06,
+      "loss": 0.3619,
+      "step": 6289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.870359012406288e-06,
+      "loss": 0.3551,
+      "step": 6290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.867434637002454e-06,
+      "loss": 0.3597,
+      "step": 6291
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.864510452651876e-06,
+      "loss": 0.3702,
+      "step": 6292
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86158645961653e-06,
+      "loss": 0.3275,
+      "step": 6293
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.85866265815837e-06,
+      "loss": 0.3393,
+      "step": 6294
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85573904853934e-06,
+      "loss": 0.3573,
+      "step": 6295
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.852815631021358e-06,
+      "loss": 0.4133,
+      "step": 6296
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849892405866327e-06,
+      "loss": 0.3761,
+      "step": 6297
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846969373336136e-06,
+      "loss": 0.3502,
+      "step": 6298
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844046533692655e-06,
+      "loss": 0.3536,
+      "step": 6299
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84112388719774e-06,
+      "loss": 0.3901,
+      "step": 6300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83820143411322e-06,
+      "loss": 0.368,
+      "step": 6301
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835279174700917e-06,
+      "loss": 0.3878,
+      "step": 6302
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83235710922263e-06,
+      "loss": 0.3533,
+      "step": 6303
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829435237940145e-06,
+      "loss": 0.3902,
+      "step": 6304
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.826513561115227e-06,
+      "loss": 0.4091,
+      "step": 6305
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823592079009623e-06,
+      "loss": 0.3972,
+      "step": 6306
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820670791885066e-06,
+      "loss": 0.3739,
+      "step": 6307
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.817749700003266e-06,
+      "loss": 0.3137,
+      "step": 6308
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814828803625926e-06,
+      "loss": 0.3669,
+      "step": 6309
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811908103014716e-06,
+      "loss": 0.3421,
+      "step": 6310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808987598431303e-06,
+      "loss": 0.3812,
+      "step": 6311
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806067290137327e-06,
+      "loss": 0.3845,
+      "step": 6312
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803147178394419e-06,
+      "loss": 0.3511,
+      "step": 6313
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800227263464187e-06,
+      "loss": 0.3929,
+      "step": 6314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.797307545608216e-06,
+      "loss": 0.3738,
+      "step": 6315
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794388025088079e-06,
+      "loss": 0.3723,
+      "step": 6316
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791468702165337e-06,
+      "loss": 0.4237,
+      "step": 6317
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788549577101529e-06,
+      "loss": 0.3939,
+      "step": 6318
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785630650158166e-06,
+      "loss": 0.373,
+      "step": 6319
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.782711921596754e-06,
+      "loss": 0.3946,
+      "step": 6320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77979339167878e-06,
+      "loss": 0.3614,
+      "step": 6321
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77687506066571e-06,
+      "loss": 0.3531,
+      "step": 6322
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.773956928818993e-06,
+      "loss": 0.4006,
+      "step": 6323
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.771038996400058e-06,
+      "loss": 0.3798,
+      "step": 6324
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.768121263670318e-06,
+      "loss": 0.3801,
+      "step": 6325
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76520373089117e-06,
+      "loss": 0.4128,
+      "step": 6326
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762286398323995e-06,
+      "loss": 0.3778,
+      "step": 6327
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.75936926623014e-06,
+      "loss": 0.3915,
+      "step": 6328
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756452334870959e-06,
+      "loss": 0.3789,
+      "step": 6329
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753535604507769e-06,
+      "loss": 0.3583,
+      "step": 6330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.750619075401877e-06,
+      "loss": 0.3784,
+      "step": 6331
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747702747814573e-06,
+      "loss": 0.3451,
+      "step": 6332
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744786622007123e-06,
+      "loss": 0.3681,
+      "step": 6333
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.741870698240777e-06,
+      "loss": 0.3341,
+      "step": 6334
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73895497677677e-06,
+      "loss": 0.362,
+      "step": 6335
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73603945787632e-06,
+      "loss": 0.3375,
+      "step": 6336
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733124141800618e-06,
+      "loss": 0.3591,
+      "step": 6337
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730209028810847e-06,
+      "loss": 0.3593,
+      "step": 6338
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.727294119168166e-06,
+      "loss": 0.3338,
+      "step": 6339
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724379413133718e-06,
+      "loss": 0.3352,
+      "step": 6340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.721464910968628e-06,
+      "loss": 0.4095,
+      "step": 6341
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718550612933997e-06,
+      "loss": 0.3705,
+      "step": 6342
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.715636519290917e-06,
+      "loss": 0.3791,
+      "step": 6343
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712722630300455e-06,
+      "loss": 0.3574,
+      "step": 6344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709808946223662e-06,
+      "loss": 0.377,
+      "step": 6345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706895467321577e-06,
+      "loss": 0.3593,
+      "step": 6346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703982193855205e-06,
+      "loss": 0.3789,
+      "step": 6347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70106912608554e-06,
+      "loss": 0.3769,
+      "step": 6348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69815626427357e-06,
+      "loss": 0.3461,
+      "step": 6349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69524360868025e-06,
+      "loss": 0.4084,
+      "step": 6350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692331159566515e-06,
+      "loss": 0.3674,
+      "step": 6351
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68941891719329e-06,
+      "loss": 0.3589,
+      "step": 6352
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.686506881821478e-06,
+      "loss": 0.3627,
+      "step": 6353
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683595053711967e-06,
+      "loss": 0.3973,
+      "step": 6354
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680683433125623e-06,
+      "loss": 0.3613,
+      "step": 6355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677772020323286e-06,
+      "loss": 0.3519,
+      "step": 6356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674860815565792e-06,
+      "loss": 0.345,
+      "step": 6357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67194981911395e-06,
+      "loss": 0.395,
+      "step": 6358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669039031228554e-06,
+      "loss": 0.3933,
+      "step": 6359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666128452170371e-06,
+      "loss": 0.3887,
+      "step": 6360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663218082200159e-06,
+      "loss": 0.3707,
+      "step": 6361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660307921578654e-06,
+      "loss": 0.3396,
+      "step": 6362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65739797056657e-06,
+      "loss": 0.3522,
+      "step": 6363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654488229424613e-06,
+      "loss": 0.3659,
+      "step": 6364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651578698413452e-06,
+      "loss": 0.405,
+      "step": 6365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648669377793752e-06,
+      "loss": 0.3845,
+      "step": 6366
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.645760267826156e-06,
+      "loss": 0.3445,
+      "step": 6367
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642851368771284e-06,
+      "loss": 0.3385,
+      "step": 6368
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639942680889741e-06,
+      "loss": 0.3937,
+      "step": 6369
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637034204442111e-06,
+      "loss": 0.3646,
+      "step": 6370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634125939688962e-06,
+      "loss": 0.352,
+      "step": 6371
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631217886890839e-06,
+      "loss": 0.3905,
+      "step": 6372
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628310046308272e-06,
+      "loss": 0.374,
+      "step": 6373
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.625402418201768e-06,
+      "loss": 0.3628,
+      "step": 6374
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622495002831817e-06,
+      "loss": 0.3852,
+      "step": 6375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.619587800458891e-06,
+      "loss": 0.408,
+      "step": 6376
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616680811343437e-06,
+      "loss": 0.3657,
+      "step": 6377
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.613774035745901e-06,
+      "loss": 0.4266,
+      "step": 6378
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.610867473926682e-06,
+      "loss": 0.3687,
+      "step": 6379
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.607961126146181e-06,
+      "loss": 0.3441,
+      "step": 6380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605054992664769e-06,
+      "loss": 0.3277,
+      "step": 6381
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6021490737428115e-06,
+      "loss": 0.3361,
+      "step": 6382
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599243369640636e-06,
+      "loss": 0.3624,
+      "step": 6383
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.596337880618559e-06,
+      "loss": 0.353,
+      "step": 6384
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593432606936887e-06,
+      "loss": 0.3689,
+      "step": 6385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590527548855894e-06,
+      "loss": 0.3517,
+      "step": 6386
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.587622706635844e-06,
+      "loss": 0.3999,
+      "step": 6387
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5847180805369705e-06,
+      "loss": 0.3882,
+      "step": 6388
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.581813670819497e-06,
+      "loss": 0.3241,
+      "step": 6389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578909477743628e-06,
+      "loss": 0.371,
+      "step": 6390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576005501569548e-06,
+      "loss": 0.3927,
+      "step": 6391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.573101742557408e-06,
+      "loss": 0.3621,
+      "step": 6392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570198200967363e-06,
+      "loss": 0.3654,
+      "step": 6393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56729487705953e-06,
+      "loss": 0.3917,
+      "step": 6394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564391771094017e-06,
+      "loss": 0.3839,
+      "step": 6395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561488883330912e-06,
+      "loss": 0.435,
+      "step": 6396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558586214030272e-06,
+      "loss": 0.3798,
+      "step": 6397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555683763452147e-06,
+      "loss": 0.3601,
+      "step": 6398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.552781531856564e-06,
+      "loss": 0.3417,
+      "step": 6399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549879519503529e-06,
+      "loss": 0.3686,
+      "step": 6400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546977726653027e-06,
+      "loss": 0.4097,
+      "step": 6401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5440761535650266e-06,
+      "loss": 0.3724,
+      "step": 6402
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.541174800499474e-06,
+      "loss": 0.4047,
+      "step": 6403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538273667716299e-06,
+      "loss": 0.375,
+      "step": 6404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535372755475411e-06,
+      "loss": 0.3568,
+      "step": 6405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.532472064036693e-06,
+      "loss": 0.3683,
+      "step": 6406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.529571593660017e-06,
+      "loss": 0.4074,
+      "step": 6407
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526671344605231e-06,
+      "loss": 0.3027,
+      "step": 6408
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523771317132166e-06,
+      "loss": 0.39,
+      "step": 6409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.520871511500627e-06,
+      "loss": 0.3905,
+      "step": 6410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.517971927970404e-06,
+      "loss": 0.4173,
+      "step": 6411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5150725668012695e-06,
+      "loss": 0.3397,
+      "step": 6412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.512173428252967e-06,
+      "loss": 0.3491,
+      "step": 6413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.509274512585237e-06,
+      "loss": 0.3344,
+      "step": 6414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506375820057776e-06,
+      "loss": 0.3189,
+      "step": 6415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.503477350930281e-06,
+      "loss": 0.4192,
+      "step": 6416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500579105462415e-06,
+      "loss": 0.3382,
+      "step": 6417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4976810839138345e-06,
+      "loss": 0.3801,
+      "step": 6418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.494783286544171e-06,
+      "loss": 0.3736,
+      "step": 6419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.491885713613023e-06,
+      "loss": 0.3786,
+      "step": 6420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488988365379985e-06,
+      "loss": 0.3658,
+      "step": 6421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.486091242104628e-06,
+      "loss": 0.3727,
+      "step": 6422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483194344046502e-06,
+      "loss": 0.4226,
+      "step": 6423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.480297671465127e-06,
+      "loss": 0.3469,
+      "step": 6424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477401224620019e-06,
+      "loss": 0.3618,
+      "step": 6425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.474505003770664e-06,
+      "loss": 0.3484,
+      "step": 6426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471609009176532e-06,
+      "loss": 0.3175,
+      "step": 6427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.468713241097071e-06,
+      "loss": 0.3406,
+      "step": 6428
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4658176997917044e-06,
+      "loss": 0.3526,
+      "step": 6429
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46292238551984e-06,
+      "loss": 0.3919,
+      "step": 6430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460027298540867e-06,
+      "loss": 0.393,
+      "step": 6431
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457132439114154e-06,
+      "loss": 0.3561,
+      "step": 6432
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.454237807499041e-06,
+      "loss": 0.3269,
+      "step": 6433
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451343403954856e-06,
+      "loss": 0.3747,
+      "step": 6434
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.448449228740904e-06,
+      "loss": 0.3748,
+      "step": 6435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445555282116471e-06,
+      "loss": 0.3859,
+      "step": 6436
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442661564340823e-06,
+      "loss": 0.3571,
+      "step": 6437
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4397680756732e-06,
+      "loss": 0.3564,
+      "step": 6438
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.436874816372826e-06,
+      "loss": 0.3577,
+      "step": 6439
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433981786698903e-06,
+      "loss": 0.3807,
+      "step": 6440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431088986910616e-06,
+      "loss": 0.3725,
+      "step": 6441
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428196417267124e-06,
+      "loss": 0.338,
+      "step": 6442
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.425304078027569e-06,
+      "loss": 0.4111,
+      "step": 6443
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422411969451069e-06,
+      "loss": 0.3437,
+      "step": 6444
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419520091796727e-06,
+      "loss": 0.4246,
+      "step": 6445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416628445323622e-06,
+      "loss": 0.3742,
+      "step": 6446
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.413737030290808e-06,
+      "loss": 0.3677,
+      "step": 6447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410845846957327e-06,
+      "loss": 0.3937,
+      "step": 6448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.407954895582189e-06,
+      "loss": 0.3377,
+      "step": 6449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.405064176424402e-06,
+      "loss": 0.3054,
+      "step": 6450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402173689742929e-06,
+      "loss": 0.3839,
+      "step": 6451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399283435796729e-06,
+      "loss": 0.3732,
+      "step": 6452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3963934148447325e-06,
+      "loss": 0.3471,
+      "step": 6453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3935036271458575e-06,
+      "loss": 0.3805,
+      "step": 6454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390614072958997e-06,
+      "loss": 0.3696,
+      "step": 6455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.387724752543014e-06,
+      "loss": 0.3911,
+      "step": 6456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.38483566615676e-06,
+      "loss": 0.3805,
+      "step": 6457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.381946814059066e-06,
+      "loss": 0.3645,
+      "step": 6458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379058196508741e-06,
+      "loss": 0.4132,
+      "step": 6459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376169813764573e-06,
+      "loss": 0.4081,
+      "step": 6460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373281666085324e-06,
+      "loss": 0.366,
+      "step": 6461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370393753729738e-06,
+      "loss": 0.3622,
+      "step": 6462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367506076956541e-06,
+      "loss": 0.3612,
+      "step": 6463
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364618636024436e-06,
+      "loss": 0.4091,
+      "step": 6464
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361731431192104e-06,
+      "loss": 0.3538,
+      "step": 6465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358844462718203e-06,
+      "loss": 0.3133,
+      "step": 6466
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355957730861374e-06,
+      "loss": 0.352,
+      "step": 6467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353071235880236e-06,
+      "loss": 0.3741,
+      "step": 6468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350184978033386e-06,
+      "loss": 0.3631,
+      "step": 6469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347298957579394e-06,
+      "loss": 0.3266,
+      "step": 6470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.344413174776819e-06,
+      "loss": 0.3574,
+      "step": 6471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.341527629884193e-06,
+      "loss": 0.3755,
+      "step": 6472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338642323160029e-06,
+      "loss": 0.3734,
+      "step": 6473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335757254862814e-06,
+      "loss": 0.3614,
+      "step": 6474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332872425251017e-06,
+      "loss": 0.3911,
+      "step": 6475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.329987834583088e-06,
+      "loss": 0.3653,
+      "step": 6476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327103483117453e-06,
+      "loss": 0.3885,
+      "step": 6477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.324219371112515e-06,
+      "loss": 0.37,
+      "step": 6478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321335498826657e-06,
+      "loss": 0.3691,
+      "step": 6479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.318451866518241e-06,
+      "loss": 0.3886,
+      "step": 6480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315568474445607e-06,
+      "loss": 0.3647,
+      "step": 6481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3126853228670765e-06,
+      "loss": 0.3633,
+      "step": 6482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309802412040944e-06,
+      "loss": 0.3605,
+      "step": 6483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3069197422254845e-06,
+      "loss": 0.3658,
+      "step": 6484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304037313678949e-06,
+      "loss": 0.3415,
+      "step": 6485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301155126659578e-06,
+      "loss": 0.376,
+      "step": 6486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.298273181425582e-06,
+      "loss": 0.365,
+      "step": 6487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29539147823514e-06,
+      "loss": 0.3997,
+      "step": 6488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.292510017346423e-06,
+      "loss": 0.3688,
+      "step": 6489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.289628799017583e-06,
+      "loss": 0.3921,
+      "step": 6490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286747823506744e-06,
+      "loss": 0.3476,
+      "step": 6491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.283867091071999e-06,
+      "loss": 0.3536,
+      "step": 6492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280986601971432e-06,
+      "loss": 0.3525,
+      "step": 6493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278106356463104e-06,
+      "loss": 0.3274,
+      "step": 6494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.275226354805051e-06,
+      "loss": 0.3696,
+      "step": 6495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.272346597255292e-06,
+      "loss": 0.355,
+      "step": 6496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.269467084071813e-06,
+      "loss": 0.3686,
+      "step": 6497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266587815512587e-06,
+      "loss": 0.3204,
+      "step": 6498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263708791835564e-06,
+      "loss": 0.3337,
+      "step": 6499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2608300132986735e-06,
+      "loss": 0.3841,
+      "step": 6500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.25795148015982e-06,
+      "loss": 0.37,
+      "step": 6501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.255073192676882e-06,
+      "loss": 0.356,
+      "step": 6502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.252195151107727e-06,
+      "loss": 0.3701,
+      "step": 6503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249317355710191e-06,
+      "loss": 0.3929,
+      "step": 6504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.246439806742095e-06,
+      "loss": 0.3502,
+      "step": 6505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243562504461229e-06,
+      "loss": 0.392,
+      "step": 6506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2406854491253685e-06,
+      "loss": 0.3755,
+      "step": 6507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.237808640992266e-06,
+      "loss": 0.3665,
+      "step": 6508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234932080319645e-06,
+      "loss": 0.3832,
+      "step": 6509
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232055767365223e-06,
+      "loss": 0.3311,
+      "step": 6510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229179702386672e-06,
+      "loss": 0.3167,
+      "step": 6511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2263038856416615e-06,
+      "loss": 0.3717,
+      "step": 6512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.223428317387829e-06,
+      "loss": 0.3982,
+      "step": 6513
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220552997882795e-06,
+      "loss": 0.3583,
+      "step": 6514
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.21767792738415e-06,
+      "loss": 0.3245,
+      "step": 6515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214803106149471e-06,
+      "loss": 0.3255,
+      "step": 6516
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.211928534436307e-06,
+      "loss": 0.4289,
+      "step": 6517
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209054212502187e-06,
+      "loss": 0.3589,
+      "step": 6518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.206180140604624e-06,
+      "loss": 0.3622,
+      "step": 6519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.203306319001089e-06,
+      "loss": 0.3903,
+      "step": 6520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.200432747949049e-06,
+      "loss": 0.3915,
+      "step": 6521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197559427705945e-06,
+      "loss": 0.3358,
+      "step": 6522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.194686358529195e-06,
+      "loss": 0.3806,
+      "step": 6523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1918135406761854e-06,
+      "loss": 0.3958,
+      "step": 6524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18894097440429e-06,
+      "loss": 0.3955,
+      "step": 6525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186068659970862e-06,
+      "loss": 0.3741,
+      "step": 6526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183196597633224e-06,
+      "loss": 0.3729,
+      "step": 6527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180324787648685e-06,
+      "loss": 0.3259,
+      "step": 6528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1774532302745135e-06,
+      "loss": 0.3614,
+      "step": 6529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1745819257679805e-06,
+      "loss": 0.429,
+      "step": 6530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171710874386317e-06,
+      "loss": 0.3287,
+      "step": 6531
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168840076386741e-06,
+      "loss": 0.3519,
+      "step": 6532
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.16596953202643e-06,
+      "loss": 0.3694,
+      "step": 6533
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163099241562563e-06,
+      "loss": 0.4319,
+      "step": 6534
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160229205252282e-06,
+      "loss": 0.3301,
+      "step": 6535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157359423352708e-06,
+      "loss": 0.3283,
+      "step": 6536
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.154489896120944e-06,
+      "loss": 0.3865,
+      "step": 6537
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1516206238140616e-06,
+      "loss": 0.4054,
+      "step": 6538
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.148751606689118e-06,
+      "loss": 0.3862,
+      "step": 6539
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1458828450031425e-06,
+      "loss": 0.339,
+      "step": 6540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143014339013142e-06,
+      "loss": 0.405,
+      "step": 6541
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1401460889761055e-06,
+      "loss": 0.3529,
+      "step": 6542
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1372780951489915e-06,
+      "loss": 0.3767,
+      "step": 6543
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.134410357788739e-06,
+      "loss": 0.3505,
+      "step": 6544
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131542877152267e-06,
+      "loss": 0.3317,
+      "step": 6545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128675653496468e-06,
+      "loss": 0.3536,
+      "step": 6546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1258086870782086e-06,
+      "loss": 0.3513,
+      "step": 6547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.122941978154338e-06,
+      "loss": 0.3191,
+      "step": 6548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120075526981683e-06,
+      "loss": 0.3338,
+      "step": 6549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117209333817039e-06,
+      "loss": 0.3593,
+      "step": 6550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.114343398917194e-06,
+      "loss": 0.3486,
+      "step": 6551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111477722538892e-06,
+      "loss": 0.3289,
+      "step": 6552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108612304938869e-06,
+      "loss": 0.3643,
+      "step": 6553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.105747146373829e-06,
+      "loss": 0.3212,
+      "step": 6554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102882247100469e-06,
+      "loss": 0.3442,
+      "step": 6555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100017607375438e-06,
+      "loss": 0.3658,
+      "step": 6556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 0.4004,
+      "step": 6557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.094289107596905e-06,
+      "loss": 0.3403,
+      "step": 6558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091425248056616e-06,
+      "loss": 0.3385,
+      "step": 6559
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.088561649091079e-06,
+      "loss": 0.3742,
+      "step": 6560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.08569831095683e-06,
+      "loss": 0.3952,
+      "step": 6561
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.082835233910399e-06,
+      "loss": 0.3613,
+      "step": 6562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079972418208283e-06,
+      "loss": 0.3292,
+      "step": 6563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077109864106964e-06,
+      "loss": 0.3831,
+      "step": 6564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074247571862877e-06,
+      "loss": 0.3712,
+      "step": 6565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071385541732465e-06,
+      "loss": 0.3813,
+      "step": 6566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0685237739721295e-06,
+      "loss": 0.3614,
+      "step": 6567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065662268838249e-06,
+      "loss": 0.3658,
+      "step": 6568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.062801026587185e-06,
+      "loss": 0.3681,
+      "step": 6569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0599400474752686e-06,
+      "loss": 0.368,
+      "step": 6570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057079331758811e-06,
+      "loss": 0.3223,
+      "step": 6571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0542188796941026e-06,
+      "loss": 0.3845,
+      "step": 6572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.051358691537405e-06,
+      "loss": 0.3759,
+      "step": 6573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048498767544955e-06,
+      "loss": 0.3654,
+      "step": 6574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.045639107972974e-06,
+      "loss": 0.3603,
+      "step": 6575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.04277971307765e-06,
+      "loss": 0.3655,
+      "step": 6576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039920583115154e-06,
+      "loss": 0.3914,
+      "step": 6577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.037061718341632e-06,
+      "loss": 0.3568,
+      "step": 6578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034203119013203e-06,
+      "loss": 0.3579,
+      "step": 6579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031344785385965e-06,
+      "loss": 0.3639,
+      "step": 6580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.028486717715994e-06,
+      "loss": 0.3636,
+      "step": 6581
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.025628916259335e-06,
+      "loss": 0.3208,
+      "step": 6582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022771381272025e-06,
+      "loss": 0.3578,
+      "step": 6583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.019914113010051e-06,
+      "loss": 0.3754,
+      "step": 6584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.017057111729401e-06,
+      "loss": 0.3652,
+      "step": 6585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014200377686024e-06,
+      "loss": 0.3603,
+      "step": 6586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011343911135861e-06,
+      "loss": 0.3773,
+      "step": 6587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008487712334809e-06,
+      "loss": 0.3402,
+      "step": 6588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.005631781538749e-06,
+      "loss": 0.3913,
+      "step": 6589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002776119003542e-06,
+      "loss": 0.3215,
+      "step": 6590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9999207249850255e-06,
+      "loss": 0.3551,
+      "step": 6591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9970655997390106e-06,
+      "loss": 0.3694,
+      "step": 6592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994210743521278e-06,
+      "loss": 0.3924,
+      "step": 6593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.99135615658759e-06,
+      "loss": 0.3614,
+      "step": 6594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988501839193689e-06,
+      "loss": 0.4153,
+      "step": 6595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.985647791595292e-06,
+      "loss": 0.3464,
+      "step": 6596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9827940140480776e-06,
+      "loss": 0.3665,
+      "step": 6597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97994050680772e-06,
+      "loss": 0.4033,
+      "step": 6598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.977087270129859e-06,
+      "loss": 0.4195,
+      "step": 6599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974234304270109e-06,
+      "loss": 0.368,
+      "step": 6600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.971381609484067e-06,
+      "loss": 0.304,
+      "step": 6601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.968529186027297e-06,
+      "loss": 0.3196,
+      "step": 6602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965677034155347e-06,
+      "loss": 0.3787,
+      "step": 6603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9628251541237355e-06,
+      "loss": 0.3895,
+      "step": 6604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.959973546187959e-06,
+      "loss": 0.3269,
+      "step": 6605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957122210603486e-06,
+      "loss": 0.3685,
+      "step": 6606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.954271147625765e-06,
+      "loss": 0.3522,
+      "step": 6607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951420357510219e-06,
+      "loss": 0.3506,
+      "step": 6608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9485698405122446e-06,
+      "loss": 0.3233,
+      "step": 6609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.945719596887217e-06,
+      "loss": 0.3294,
+      "step": 6610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942869626890483e-06,
+      "loss": 0.3486,
+      "step": 6611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9400199307773684e-06,
+      "loss": 0.4014,
+      "step": 6612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937170508803172e-06,
+      "loss": 0.3612,
+      "step": 6613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.934321361223173e-06,
+      "loss": 0.357,
+      "step": 6614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.931472488292616e-06,
+      "loss": 0.397,
+      "step": 6615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.92862389026673e-06,
+      "loss": 0.3727,
+      "step": 6616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.925775567400717e-06,
+      "loss": 0.3837,
+      "step": 6617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.922927519949753e-06,
+      "loss": 0.3847,
+      "step": 6618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.920079748168996e-06,
+      "loss": 0.3154,
+      "step": 6619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.917232252313562e-06,
+      "loss": 0.3509,
+      "step": 6620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914385032638562e-06,
+      "loss": 0.4128,
+      "step": 6621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9115380893990705e-06,
+      "loss": 0.3219,
+      "step": 6622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.908691422850144e-06,
+      "loss": 0.3864,
+      "step": 6623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.905845033246814e-06,
+      "loss": 0.3679,
+      "step": 6624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902998920844075e-06,
+      "loss": 0.3329,
+      "step": 6625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900153085896909e-06,
+      "loss": 0.365,
+      "step": 6626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897307528660273e-06,
+      "loss": 0.4229,
+      "step": 6627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894462249389098e-06,
+      "loss": 0.3783,
+      "step": 6628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891617248338282e-06,
+      "loss": 0.3407,
+      "step": 6629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888772525762703e-06,
+      "loss": 0.3851,
+      "step": 6630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.885928081917223e-06,
+      "loss": 0.3748,
+      "step": 6631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8830839170566676e-06,
+      "loss": 0.3419,
+      "step": 6632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880240031435843e-06,
+      "loss": 0.3603,
+      "step": 6633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877396425309526e-06,
+      "loss": 0.3714,
+      "step": 6634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874553098932471e-06,
+      "loss": 0.3489,
+      "step": 6635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.871710052559406e-06,
+      "loss": 0.3799,
+      "step": 6636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868867286445041e-06,
+      "loss": 0.3845,
+      "step": 6637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.866024800844049e-06,
+      "loss": 0.3952,
+      "step": 6638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8631825960110866e-06,
+      "loss": 0.403,
+      "step": 6639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.86034067220078e-06,
+      "loss": 0.3584,
+      "step": 6640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857499029667738e-06,
+      "loss": 0.3845,
+      "step": 6641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854657668666535e-06,
+      "loss": 0.3607,
+      "step": 6642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.851816589451725e-06,
+      "loss": 0.3485,
+      "step": 6643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8489757922778355e-06,
+      "loss": 0.3622,
+      "step": 6644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846135277399368e-06,
+      "loss": 0.4052,
+      "step": 6645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.843295045070806e-06,
+      "loss": 0.3618,
+      "step": 6646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840455095546594e-06,
+      "loss": 0.3445,
+      "step": 6647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837615429081161e-06,
+      "loss": 0.3326,
+      "step": 6648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.83477604592891e-06,
+      "loss": 0.3715,
+      "step": 6649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.831936946344216e-06,
+      "loss": 0.3316,
+      "step": 6650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8290981305814315e-06,
+      "loss": 0.3637,
+      "step": 6651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.826259598894879e-06,
+      "loss": 0.3854,
+      "step": 6652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823421351538857e-06,
+      "loss": 0.353,
+      "step": 6653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.820583388767643e-06,
+      "loss": 0.3332,
+      "step": 6654
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.817745710835486e-06,
+      "loss": 0.3437,
+      "step": 6655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814908317996607e-06,
+      "loss": 0.3265,
+      "step": 6656
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8120712105052026e-06,
+      "loss": 0.3799,
+      "step": 6657
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8092343886154454e-06,
+      "loss": 0.353,
+      "step": 6658
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806397852581486e-06,
+      "loss": 0.4038,
+      "step": 6659
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803561602657445e-06,
+      "loss": 0.3612,
+      "step": 6660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800725639097412e-06,
+      "loss": 0.3856,
+      "step": 6661
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797889962155457e-06,
+      "loss": 0.3437,
+      "step": 6662
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795054572085631e-06,
+      "loss": 0.337,
+      "step": 6663
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.792219469141945e-06,
+      "loss": 0.3576,
+      "step": 6664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.789384653578401e-06,
+      "loss": 0.3811,
+      "step": 6665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7865501256489515e-06,
+      "loss": 0.389,
+      "step": 6666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7837158856075495e-06,
+      "loss": 0.3906,
+      "step": 6667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7808819337081045e-06,
+      "loss": 0.3224,
+      "step": 6668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7780482702045095e-06,
+      "loss": 0.3826,
+      "step": 6669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.775214895350624e-06,
+      "loss": 0.3632,
+      "step": 6670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772381809400288e-06,
+      "loss": 0.3486,
+      "step": 6671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769549012607313e-06,
+      "loss": 0.3662,
+      "step": 6672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.766716505225484e-06,
+      "loss": 0.3487,
+      "step": 6673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763884287508565e-06,
+      "loss": 0.39,
+      "step": 6674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7610523597102854e-06,
+      "loss": 0.3165,
+      "step": 6675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758220722084355e-06,
+      "loss": 0.3248,
+      "step": 6676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7553893748844535e-06,
+      "loss": 0.3535,
+      "step": 6677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.752558318364242e-06,
+      "loss": 0.357,
+      "step": 6678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7497275527773475e-06,
+      "loss": 0.3857,
+      "step": 6679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746897078377372e-06,
+      "loss": 0.3564,
+      "step": 6680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.744066895417896e-06,
+      "loss": 0.3927,
+      "step": 6681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741237004152471e-06,
+      "loss": 0.3582,
+      "step": 6682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.738407404834625e-06,
+      "loss": 0.3624,
+      "step": 6683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735578097717852e-06,
+      "loss": 0.3941,
+      "step": 6684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732749083055627e-06,
+      "loss": 0.3648,
+      "step": 6685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729920361101401e-06,
+      "loss": 0.3623,
+      "step": 6686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.727091932108592e-06,
+      "loss": 0.3574,
+      "step": 6687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724263796330592e-06,
+      "loss": 0.3519,
+      "step": 6688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.721435954020774e-06,
+      "loss": 0.3555,
+      "step": 6689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718608405432477e-06,
+      "loss": 0.3571,
+      "step": 6690
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.715781150819015e-06,
+      "loss": 0.3277,
+      "step": 6691
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.712954190433687e-06,
+      "loss": 0.3661,
+      "step": 6692
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.710127524529746e-06,
+      "loss": 0.3385,
+      "step": 6693
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707301153360428e-06,
+      "loss": 0.3477,
+      "step": 6694
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.704475077178951e-06,
+      "loss": 0.3503,
+      "step": 6695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701649296238499e-06,
+      "loss": 0.317,
+      "step": 6696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.698823810792219e-06,
+      "loss": 0.3823,
+      "step": 6697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.695998621093248e-06,
+      "loss": 0.3669,
+      "step": 6698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693173727394693e-06,
+      "loss": 0.3362,
+      "step": 6699
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690349129949629e-06,
+      "loss": 0.3269,
+      "step": 6700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6875248290111115e-06,
+      "loss": 0.3778,
+      "step": 6701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684700824832156e-06,
+      "loss": 0.3584,
+      "step": 6702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.681877117665769e-06,
+      "loss": 0.3657,
+      "step": 6703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.679053707764917e-06,
+      "loss": 0.3699,
+      "step": 6704
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676230595382551e-06,
+      "loss": 0.3464,
+      "step": 6705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673407780771586e-06,
+      "loss": 0.3995,
+      "step": 6706
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670585264184911e-06,
+      "loss": 0.3727,
+      "step": 6707
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667763045875394e-06,
+      "loss": 0.3645,
+      "step": 6708
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664941126095872e-06,
+      "loss": 0.347,
+      "step": 6709
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6621195050991605e-06,
+      "loss": 0.3404,
+      "step": 6710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6592981831380365e-06,
+      "loss": 0.3489,
+      "step": 6711
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.656477160465263e-06,
+      "loss": 0.3527,
+      "step": 6712
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653656437333571e-06,
+      "loss": 0.3858,
+      "step": 6713
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.650836013995663e-06,
+      "loss": 0.3912,
+      "step": 6714
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648015890704219e-06,
+      "loss": 0.3647,
+      "step": 6715
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645196067711887e-06,
+      "loss": 0.353,
+      "step": 6716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6423765452712895e-06,
+      "loss": 0.3449,
+      "step": 6717
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639557323635027e-06,
+      "loss": 0.3363,
+      "step": 6718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6367384030556695e-06,
+      "loss": 0.343,
+      "step": 6719
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633919783785754e-06,
+      "loss": 0.3501,
+      "step": 6720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631101466077801e-06,
+      "loss": 0.3913,
+      "step": 6721
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628283450184299e-06,
+      "loss": 0.3524,
+      "step": 6722
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625465736357706e-06,
+      "loss": 0.3901,
+      "step": 6723
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622648324850465e-06,
+      "loss": 0.3857,
+      "step": 6724
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619831215914974e-06,
+      "loss": 0.3983,
+      "step": 6725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6170144098036195e-06,
+      "loss": 0.3629,
+      "step": 6726
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.614197906768749e-06,
+      "loss": 0.3362,
+      "step": 6727
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6113817070627e-06,
+      "loss": 0.3684,
+      "step": 6728
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6085658109377595e-06,
+      "loss": 0.3481,
+      "step": 6729
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605750218646203e-06,
+      "loss": 0.3969,
+      "step": 6730
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602934930440276e-06,
+      "loss": 0.3767,
+      "step": 6731
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600119946572197e-06,
+      "loss": 0.3742,
+      "step": 6732
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5973052672941585e-06,
+      "loss": 0.3441,
+      "step": 6733
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594490892858314e-06,
+      "loss": 0.4057,
+      "step": 6734
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.591676823516807e-06,
+      "loss": 0.3406,
+      "step": 6735
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588863059521744e-06,
+      "loss": 0.3572,
+      "step": 6736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586049601125208e-06,
+      "loss": 0.3751,
+      "step": 6737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.583236448579244e-06,
+      "loss": 0.3927,
+      "step": 6738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5804236021358865e-06,
+      "loss": 0.3852,
+      "step": 6739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5776110620471315e-06,
+      "loss": 0.3642,
+      "step": 6740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574798828564948e-06,
+      "loss": 0.3578,
+      "step": 6741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5719869019412854e-06,
+      "loss": 0.3497,
+      "step": 6742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569175282428053e-06,
+      "loss": 0.3049,
+      "step": 6743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.566363970277143e-06,
+      "loss": 0.3879,
+      "step": 6744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.563552965740417e-06,
+      "loss": 0.3478,
+      "step": 6745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.560742269069706e-06,
+      "loss": 0.3659,
+      "step": 6746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557931880516821e-06,
+      "loss": 0.3498,
+      "step": 6747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555121800333535e-06,
+      "loss": 0.3497,
+      "step": 6748
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5523120287715995e-06,
+      "loss": 0.3718,
+      "step": 6749
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54950256608274e-06,
+      "loss": 0.3245,
+      "step": 6750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5466934125186524e-06,
+      "loss": 0.4125,
+      "step": 6751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543884568331002e-06,
+      "loss": 0.3526,
+      "step": 6752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54107603377143e-06,
+      "loss": 0.3496,
+      "step": 6753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5382678090915475e-06,
+      "loss": 0.3624,
+      "step": 6754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.53545989454294e-06,
+      "loss": 0.3415,
+      "step": 6755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5326522903771665e-06,
+      "loss": 0.3844,
+      "step": 6756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529844996845751e-06,
+      "loss": 0.3808,
+      "step": 6757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527038014200199e-06,
+      "loss": 0.3826,
+      "step": 6758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52423134269198e-06,
+      "loss": 0.4013,
+      "step": 6759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.521424982572546e-06,
+      "loss": 0.3397,
+      "step": 6760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.518618934093308e-06,
+      "loss": 0.3523,
+      "step": 6761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515813197505656e-06,
+      "loss": 0.3304,
+      "step": 6762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513007773060953e-06,
+      "loss": 0.352,
+      "step": 6763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510202661010534e-06,
+      "loss": 0.3719,
+      "step": 6764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507397861605707e-06,
+      "loss": 0.3845,
+      "step": 6765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.504593375097743e-06,
+      "loss": 0.3416,
+      "step": 6766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501789201737892e-06,
+      "loss": 0.4008,
+      "step": 6767
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4989853417773815e-06,
+      "loss": 0.3206,
+      "step": 6768
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496181795467407e-06,
+      "loss": 0.4113,
+      "step": 6769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493378563059121e-06,
+      "loss": 0.3921,
+      "step": 6770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490575644803672e-06,
+      "loss": 0.363,
+      "step": 6771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487773040952165e-06,
+      "loss": 0.3456,
+      "step": 6772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484970751755682e-06,
+      "loss": 0.3239,
+      "step": 6773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.482168777465279e-06,
+      "loss": 0.3721,
+      "step": 6774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479367118331974e-06,
+      "loss": 0.3559,
+      "step": 6775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476565774606767e-06,
+      "loss": 0.3412,
+      "step": 6776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.473764746540625e-06,
+      "loss": 0.3864,
+      "step": 6777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4709640343844904e-06,
+      "loss": 0.3797,
+      "step": 6778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468163638389272e-06,
+      "loss": 0.3379,
+      "step": 6779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465363558805853e-06,
+      "loss": 0.3538,
+      "step": 6780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462563795885088e-06,
+      "loss": 0.3898,
+      "step": 6781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459764349877806e-06,
+      "loss": 0.3625,
+      "step": 6782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4569652210348056e-06,
+      "loss": 0.4042,
+      "step": 6783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454166409606852e-06,
+      "loss": 0.3328,
+      "step": 6784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451367915844689e-06,
+      "loss": 0.3936,
+      "step": 6785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448569739999029e-06,
+      "loss": 0.376,
+      "step": 6786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445771882320555e-06,
+      "loss": 0.3887,
+      "step": 6787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442974343059928e-06,
+      "loss": 0.3254,
+      "step": 6788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.440177122467769e-06,
+      "loss": 0.321,
+      "step": 6789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437380220794679e-06,
+      "loss": 0.3625,
+      "step": 6790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434583638291227e-06,
+      "loss": 0.3834,
+      "step": 6791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431787375207961e-06,
+      "loss": 0.3559,
+      "step": 6792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428991431795384e-06,
+      "loss": 0.3993,
+      "step": 6793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426195808303986e-06,
+      "loss": 0.4021,
+      "step": 6794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4234005049842204e-06,
+      "loss": 0.3392,
+      "step": 6795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420605522086516e-06,
+      "loss": 0.3909,
+      "step": 6796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.417810859861275e-06,
+      "loss": 0.384,
+      "step": 6797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.415016518558858e-06,
+      "loss": 0.3974,
+      "step": 6798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.412222498429609e-06,
+      "loss": 0.3631,
+      "step": 6799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409428799723842e-06,
+      "loss": 0.351,
+      "step": 6800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.406635422691844e-06,
+      "loss": 0.3347,
+      "step": 6801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403842367583861e-06,
+      "loss": 0.3629,
+      "step": 6802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401049634650119e-06,
+      "loss": 0.3727,
+      "step": 6803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.39825722414082e-06,
+      "loss": 0.3373,
+      "step": 6804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395465136306129e-06,
+      "loss": 0.3642,
+      "step": 6805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.392673371396189e-06,
+      "loss": 0.376,
+      "step": 6806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389881929661104e-06,
+      "loss": 0.4319,
+      "step": 6807
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3870908113509554e-06,
+      "loss": 0.3623,
+      "step": 6808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384300016715799e-06,
+      "loss": 0.3562,
+      "step": 6809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381509546005658e-06,
+      "loss": 0.3649,
+      "step": 6810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.378719399470521e-06,
+      "loss": 0.3339,
+      "step": 6811
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.375929577360358e-06,
+      "loss": 0.2974,
+      "step": 6812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373140079925102e-06,
+      "loss": 0.3893,
+      "step": 6813
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37035090741466e-06,
+      "loss": 0.3552,
+      "step": 6814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.367562060078914e-06,
+      "loss": 0.3806,
+      "step": 6815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364773538167707e-06,
+      "loss": 0.3487,
+      "step": 6816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361985341930861e-06,
+      "loss": 0.3844,
+      "step": 6817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3591974716181645e-06,
+      "loss": 0.332,
+      "step": 6818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356409927479383e-06,
+      "loss": 0.3674,
+      "step": 6819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353622709764242e-06,
+      "loss": 0.3559,
+      "step": 6820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350835818722449e-06,
+      "loss": 0.3417,
+      "step": 6821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348049254603675e-06,
+      "loss": 0.3402,
+      "step": 6822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3452630176575636e-06,
+      "loss": 0.375,
+      "step": 6823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342477108133732e-06,
+      "loss": 0.387,
+      "step": 6824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3396915262817626e-06,
+      "loss": 0.3357,
+      "step": 6825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336906272351213e-06,
+      "loss": 0.4129,
+      "step": 6826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.334121346591609e-06,
+      "loss": 0.3426,
+      "step": 6827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331336749252447e-06,
+      "loss": 0.3162,
+      "step": 6828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.328552480583201e-06,
+      "loss": 0.3271,
+      "step": 6829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.325768540833303e-06,
+      "loss": 0.3706,
+      "step": 6830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.322984930252159e-06,
+      "loss": 0.373,
+      "step": 6831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320201649089156e-06,
+      "loss": 0.3534,
+      "step": 6832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317418697593645e-06,
+      "loss": 0.3708,
+      "step": 6833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.314636076014937e-06,
+      "loss": 0.3746,
+      "step": 6834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3118537846023265e-06,
+      "loss": 0.3669,
+      "step": 6835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.309071823605077e-06,
+      "loss": 0.3836,
+      "step": 6836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.306290193272422e-06,
+      "loss": 0.3769,
+      "step": 6837
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3035088938535626e-06,
+      "loss": 0.3849,
+      "step": 6838
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.300727925597663e-06,
+      "loss": 0.3492,
+      "step": 6839
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.297947288753875e-06,
+      "loss": 0.3664,
+      "step": 6840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295166983571308e-06,
+      "loss": 0.3651,
+      "step": 6841
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.292387010299049e-06,
+      "loss": 0.3674,
+      "step": 6842
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2896073691861455e-06,
+      "loss": 0.3857,
+      "step": 6843
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.286828060481626e-06,
+      "loss": 0.3294,
+      "step": 6844
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2840490844344805e-06,
+      "loss": 0.3655,
+      "step": 6845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281270441293677e-06,
+      "loss": 0.3626,
+      "step": 6846
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.27849213130815e-06,
+      "loss": 0.3406,
+      "step": 6847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275714154726801e-06,
+      "loss": 0.3976,
+      "step": 6848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272936511798506e-06,
+      "loss": 0.4076,
+      "step": 6849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270159202772109e-06,
+      "loss": 0.3809,
+      "step": 6850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.267382227896428e-06,
+      "loss": 0.3501,
+      "step": 6851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2646055874202446e-06,
+      "loss": 0.3451,
+      "step": 6852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.261829281592313e-06,
+      "loss": 0.3411,
+      "step": 6853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259053310661362e-06,
+      "loss": 0.3379,
+      "step": 6854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2562776748760835e-06,
+      "loss": 0.3588,
+      "step": 6855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253502374485146e-06,
+      "loss": 0.3864,
+      "step": 6856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250727409737179e-06,
+      "loss": 0.3432,
+      "step": 6857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.24795278088079e-06,
+      "loss": 0.3391,
+      "step": 6858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.245178488164554e-06,
+      "loss": 0.3235,
+      "step": 6859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242404531837019e-06,
+      "loss": 0.299,
+      "step": 6860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239630912146692e-06,
+      "loss": 0.3385,
+      "step": 6861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236857629342063e-06,
+      "loss": 0.3938,
+      "step": 6862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234084683671585e-06,
+      "loss": 0.3176,
+      "step": 6863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231312075383679e-06,
+      "loss": 0.3635,
+      "step": 6864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.228539804726747e-06,
+      "loss": 0.3497,
+      "step": 6865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225767871949142e-06,
+      "loss": 0.3438,
+      "step": 6866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222996277299201e-06,
+      "loss": 0.3639,
+      "step": 6867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22022502102523e-06,
+      "loss": 0.3604,
+      "step": 6868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217454103375498e-06,
+      "loss": 0.3433,
+      "step": 6869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.214683524598255e-06,
+      "loss": 0.3674,
+      "step": 6870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.211913284941698e-06,
+      "loss": 0.3613,
+      "step": 6871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2091433846540215e-06,
+      "loss": 0.353,
+      "step": 6872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2063738239833705e-06,
+      "loss": 0.3509,
+      "step": 6873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203604603177873e-06,
+      "loss": 0.3758,
+      "step": 6874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2008357224856054e-06,
+      "loss": 0.3722,
+      "step": 6875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.198067182154639e-06,
+      "loss": 0.3258,
+      "step": 6876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.195298982433e-06,
+      "loss": 0.3627,
+      "step": 6877
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.192531123568687e-06,
+      "loss": 0.3572,
+      "step": 6878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.18976360580967e-06,
+      "loss": 0.3595,
+      "step": 6879
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186996429403882e-06,
+      "loss": 0.3902,
+      "step": 6880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184229594599234e-06,
+      "loss": 0.3509,
+      "step": 6881
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181463101643601e-06,
+      "loss": 0.3663,
+      "step": 6882
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178696950784833e-06,
+      "loss": 0.3721,
+      "step": 6883
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.175931142270737e-06,
+      "loss": 0.35,
+      "step": 6884
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.3743,
+      "step": 6885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1704005532676845e-06,
+      "loss": 0.3532,
+      "step": 6886
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167635773274203e-06,
+      "loss": 0.367,
+      "step": 6887
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.164871336616356e-06,
+      "loss": 0.3424,
+      "step": 6888
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162107243541797e-06,
+      "loss": 0.368,
+      "step": 6889
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159343494298161e-06,
+      "loss": 0.3618,
+      "step": 6890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156580089133048e-06,
+      "loss": 0.3313,
+      "step": 6891
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153817028294031e-06,
+      "loss": 0.4311,
+      "step": 6892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151054312028641e-06,
+      "loss": 0.3684,
+      "step": 6893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14829194058439e-06,
+      "loss": 0.3477,
+      "step": 6894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145529914208752e-06,
+      "loss": 0.3492,
+      "step": 6895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1427682331491745e-06,
+      "loss": 0.3769,
+      "step": 6896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.140006897653077e-06,
+      "loss": 0.3537,
+      "step": 6897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.137245907967835e-06,
+      "loss": 0.3771,
+      "step": 6898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.134485264340804e-06,
+      "loss": 0.3556,
+      "step": 6899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.131724967019306e-06,
+      "loss": 0.3477,
+      "step": 6900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.128965016250637e-06,
+      "loss": 0.3669,
+      "step": 6901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126205412282049e-06,
+      "loss": 0.4044,
+      "step": 6902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1234461553607746e-06,
+      "loss": 0.362,
+      "step": 6903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.120687245734007e-06,
+      "loss": 0.3527,
+      "step": 6904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117928683648921e-06,
+      "loss": 0.3656,
+      "step": 6905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.115170469352649e-06,
+      "loss": 0.3886,
+      "step": 6906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.112412603092288e-06,
+      "loss": 0.3747,
+      "step": 6907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.109655085114918e-06,
+      "loss": 0.3303,
+      "step": 6908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1068979156675815e-06,
+      "loss": 0.3621,
+      "step": 6909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104141094997285e-06,
+      "loss": 0.3735,
+      "step": 6910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1013846233510125e-06,
+      "loss": 0.3841,
+      "step": 6911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.098628500975707e-06,
+      "loss": 0.3578,
+      "step": 6912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095872728118287e-06,
+      "loss": 0.3642,
+      "step": 6913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.093117305025639e-06,
+      "loss": 0.3913,
+      "step": 6914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09036223194462e-06,
+      "loss": 0.3173,
+      "step": 6915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0876075091220456e-06,
+      "loss": 0.3504,
+      "step": 6916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084853136804711e-06,
+      "loss": 0.3581,
+      "step": 6917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0820991152393775e-06,
+      "loss": 0.34,
+      "step": 6918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079345444672772e-06,
+      "loss": 0.3684,
+      "step": 6919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076592125351594e-06,
+      "loss": 0.3408,
+      "step": 6920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073839157522504e-06,
+      "loss": 0.3827,
+      "step": 6921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0710865414321405e-06,
+      "loss": 0.4122,
+      "step": 6922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068334277327105e-06,
+      "loss": 0.3275,
+      "step": 6923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.06558236545397e-06,
+      "loss": 0.3538,
+      "step": 6924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062830806059272e-06,
+      "loss": 0.3767,
+      "step": 6925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.060079599389521e-06,
+      "loss": 0.3666,
+      "step": 6926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0573287456911924e-06,
+      "loss": 0.3632,
+      "step": 6927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054578245210732e-06,
+      "loss": 0.3935,
+      "step": 6928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0518280981945545e-06,
+      "loss": 0.36,
+      "step": 6929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049078304889036e-06,
+      "loss": 0.3523,
+      "step": 6930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0463288655405316e-06,
+      "loss": 0.3827,
+      "step": 6931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043579780395354e-06,
+      "loss": 0.38,
+      "step": 6932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040831049699799e-06,
+      "loss": 0.4125,
+      "step": 6933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.038082673700112e-06,
+      "loss": 0.3848,
+      "step": 6934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.035334652642517e-06,
+      "loss": 0.3823,
+      "step": 6935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032586986773207e-06,
+      "loss": 0.362,
+      "step": 6936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029839676338341e-06,
+      "loss": 0.3114,
+      "step": 6937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027092721584052e-06,
+      "loss": 0.3412,
+      "step": 6938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.024346122756426e-06,
+      "loss": 0.3299,
+      "step": 6939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0215998801015276e-06,
+      "loss": 0.349,
+      "step": 6940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018853993865392e-06,
+      "loss": 0.376,
+      "step": 6941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.016108464294024e-06,
+      "loss": 0.3413,
+      "step": 6942
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.013363291633378e-06,
+      "loss": 0.3518,
+      "step": 6943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0106184761294e-06,
+      "loss": 0.3712,
+      "step": 6944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007874018027991e-06,
+      "loss": 0.3691,
+      "step": 6945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.005129917575023e-06,
+      "loss": 0.3362,
+      "step": 6946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002386175016339e-06,
+      "loss": 0.3419,
+      "step": 6947
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99964279059774e-06,
+      "loss": 0.3604,
+      "step": 6948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996899764565005e-06,
+      "loss": 0.325,
+      "step": 6949
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994157097163877e-06,
+      "loss": 0.4128,
+      "step": 6950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9914147886400696e-06,
+      "loss": 0.3698,
+      "step": 6951
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988672839239262e-06,
+      "loss": 0.386,
+      "step": 6952
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985931249207095e-06,
+      "loss": 0.3698,
+      "step": 6953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.983190018789191e-06,
+      "loss": 0.3636,
+      "step": 6954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.980449148231128e-06,
+      "loss": 0.367,
+      "step": 6955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97770863777846e-06,
+      "loss": 0.3959,
+      "step": 6956
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9749684876767015e-06,
+      "loss": 0.3525,
+      "step": 6957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972228698171339e-06,
+      "loss": 0.3515,
+      "step": 6958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969489269507828e-06,
+      "loss": 0.3561,
+      "step": 6959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966750201931588e-06,
+      "loss": 0.354,
+      "step": 6960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964011495688012e-06,
+      "loss": 0.3626,
+      "step": 6961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961273151022449e-06,
+      "loss": 0.3303,
+      "step": 6962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958535168180229e-06,
+      "loss": 0.3623,
+      "step": 6963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955797547406639e-06,
+      "loss": 0.3058,
+      "step": 6964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953060288946946e-06,
+      "loss": 0.3957,
+      "step": 6965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9503233930463675e-06,
+      "loss": 0.3433,
+      "step": 6966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947586859950103e-06,
+      "loss": 0.3287,
+      "step": 6967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944850689903312e-06,
+      "loss": 0.3749,
+      "step": 6968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942114883151127e-06,
+      "loss": 0.3358,
+      "step": 6969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.939379439938646e-06,
+      "loss": 0.4034,
+      "step": 6970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936644360510927e-06,
+      "loss": 0.3393,
+      "step": 6971
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933909645113003e-06,
+      "loss": 0.3901,
+      "step": 6972
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931175293989876e-06,
+      "loss": 0.376,
+      "step": 6973
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9284413073865146e-06,
+      "loss": 0.3155,
+      "step": 6974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.925707685547845e-06,
+      "loss": 0.3363,
+      "step": 6975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922974428718771e-06,
+      "loss": 0.3673,
+      "step": 6976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920241537144162e-06,
+      "loss": 0.3824,
+      "step": 6977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917509011068856e-06,
+      "loss": 0.3917,
+      "step": 6978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.914776850737653e-06,
+      "loss": 0.3535,
+      "step": 6979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9120450563953216e-06,
+      "loss": 0.357,
+      "step": 6980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9093136282866014e-06,
+      "loss": 0.3585,
+      "step": 6981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906582566656196e-06,
+      "loss": 0.3695,
+      "step": 6982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9038518717487804e-06,
+      "loss": 0.3936,
+      "step": 6983
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901121543808987e-06,
+      "loss": 0.3553,
+      "step": 6984
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898391583081424e-06,
+      "loss": 0.3915,
+      "step": 6985
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.895661989810666e-06,
+      "loss": 0.3539,
+      "step": 6986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.892932764241253e-06,
+      "loss": 0.3588,
+      "step": 6987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.890203906617693e-06,
+      "loss": 0.3175,
+      "step": 6988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8874754171844576e-06,
+      "loss": 0.3228,
+      "step": 6989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884747296185988e-06,
+      "loss": 0.3326,
+      "step": 6990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8820195438666935e-06,
+      "loss": 0.3497,
+      "step": 6991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87929216047095e-06,
+      "loss": 0.3593,
+      "step": 6992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8765651462431e-06,
+      "loss": 0.3345,
+      "step": 6993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8738385014274495e-06,
+      "loss": 0.3753,
+      "step": 6994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.871112226268276e-06,
+      "loss": 0.3479,
+      "step": 6995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8683863210098224e-06,
+      "loss": 0.384,
+      "step": 6996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8656607858963014e-06,
+      "loss": 0.3586,
+      "step": 6997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.862935621171884e-06,
+      "loss": 0.3291,
+      "step": 6998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8602108270807144e-06,
+      "loss": 0.3769,
+      "step": 6999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.857486403866906e-06,
+      "loss": 0.3227,
+      "step": 7000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.854762351774533e-06,
+      "loss": 0.3615,
+      "step": 7001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852038671047645e-06,
+      "loss": 0.3401,
+      "step": 7002
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849315361930243e-06,
+      "loss": 0.3469,
+      "step": 7003
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.846592424666309e-06,
+      "loss": 0.3769,
+      "step": 7004
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843869859499787e-06,
+      "loss": 0.3807,
+      "step": 7005
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.84114766667459e-06,
+      "loss": 0.3817,
+      "step": 7006
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83842584643459e-06,
+      "loss": 0.3596,
+      "step": 7007
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.835704399023631e-06,
+      "loss": 0.3464,
+      "step": 7008
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.832983324685525e-06,
+      "loss": 0.3652,
+      "step": 7009
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830262623664049e-06,
+      "loss": 0.392,
+      "step": 7010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827542296202946e-06,
+      "loss": 0.3719,
+      "step": 7011
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824822342545926e-06,
+      "loss": 0.357,
+      "step": 7012
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822102762936667e-06,
+      "loss": 0.3445,
+      "step": 7013
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819383557618809e-06,
+      "loss": 0.3686,
+      "step": 7014
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.816664726835968e-06,
+      "loss": 0.3761,
+      "step": 7015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.813946270831711e-06,
+      "loss": 0.3556,
+      "step": 7016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811228189849583e-06,
+      "loss": 0.337,
+      "step": 7017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808510484133097e-06,
+      "loss": 0.347,
+      "step": 7018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8057931539257185e-06,
+      "loss": 0.3591,
+      "step": 7019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803076199470903e-06,
+      "loss": 0.3754,
+      "step": 7020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8003596210120486e-06,
+      "loss": 0.3572,
+      "step": 7021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7976434187925305e-06,
+      "loss": 0.3684,
+      "step": 7022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7949275930556905e-06,
+      "loss": 0.3824,
+      "step": 7023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.792212144044838e-06,
+      "loss": 0.3497,
+      "step": 7024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789497072003239e-06,
+      "loss": 0.3102,
+      "step": 7025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.786782377174134e-06,
+      "loss": 0.3374,
+      "step": 7026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784068059800733e-06,
+      "loss": 0.3499,
+      "step": 7027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781354120126206e-06,
+      "loss": 0.3838,
+      "step": 7028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.778640558393693e-06,
+      "loss": 0.3628,
+      "step": 7029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.775927374846293e-06,
+      "loss": 0.3523,
+      "step": 7030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773214569727077e-06,
+      "loss": 0.386,
+      "step": 7031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.770502143279081e-06,
+      "loss": 0.3873,
+      "step": 7032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.767790095745308e-06,
+      "loss": 0.3579,
+      "step": 7033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765078427368727e-06,
+      "loss": 0.2893,
+      "step": 7034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76236713839227e-06,
+      "loss": 0.3203,
+      "step": 7035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759656229058841e-06,
+      "loss": 0.3304,
+      "step": 7036
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.756945699611302e-06,
+      "loss": 0.3598,
+      "step": 7037
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754235550292491e-06,
+      "loss": 0.3775,
+      "step": 7038
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.751525781345197e-06,
+      "loss": 0.363,
+      "step": 7039
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.748816393012191e-06,
+      "loss": 0.3201,
+      "step": 7040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7461073855362014e-06,
+      "loss": 0.3404,
+      "step": 7041
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743398759159922e-06,
+      "loss": 0.3386,
+      "step": 7042
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740690514126017e-06,
+      "loss": 0.3599,
+      "step": 7043
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.737982650677113e-06,
+      "loss": 0.4508,
+      "step": 7044
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7352751690558025e-06,
+      "loss": 0.386,
+      "step": 7045
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732568069504647e-06,
+      "loss": 0.3523,
+      "step": 7046
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729861352266173e-06,
+      "loss": 0.3914,
+      "step": 7047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727155017582865e-06,
+      "loss": 0.3601,
+      "step": 7048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.724449065697182e-06,
+      "loss": 0.3507,
+      "step": 7049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.721743496851546e-06,
+      "loss": 0.3479,
+      "step": 7050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.719038311288342e-06,
+      "loss": 0.3574,
+      "step": 7051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.716333509249934e-06,
+      "loss": 0.3817,
+      "step": 7052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.71362909097863e-06,
+      "loss": 0.353,
+      "step": 7053
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710925056716719e-06,
+      "loss": 0.3673,
+      "step": 7054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.708221406706449e-06,
+      "loss": 0.3861,
+      "step": 7055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705518141190043e-06,
+      "loss": 0.3355,
+      "step": 7056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702815260409673e-06,
+      "loss": 0.3497,
+      "step": 7057
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70011276460749e-06,
+      "loss": 0.3277,
+      "step": 7058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.697410654025602e-06,
+      "loss": 0.3592,
+      "step": 7059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694708928906097e-06,
+      "loss": 0.3948,
+      "step": 7060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692007589491014e-06,
+      "loss": 0.3691,
+      "step": 7061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689306636022358e-06,
+      "loss": 0.3371,
+      "step": 7062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.686606068742105e-06,
+      "loss": 0.3577,
+      "step": 7063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683905887892196e-06,
+      "loss": 0.3264,
+      "step": 7064
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68120609371454e-06,
+      "loss": 0.3513,
+      "step": 7065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.678506686450995e-06,
+      "loss": 0.3968,
+      "step": 7066
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67580766634341e-06,
+      "loss": 0.3307,
+      "step": 7067
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673109033633578e-06,
+      "loss": 0.3307,
+      "step": 7068
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67041078856327e-06,
+      "loss": 0.3411,
+      "step": 7069
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6677129313742195e-06,
+      "loss": 0.3867,
+      "step": 7070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.665015462308115e-06,
+      "loss": 0.3259,
+      "step": 7071
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.662318381606624e-06,
+      "loss": 0.3508,
+      "step": 7072
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659621689511374e-06,
+      "loss": 0.3829,
+      "step": 7073
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.656925386263956e-06,
+      "loss": 0.3711,
+      "step": 7074
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654229472105928e-06,
+      "loss": 0.3572,
+      "step": 7075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651533947278813e-06,
+      "loss": 0.4074,
+      "step": 7076
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6488388120241e-06,
+      "loss": 0.4083,
+      "step": 7077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646144066583241e-06,
+      "loss": 0.3407,
+      "step": 7078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643449711197657e-06,
+      "loss": 0.3836,
+      "step": 7079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.640755746108726e-06,
+      "loss": 0.3391,
+      "step": 7080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638062171557798e-06,
+      "loss": 0.3357,
+      "step": 7081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635368987786186e-06,
+      "loss": 0.3759,
+      "step": 7082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632676195035169e-06,
+      "loss": 0.3583,
+      "step": 7083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629983793545993e-06,
+      "loss": 0.3331,
+      "step": 7084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627291783559861e-06,
+      "loss": 0.3657,
+      "step": 7085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624600165317947e-06,
+      "loss": 0.3336,
+      "step": 7086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.621908939061392e-06,
+      "loss": 0.3698,
+      "step": 7087
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.619218105031301e-06,
+      "loss": 0.3348,
+      "step": 7088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.616527663468734e-06,
+      "loss": 0.3328,
+      "step": 7089
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.613837614614726e-06,
+      "loss": 0.3945,
+      "step": 7090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611147958710274e-06,
+      "loss": 0.3546,
+      "step": 7091
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.608458695996344e-06,
+      "loss": 0.3852,
+      "step": 7092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.605769826713865e-06,
+      "loss": 0.39,
+      "step": 7093
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6030813511037204e-06,
+      "loss": 0.3654,
+      "step": 7094
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.600393269406772e-06,
+      "loss": 0.368,
+      "step": 7095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5977055818638385e-06,
+      "loss": 0.3906,
+      "step": 7096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.59501828871571e-06,
+      "loss": 0.3817,
+      "step": 7097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592331390203127e-06,
+      "loss": 0.3819,
+      "step": 7098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.589644886566814e-06,
+      "loss": 0.3448,
+      "step": 7099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5869587780474485e-06,
+      "loss": 0.3408,
+      "step": 7100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.584273064885676e-06,
+      "loss": 0.3768,
+      "step": 7101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.581587747322106e-06,
+      "loss": 0.3886,
+      "step": 7102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578902825597307e-06,
+      "loss": 0.3477,
+      "step": 7103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576218299951819e-06,
+      "loss": 0.349,
+      "step": 7104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573534170626145e-06,
+      "loss": 0.3515,
+      "step": 7105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570850437860753e-06,
+      "loss": 0.3365,
+      "step": 7106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.568167101896074e-06,
+      "loss": 0.3723,
+      "step": 7107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565484162972503e-06,
+      "loss": 0.36,
+      "step": 7108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562801621330402e-06,
+      "loss": 0.3745,
+      "step": 7109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560119477210095e-06,
+      "loss": 0.3597,
+      "step": 7110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.557437730851876e-06,
+      "loss": 0.3443,
+      "step": 7111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55475638249599e-06,
+      "loss": 0.3681,
+      "step": 7112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552075432382657e-06,
+      "loss": 0.398,
+      "step": 7113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549394880752062e-06,
+      "loss": 0.3611,
+      "step": 7114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.546714727844349e-06,
+      "loss": 0.3957,
+      "step": 7115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.54403497389963e-06,
+      "loss": 0.3333,
+      "step": 7116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541355619157981e-06,
+      "loss": 0.3538,
+      "step": 7117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538676663859442e-06,
+      "loss": 0.3743,
+      "step": 7118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.535998108244012e-06,
+      "loss": 0.3619,
+      "step": 7119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.533319952551667e-06,
+      "loss": 0.3457,
+      "step": 7120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530642197022329e-06,
+      "loss": 0.3683,
+      "step": 7121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527964841895898e-06,
+      "loss": 0.3519,
+      "step": 7122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525287887412233e-06,
+      "loss": 0.3675,
+      "step": 7123
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522611333811162e-06,
+      "loss": 0.3838,
+      "step": 7124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5199351813324764e-06,
+      "loss": 0.3451,
+      "step": 7125
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517259430215918e-06,
+      "loss": 0.3624,
+      "step": 7126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514584080701208e-06,
+      "loss": 0.3535,
+      "step": 7127
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5119091330280305e-06,
+      "loss": 0.3628,
+      "step": 7128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.509234587436029e-06,
+      "loss": 0.382,
+      "step": 7129
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5065604441648055e-06,
+      "loss": 0.36,
+      "step": 7130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503886703453933e-06,
+      "loss": 0.3559,
+      "step": 7131
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501213365542957e-06,
+      "loss": 0.3892,
+      "step": 7132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4985404306713705e-06,
+      "loss": 0.3611,
+      "step": 7133
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495867899078644e-06,
+      "loss": 0.3508,
+      "step": 7134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.493195771004198e-06,
+      "loss": 0.3527,
+      "step": 7135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490524046687427e-06,
+      "loss": 0.3364,
+      "step": 7136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.487852726367686e-06,
+      "loss": 0.3953,
+      "step": 7137
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.485181810284296e-06,
+      "loss": 0.3835,
+      "step": 7138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48251129867654e-06,
+      "loss": 0.3633,
+      "step": 7139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479841191783664e-06,
+      "loss": 0.3657,
+      "step": 7140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4771714898448816e-06,
+      "loss": 0.3648,
+      "step": 7141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4745021930993625e-06,
+      "loss": 0.3398,
+      "step": 7142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4718333017862535e-06,
+      "loss": 0.3852,
+      "step": 7143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469164816144647e-06,
+      "loss": 0.3447,
+      "step": 7144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466496736413611e-06,
+      "loss": 0.3882,
+      "step": 7145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463829062832177e-06,
+      "loss": 0.387,
+      "step": 7146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461161795639336e-06,
+      "loss": 0.3979,
+      "step": 7147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458494935074045e-06,
+      "loss": 0.3407,
+      "step": 7148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455828481375225e-06,
+      "loss": 0.3467,
+      "step": 7149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453162434781757e-06,
+      "loss": 0.3553,
+      "step": 7150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450496795532489e-06,
+      "loss": 0.3442,
+      "step": 7151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4478315638662386e-06,
+      "loss": 0.4008,
+      "step": 7152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445166740021768e-06,
+      "loss": 0.34,
+      "step": 7153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442502324237819e-06,
+      "loss": 0.3401,
+      "step": 7154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.439838316753094e-06,
+      "loss": 0.4018,
+      "step": 7155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4371747178062524e-06,
+      "loss": 0.3741,
+      "step": 7156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.434511527635935e-06,
+      "loss": 0.349,
+      "step": 7157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4318487464807215e-06,
+      "loss": 0.3619,
+      "step": 7158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429186374579168e-06,
+      "loss": 0.3428,
+      "step": 7159
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.426524412169792e-06,
+      "loss": 0.3779,
+      "step": 7160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.423862859491082e-06,
+      "loss": 0.3365,
+      "step": 7161
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4212017167814715e-06,
+      "loss": 0.3598,
+      "step": 7162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418540984279369e-06,
+      "loss": 0.3511,
+      "step": 7163
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415880662223154e-06,
+      "loss": 0.3959,
+      "step": 7164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413220750851156e-06,
+      "loss": 0.3691,
+      "step": 7165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410561250401678e-06,
+      "loss": 0.3945,
+      "step": 7166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40790216111297e-06,
+      "loss": 0.3945,
+      "step": 7167
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405243483223262e-06,
+      "loss": 0.3132,
+      "step": 7168
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402585216970737e-06,
+      "loss": 0.3657,
+      "step": 7169
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.399927362593551e-06,
+      "loss": 0.3634,
+      "step": 7170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397269920329813e-06,
+      "loss": 0.3419,
+      "step": 7171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3946128904176e-06,
+      "loss": 0.3703,
+      "step": 7172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391956273094952e-06,
+      "loss": 0.3711,
+      "step": 7173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389300068599869e-06,
+      "loss": 0.3939,
+      "step": 7174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.386644277170323e-06,
+      "loss": 0.3577,
+      "step": 7175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.383988899044232e-06,
+      "loss": 0.3688,
+      "step": 7176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381333934459494e-06,
+      "loss": 0.3521,
+      "step": 7177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37867938365396e-06,
+      "loss": 0.3689,
+      "step": 7178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37602524686545e-06,
+      "loss": 0.3619,
+      "step": 7179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.373371524331743e-06,
+      "loss": 0.366,
+      "step": 7180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3707182162905815e-06,
+      "loss": 0.3578,
+      "step": 7181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.368065322979671e-06,
+      "loss": 0.3391,
+      "step": 7182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3654128446366815e-06,
+      "loss": 0.335,
+      "step": 7183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.362760781499247e-06,
+      "loss": 0.355,
+      "step": 7184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360109133804954e-06,
+      "loss": 0.3272,
+      "step": 7185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3574579017913655e-06,
+      "loss": 0.3457,
+      "step": 7186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354807085695998e-06,
+      "loss": 0.3577,
+      "step": 7187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.352156685756336e-06,
+      "loss": 0.365,
+      "step": 7188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3495067022098255e-06,
+      "loss": 0.3516,
+      "step": 7189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346857135293873e-06,
+      "loss": 0.3703,
+      "step": 7190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.34420798524585e-06,
+      "loss": 0.3917,
+      "step": 7191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341559252303089e-06,
+      "loss": 0.3359,
+      "step": 7192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3389109367028905e-06,
+      "loss": 0.4068,
+      "step": 7193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.336263038682505e-06,
+      "loss": 0.3093,
+      "step": 7194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333615558479153e-06,
+      "loss": 0.3937,
+      "step": 7195
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.330968496330026e-06,
+      "loss": 0.3261,
+      "step": 7196
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328321852472269e-06,
+      "loss": 0.3242,
+      "step": 7197
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.325675627142992e-06,
+      "loss": 0.347,
+      "step": 7198
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3230298205792594e-06,
+      "loss": 0.3565,
+      "step": 7199
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3203844330181075e-06,
+      "loss": 0.3277,
+      "step": 7200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317739464696534e-06,
+      "loss": 0.3941,
+      "step": 7201
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315094915851501e-06,
+      "loss": 0.3683,
+      "step": 7202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312450786719918e-06,
+      "loss": 0.374,
+      "step": 7203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3098070775386805e-06,
+      "loss": 0.3438,
+      "step": 7204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30716378854463e-06,
+      "loss": 0.3751,
+      "step": 7205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3045209199745715e-06,
+      "loss": 0.3861,
+      "step": 7206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.301878472065286e-06,
+      "loss": 0.3625,
+      "step": 7207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.299236445053495e-06,
+      "loss": 0.3424,
+      "step": 7208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296594839175897e-06,
+      "loss": 0.3552,
+      "step": 7209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2939536546691495e-06,
+      "loss": 0.3232,
+      "step": 7210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.291312891769873e-06,
+      "loss": 0.3551,
+      "step": 7211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.288672550714651e-06,
+      "loss": 0.3177,
+      "step": 7212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.3754,
+      "step": 7213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2833931350825e-06,
+      "loss": 0.2894,
+      "step": 7214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280754060978548e-06,
+      "loss": 0.2189,
+      "step": 7215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278115409664602e-06,
+      "loss": 0.2861,
+      "step": 7216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.275477181377048e-06,
+      "loss": 0.2223,
+      "step": 7217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.272839376352243e-06,
+      "loss": 0.2341,
+      "step": 7218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2702019948265045e-06,
+      "loss": 0.2646,
+      "step": 7219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267565037036111e-06,
+      "loss": 0.2285,
+      "step": 7220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2649285032173045e-06,
+      "loss": 0.2217,
+      "step": 7221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.262292393606288e-06,
+      "loss": 0.2291,
+      "step": 7222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2596567084392254e-06,
+      "loss": 0.2268,
+      "step": 7223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257021447952245e-06,
+      "loss": 0.2274,
+      "step": 7224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2543866123814394e-06,
+      "loss": 0.2321,
+      "step": 7225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2517522019628494e-06,
+      "loss": 0.2171,
+      "step": 7226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249118216932496e-06,
+      "loss": 0.2258,
+      "step": 7227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246484657526347e-06,
+      "loss": 0.226,
+      "step": 7228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.24385152398035e-06,
+      "loss": 0.207,
+      "step": 7229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.241218816530394e-06,
+      "loss": 0.2125,
+      "step": 7230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2385865354123425e-06,
+      "loss": 0.2306,
+      "step": 7231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2359546808620155e-06,
+      "loss": 0.2183,
+      "step": 7232
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.233323253115199e-06,
+      "loss": 0.2005,
+      "step": 7233
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.230692252407642e-06,
+      "loss": 0.2114,
+      "step": 7234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.228061678975041e-06,
+      "loss": 0.2359,
+      "step": 7235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.225431533053073e-06,
+      "loss": 0.2153,
+      "step": 7236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22280181487737e-06,
+      "loss": 0.1966,
+      "step": 7237
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220172524683521e-06,
+      "loss": 0.2178,
+      "step": 7238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.217543662707085e-06,
+      "loss": 0.2214,
+      "step": 7239
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.214915229183571e-06,
+      "loss": 0.2348,
+      "step": 7240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212287224348457e-06,
+      "loss": 0.2332,
+      "step": 7241
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209659648437184e-06,
+      "loss": 0.237,
+      "step": 7242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.207032501685154e-06,
+      "loss": 0.2407,
+      "step": 7243
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2044057843277275e-06,
+      "loss": 0.2298,
+      "step": 7244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.201779496600228e-06,
+      "loss": 0.2277,
+      "step": 7245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.199153638737939e-06,
+      "loss": 0.2217,
+      "step": 7246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.19652821097611e-06,
+      "loss": 0.2103,
+      "step": 7247
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193903213549953e-06,
+      "loss": 0.2039,
+      "step": 7248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191278646694627e-06,
+      "loss": 0.202,
+      "step": 7249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.188654510645268e-06,
+      "loss": 0.2182,
+      "step": 7250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1860308056369694e-06,
+      "loss": 0.2423,
+      "step": 7251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1834075319047825e-06,
+      "loss": 0.2285,
+      "step": 7252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180784689683727e-06,
+      "loss": 0.2219,
+      "step": 7253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178162279208774e-06,
+      "loss": 0.2394,
+      "step": 7254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175540300714863e-06,
+      "loss": 0.1825,
+      "step": 7255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172918754436894e-06,
+      "loss": 0.2489,
+      "step": 7256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170297640609732e-06,
+      "loss": 0.217,
+      "step": 7257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167676959468186e-06,
+      "loss": 0.2457,
+      "step": 7258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.16505671124705e-06,
+      "loss": 0.2242,
+      "step": 7259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1624368961810574e-06,
+      "loss": 0.2008,
+      "step": 7260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.15981751450493e-06,
+      "loss": 0.2261,
+      "step": 7261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.157198566453321e-06,
+      "loss": 0.2483,
+      "step": 7262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1545800522608605e-06,
+      "loss": 0.2328,
+      "step": 7263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.151961972162136e-06,
+      "loss": 0.2178,
+      "step": 7264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149344326391701e-06,
+      "loss": 0.2107,
+      "step": 7265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146727115184069e-06,
+      "loss": 0.2032,
+      "step": 7266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.144110338773702e-06,
+      "loss": 0.2237,
+      "step": 7267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1414939973950375e-06,
+      "loss": 0.2424,
+      "step": 7268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138878091282472e-06,
+      "loss": 0.2259,
+      "step": 7269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136262620670364e-06,
+      "loss": 0.1955,
+      "step": 7270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13364758579302e-06,
+      "loss": 0.2431,
+      "step": 7271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.131032986884722e-06,
+      "loss": 0.2161,
+      "step": 7272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128418824179707e-06,
+      "loss": 0.2374,
+      "step": 7273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.125805097912175e-06,
+      "loss": 0.2195,
+      "step": 7274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123191808316285e-06,
+      "loss": 0.1966,
+      "step": 7275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120578955626156e-06,
+      "loss": 0.2188,
+      "step": 7276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117966540075874e-06,
+      "loss": 0.2179,
+      "step": 7277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115354561899476e-06,
+      "loss": 0.1971,
+      "step": 7278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.112743021330968e-06,
+      "loss": 0.2387,
+      "step": 7279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110131918604318e-06,
+      "loss": 0.245,
+      "step": 7280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10752125395344e-06,
+      "loss": 0.2237,
+      "step": 7281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.104911027612227e-06,
+      "loss": 0.1891,
+      "step": 7282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102301239814524e-06,
+      "loss": 0.2311,
+      "step": 7283
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099691890794139e-06,
+      "loss": 0.1936,
+      "step": 7284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.097082980784836e-06,
+      "loss": 0.2305,
+      "step": 7285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094474510020345e-06,
+      "loss": 0.2275,
+      "step": 7286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0918664787343565e-06,
+      "loss": 0.2177,
+      "step": 7287
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.089258887160519e-06,
+      "loss": 0.2132,
+      "step": 7288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086651735532445e-06,
+      "loss": 0.2271,
+      "step": 7289
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.084045024083699e-06,
+      "loss": 0.2194,
+      "step": 7290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081438753047817e-06,
+      "loss": 0.2287,
+      "step": 7291
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.07883292265829e-06,
+      "loss": 0.2353,
+      "step": 7292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076227533148569e-06,
+      "loss": 0.2575,
+      "step": 7293
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0736225847520685e-06,
+      "loss": 0.2262,
+      "step": 7294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071018077702161e-06,
+      "loss": 0.2307,
+      "step": 7295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.06841401223218e-06,
+      "loss": 0.2283,
+      "step": 7296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065810388575421e-06,
+      "loss": 0.2167,
+      "step": 7297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063207206965142e-06,
+      "loss": 0.2373,
+      "step": 7298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.060604467634549e-06,
+      "loss": 0.2151,
+      "step": 7299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058002170816821e-06,
+      "loss": 0.2191,
+      "step": 7300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.055400316745096e-06,
+      "loss": 0.2097,
+      "step": 7301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.052798905652474e-06,
+      "loss": 0.2296,
+      "step": 7302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.050197937772004e-06,
+      "loss": 0.2176,
+      "step": 7303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0475974133367045e-06,
+      "loss": 0.234,
+      "step": 7304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044997332579554e-06,
+      "loss": 0.226,
+      "step": 7305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.04239769573349e-06,
+      "loss": 0.2344,
+      "step": 7306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.039798503031408e-06,
+      "loss": 0.2181,
+      "step": 7307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.037199754706167e-06,
+      "loss": 0.2158,
+      "step": 7308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034601450990584e-06,
+      "loss": 0.2186,
+      "step": 7309
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.032003592117438e-06,
+      "loss": 0.2422,
+      "step": 7310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029406178319473e-06,
+      "loss": 0.2155,
+      "step": 7311
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.026809209829375e-06,
+      "loss": 0.1947,
+      "step": 7312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02421268687981e-06,
+      "loss": 0.2153,
+      "step": 7313
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021616609703395e-06,
+      "loss": 0.2138,
+      "step": 7314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.019020978532707e-06,
+      "loss": 0.2487,
+      "step": 7315
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.016425793600288e-06,
+      "loss": 0.2111,
+      "step": 7316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013831055138636e-06,
+      "loss": 0.2031,
+      "step": 7317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011236763380207e-06,
+      "loss": 0.2202,
+      "step": 7318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008642918557421e-06,
+      "loss": 0.2402,
+      "step": 7319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0060495209026585e-06,
+      "loss": 0.211,
+      "step": 7320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.00345657064826e-06,
+      "loss": 0.2203,
+      "step": 7321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000864068026515e-06,
+      "loss": 0.2239,
+      "step": 7322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.998272013269689e-06,
+      "loss": 0.201,
+      "step": 7323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9956804066099964e-06,
+      "loss": 0.2427,
+      "step": 7324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993089248279618e-06,
+      "loss": 0.2326,
+      "step": 7325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.990498538510691e-06,
+      "loss": 0.2305,
+      "step": 7326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9879082775353125e-06,
+      "loss": 0.2236,
+      "step": 7327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985318465585541e-06,
+      "loss": 0.2493,
+      "step": 7328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982729102893394e-06,
+      "loss": 0.2324,
+      "step": 7329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.980140189690851e-06,
+      "loss": 0.2443,
+      "step": 7330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977551726209843e-06,
+      "loss": 0.2282,
+      "step": 7331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974963712682263e-06,
+      "loss": 0.2135,
+      "step": 7332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.972376149339978e-06,
+      "loss": 0.231,
+      "step": 7333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.969789036414804e-06,
+      "loss": 0.2237,
+      "step": 7334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967202374138507e-06,
+      "loss": 0.2163,
+      "step": 7335
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964616162742826e-06,
+      "loss": 0.2258,
+      "step": 7336
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.962030402459457e-06,
+      "loss": 0.2142,
+      "step": 7337
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.959445093520053e-06,
+      "loss": 0.2237,
+      "step": 7338
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956860236156232e-06,
+      "loss": 0.1959,
+      "step": 7339
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.954275830599558e-06,
+      "loss": 0.2021,
+      "step": 7340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.951691877081572e-06,
+      "loss": 0.2122,
+      "step": 7341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949108375833765e-06,
+      "loss": 0.2394,
+      "step": 7342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.946525327087593e-06,
+      "loss": 0.2048,
+      "step": 7343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943942731074457e-06,
+      "loss": 0.2457,
+      "step": 7344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941360588025734e-06,
+      "loss": 0.22,
+      "step": 7345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938778898172754e-06,
+      "loss": 0.1995,
+      "step": 7346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936197661746805e-06,
+      "loss": 0.2138,
+      "step": 7347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933616878979138e-06,
+      "loss": 0.2101,
+      "step": 7348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931036550100962e-06,
+      "loss": 0.2186,
+      "step": 7349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.928456675343441e-06,
+      "loss": 0.2112,
+      "step": 7350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9258772549377066e-06,
+      "loss": 0.2159,
+      "step": 7351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923298289114845e-06,
+      "loss": 0.2044,
+      "step": 7352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920719778105896e-06,
+      "loss": 0.2598,
+      "step": 7353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9181417221418695e-06,
+      "loss": 0.2053,
+      "step": 7354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9155641214537265e-06,
+      "loss": 0.2174,
+      "step": 7355
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.912986976272393e-06,
+      "loss": 0.2198,
+      "step": 7356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91041028682875e-06,
+      "loss": 0.2097,
+      "step": 7357
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.907834053353639e-06,
+      "loss": 0.2105,
+      "step": 7358
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.905258276077862e-06,
+      "loss": 0.2263,
+      "step": 7359
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902682955232178e-06,
+      "loss": 0.2377,
+      "step": 7360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.900108091047307e-06,
+      "loss": 0.2275,
+      "step": 7361
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89753368375393e-06,
+      "loss": 0.2248,
+      "step": 7362
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8949597335826774e-06,
+      "loss": 0.2162,
+      "step": 7363
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.892386240764149e-06,
+      "loss": 0.2344,
+      "step": 7364
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889813205528895e-06,
+      "loss": 0.217,
+      "step": 7365
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.887240628107443e-06,
+      "loss": 0.2227,
+      "step": 7366
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884668508730252e-06,
+      "loss": 0.2039,
+      "step": 7367
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882096847627762e-06,
+      "loss": 0.2003,
+      "step": 7368
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879525645030362e-06,
+      "loss": 0.2366,
+      "step": 7369
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.876954901168402e-06,
+      "loss": 0.1947,
+      "step": 7370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8743846162721944e-06,
+      "loss": 0.2262,
+      "step": 7371
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.871814790571997e-06,
+      "loss": 0.2429,
+      "step": 7372
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869245424298047e-06,
+      "loss": 0.2484,
+      "step": 7373
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866676517680526e-06,
+      "loss": 0.215,
+      "step": 7374
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.864108070949584e-06,
+      "loss": 0.2008,
+      "step": 7375
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.861540084335314e-06,
+      "loss": 0.2367,
+      "step": 7376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.858972558067784e-06,
+      "loss": 0.2186,
+      "step": 7377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.856405492377013e-06,
+      "loss": 0.2278,
+      "step": 7378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853838887492982e-06,
+      "loss": 0.2354,
+      "step": 7379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851272743645628e-06,
+      "loss": 0.2124,
+      "step": 7380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848707061064849e-06,
+      "loss": 0.206,
+      "step": 7381
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.846141839980499e-06,
+      "loss": 0.2159,
+      "step": 7382
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.843577080622393e-06,
+      "loss": 0.2552,
+      "step": 7383
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.841012783220309e-06,
+      "loss": 0.2602,
+      "step": 7384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8384489480039694e-06,
+      "loss": 0.2212,
+      "step": 7385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.835885575203067e-06,
+      "loss": 0.2084,
+      "step": 7386
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833322665047253e-06,
+      "loss": 0.2287,
+      "step": 7387
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830760217766134e-06,
+      "loss": 0.2405,
+      "step": 7388
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.828198233589274e-06,
+      "loss": 0.2192,
+      "step": 7389
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.825636712746198e-06,
+      "loss": 0.2149,
+      "step": 7390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.823075655466389e-06,
+      "loss": 0.2575,
+      "step": 7391
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82051506197929e-06,
+      "loss": 0.2173,
+      "step": 7392
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8179549325143005e-06,
+      "loss": 0.2308,
+      "step": 7393
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.815395267300774e-06,
+      "loss": 0.2514,
+      "step": 7394
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.81283606656803e-06,
+      "loss": 0.2191,
+      "step": 7395
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810277330545342e-06,
+      "loss": 0.2093,
+      "step": 7396
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8077190594619425e-06,
+      "loss": 0.1967,
+      "step": 7397
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805161253547032e-06,
+      "loss": 0.2309,
+      "step": 7398
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80260391302975e-06,
+      "loss": 0.2324,
+      "step": 7399
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800047038139207e-06,
+      "loss": 0.1977,
+      "step": 7400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797490629104472e-06,
+      "loss": 0.2003,
+      "step": 7401
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794934686154568e-06,
+      "loss": 0.2545,
+      "step": 7402
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.792379209518482e-06,
+      "loss": 0.2167,
+      "step": 7403
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789824199425148e-06,
+      "loss": 0.2012,
+      "step": 7404
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.787269656103466e-06,
+      "loss": 0.1894,
+      "step": 7405
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7847155797823e-06,
+      "loss": 0.198,
+      "step": 7406
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.782161970690466e-06,
+      "loss": 0.207,
+      "step": 7407
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779608829056731e-06,
+      "loss": 0.192,
+      "step": 7408
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77705615510983e-06,
+      "loss": 0.2099,
+      "step": 7409
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774503949078455e-06,
+      "loss": 0.2421,
+      "step": 7410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771952211191252e-06,
+      "loss": 0.1998,
+      "step": 7411
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769400941676828e-06,
+      "loss": 0.2169,
+      "step": 7412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.766850140763748e-06,
+      "loss": 0.2085,
+      "step": 7413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764299808680534e-06,
+      "loss": 0.2321,
+      "step": 7414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.761749945655665e-06,
+      "loss": 0.227,
+      "step": 7415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759200551917584e-06,
+      "loss": 0.2146,
+      "step": 7416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.756651627694682e-06,
+      "loss": 0.2174,
+      "step": 7417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754103173215313e-06,
+      "loss": 0.2363,
+      "step": 7418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751555188707792e-06,
+      "loss": 0.1956,
+      "step": 7419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.749007674400388e-06,
+      "loss": 0.2196,
+      "step": 7420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7464606305213275e-06,
+      "loss": 0.2345,
+      "step": 7421
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743914057298799e-06,
+      "loss": 0.2233,
+      "step": 7422
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741367954960946e-06,
+      "loss": 0.2234,
+      "step": 7423
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738822323735867e-06,
+      "loss": 0.2355,
+      "step": 7424
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7362771638516265e-06,
+      "loss": 0.2248,
+      "step": 7425
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733732475536236e-06,
+      "loss": 0.2502,
+      "step": 7426
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.73118825901767e-06,
+      "loss": 0.2224,
+      "step": 7427
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728644514523865e-06,
+      "loss": 0.2059,
+      "step": 7428
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726101242282708e-06,
+      "loss": 0.2298,
+      "step": 7429
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.723558442522048e-06,
+      "loss": 0.2213,
+      "step": 7430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.721016115469691e-06,
+      "loss": 0.2233,
+      "step": 7431
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7184742613533994e-06,
+      "loss": 0.2059,
+      "step": 7432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7159328804008955e-06,
+      "loss": 0.2149,
+      "step": 7433
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713391972839859e-06,
+      "loss": 0.2204,
+      "step": 7434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71085153889792e-06,
+      "loss": 0.2041,
+      "step": 7435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708311578802677e-06,
+      "loss": 0.2145,
+      "step": 7436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.705772092781675e-06,
+      "loss": 0.1976,
+      "step": 7437
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703233081062431e-06,
+      "loss": 0.2354,
+      "step": 7438
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700694543872412e-06,
+      "loss": 0.2183,
+      "step": 7439
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.698156481439033e-06,
+      "loss": 0.2093,
+      "step": 7440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695618893989679e-06,
+      "loss": 0.2345,
+      "step": 7441
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.693081781751688e-06,
+      "loss": 0.2211,
+      "step": 7442
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690545144952359e-06,
+      "loss": 0.2314,
+      "step": 7443
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688008983818941e-06,
+      "loss": 0.2099,
+      "step": 7444
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685473298578649e-06,
+      "loss": 0.2171,
+      "step": 7445
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.682938089458648e-06,
+      "loss": 0.2018,
+      "step": 7446
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.680403356686066e-06,
+      "loss": 0.1957,
+      "step": 7447
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.677869100487987e-06,
+      "loss": 0.1853,
+      "step": 7448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675335321091446e-06,
+      "loss": 0.2169,
+      "step": 7449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.672802018723442e-06,
+      "loss": 0.2051,
+      "step": 7450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670269193610932e-06,
+      "loss": 0.1981,
+      "step": 7451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.667736845980826e-06,
+      "loss": 0.2314,
+      "step": 7452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665204976059994e-06,
+      "loss": 0.193,
+      "step": 7453
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.662673584075262e-06,
+      "loss": 0.2007,
+      "step": 7454
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.660142670253413e-06,
+      "loss": 0.2152,
+      "step": 7455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657612234821189e-06,
+      "loss": 0.2332,
+      "step": 7456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6550822780052906e-06,
+      "loss": 0.2224,
+      "step": 7457
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.652552800032366e-06,
+      "loss": 0.2194,
+      "step": 7458
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65002380112903e-06,
+      "loss": 0.2001,
+      "step": 7459
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647495281521853e-06,
+      "loss": 0.2174,
+      "step": 7460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64496724143736e-06,
+      "loss": 0.2311,
+      "step": 7461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6424396811020345e-06,
+      "loss": 0.2024,
+      "step": 7462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639912600742318e-06,
+      "loss": 0.245,
+      "step": 7463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637386000584605e-06,
+      "loss": 0.2392,
+      "step": 7464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634859880855252e-06,
+      "loss": 0.2162,
+      "step": 7465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632334241780575e-06,
+      "loss": 0.2142,
+      "step": 7466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.629809083586831e-06,
+      "loss": 0.2146,
+      "step": 7467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.627284406500252e-06,
+      "loss": 0.2095,
+      "step": 7468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624760210747016e-06,
+      "loss": 0.2269,
+      "step": 7469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6222364965532665e-06,
+      "loss": 0.24,
+      "step": 7470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619713264145102e-06,
+      "loss": 0.2271,
+      "step": 7471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6171905137485664e-06,
+      "loss": 0.2164,
+      "step": 7472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614668245589672e-06,
+      "loss": 0.2054,
+      "step": 7473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612146459894387e-06,
+      "loss": 0.1994,
+      "step": 7474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609625156888637e-06,
+      "loss": 0.1971,
+      "step": 7475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607104336798293e-06,
+      "loss": 0.2244,
+      "step": 7476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.604583999849193e-06,
+      "loss": 0.2222,
+      "step": 7477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.602064146267137e-06,
+      "loss": 0.2364,
+      "step": 7478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.599544776277871e-06,
+      "loss": 0.1913,
+      "step": 7479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597025890107106e-06,
+      "loss": 0.2284,
+      "step": 7480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594507487980498e-06,
+      "loss": 0.2161,
+      "step": 7481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591989570123667e-06,
+      "loss": 0.2268,
+      "step": 7482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5894721367621945e-06,
+      "loss": 0.2195,
+      "step": 7483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.58695518812161e-06,
+      "loss": 0.2192,
+      "step": 7484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5844387244274035e-06,
+      "loss": 0.2137,
+      "step": 7485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.581922745905022e-06,
+      "loss": 0.2078,
+      "step": 7486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5794072527798685e-06,
+      "loss": 0.2326,
+      "step": 7487
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5768922452773015e-06,
+      "loss": 0.2096,
+      "step": 7488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.574377723622639e-06,
+      "loss": 0.222,
+      "step": 7489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571863688041148e-06,
+      "loss": 0.2048,
+      "step": 7490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569350138758061e-06,
+      "loss": 0.1983,
+      "step": 7491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.56683707599856e-06,
+      "loss": 0.2329,
+      "step": 7492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.56432449998779e-06,
+      "loss": 0.208,
+      "step": 7493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.561812410950847e-06,
+      "loss": 0.2121,
+      "step": 7494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5593008091127845e-06,
+      "loss": 0.2198,
+      "step": 7495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5567896946986135e-06,
+      "loss": 0.237,
+      "step": 7496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5542790679333026e-06,
+      "loss": 0.2254,
+      "step": 7497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551768929041777e-06,
+      "loss": 0.1876,
+      "step": 7498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549259278248909e-06,
+      "loss": 0.2143,
+      "step": 7499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.546750115779538e-06,
+      "loss": 0.2274,
+      "step": 7500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544241441858457e-06,
+      "loss": 0.2288,
+      "step": 7501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54173325671041e-06,
+      "loss": 0.2195,
+      "step": 7502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5392255605601125e-06,
+      "loss": 0.2282,
+      "step": 7503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536718353632212e-06,
+      "loss": 0.2372,
+      "step": 7504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5342116361513334e-06,
+      "loss": 0.1866,
+      "step": 7505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531705408342045e-06,
+      "loss": 0.2233,
+      "step": 7506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.529199670428885e-06,
+      "loss": 0.2277,
+      "step": 7507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5266944226363256e-06,
+      "loss": 0.2258,
+      "step": 7508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.52418966518881e-06,
+      "loss": 0.2448,
+      "step": 7509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521685398310746e-06,
+      "loss": 0.2077,
+      "step": 7510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519181622226479e-06,
+      "loss": 0.2064,
+      "step": 7511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5166783371603255e-06,
+      "loss": 0.2237,
+      "step": 7512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.514175543336541e-06,
+      "loss": 0.2086,
+      "step": 7513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.511673240979352e-06,
+      "loss": 0.2284,
+      "step": 7514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.509171430312935e-06,
+      "loss": 0.2103,
+      "step": 7515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.506670111561429e-06,
+      "loss": 0.2099,
+      "step": 7516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504169284948909e-06,
+      "loss": 0.2535,
+      "step": 7517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501668950699436e-06,
+      "loss": 0.2159,
+      "step": 7518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.499169109037002e-06,
+      "loss": 0.2349,
+      "step": 7519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496669760185569e-06,
+      "loss": 0.2296,
+      "step": 7520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.494170904369051e-06,
+      "loss": 0.2212,
+      "step": 7521
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4916725418113084e-06,
+      "loss": 0.1981,
+      "step": 7522
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4891746727361705e-06,
+      "loss": 0.2334,
+      "step": 7523
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.486677297367417e-06,
+      "loss": 0.2014,
+      "step": 7524
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484180415928786e-06,
+      "loss": 0.228,
+      "step": 7525
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481684028643967e-06,
+      "loss": 0.2097,
+      "step": 7526
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479188135736609e-06,
+      "loss": 0.2089,
+      "step": 7527
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.476692737430312e-06,
+      "loss": 0.2221,
+      "step": 7528
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.474197833948639e-06,
+      "loss": 0.2293,
+      "step": 7529
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.471703425515107e-06,
+      "loss": 0.2273,
+      "step": 7530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469209512353177e-06,
+      "loss": 0.2387,
+      "step": 7531
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466716094686282e-06,
+      "loss": 0.2274,
+      "step": 7532
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.464223172737801e-06,
+      "loss": 0.1982,
+      "step": 7533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461730746731065e-06,
+      "loss": 0.1741,
+      "step": 7534
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.459238816889384e-06,
+      "loss": 0.2095,
+      "step": 7535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4567473834359895e-06,
+      "loss": 0.23,
+      "step": 7536
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454256446594092e-06,
+      "loss": 0.2133,
+      "step": 7537
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45176600658685e-06,
+      "loss": 0.2182,
+      "step": 7538
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44927606363738e-06,
+      "loss": 0.2011,
+      "step": 7539
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446786617968748e-06,
+      "loss": 0.2265,
+      "step": 7540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.1896,
+      "step": 7541
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441809219366056e-06,
+      "loss": 0.2098,
+      "step": 7542
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.43932126687792e-06,
+      "loss": 0.2004,
+      "step": 7543
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436833812562461e-06,
+      "loss": 0.2014,
+      "step": 7544
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434346856642522e-06,
+      "loss": 0.2223,
+      "step": 7545
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.431860399340906e-06,
+      "loss": 0.2173,
+      "step": 7546
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429374440880374e-06,
+      "loss": 0.2143,
+      "step": 7547
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426888981483641e-06,
+      "loss": 0.2327,
+      "step": 7548
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424404021373365e-06,
+      "loss": 0.2292,
+      "step": 7549
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421919560772182e-06,
+      "loss": 0.1991,
+      "step": 7550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419435599902665e-06,
+      "loss": 0.2354,
+      "step": 7551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.416952138987349e-06,
+      "loss": 0.2098,
+      "step": 7552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414469178248729e-06,
+      "loss": 0.224,
+      "step": 7553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.41198671790924e-06,
+      "loss": 0.2046,
+      "step": 7554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409504758191287e-06,
+      "loss": 0.2292,
+      "step": 7555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.407023299317223e-06,
+      "loss": 0.2192,
+      "step": 7556
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.40454234150936e-06,
+      "loss": 0.2182,
+      "step": 7557
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402061884989963e-06,
+      "loss": 0.2545,
+      "step": 7558
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3995819299812514e-06,
+      "loss": 0.2268,
+      "step": 7559
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397102476705401e-06,
+      "loss": 0.2238,
+      "step": 7560
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394623525384543e-06,
+      "loss": 0.2229,
+      "step": 7561
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.392145076240766e-06,
+      "loss": 0.2082,
+      "step": 7562
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3896671294961014e-06,
+      "loss": 0.2305,
+      "step": 7563
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.38718968537255e-06,
+      "loss": 0.2124,
+      "step": 7564
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.384712744092063e-06,
+      "loss": 0.2036,
+      "step": 7565
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.38223630587654e-06,
+      "loss": 0.2182,
+      "step": 7566
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379760370947855e-06,
+      "loss": 0.2147,
+      "step": 7567
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377284939527809e-06,
+      "loss": 0.2076,
+      "step": 7568
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.374810011838179e-06,
+      "loss": 0.1992,
+      "step": 7569
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.372335588100687e-06,
+      "loss": 0.2011,
+      "step": 7570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369861668537019e-06,
+      "loss": 0.2406,
+      "step": 7571
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3673882533688e-06,
+      "loss": 0.2388,
+      "step": 7572
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364915342817625e-06,
+      "loss": 0.218,
+      "step": 7573
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362442937105034e-06,
+      "loss": 0.2102,
+      "step": 7574
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.359971036452533e-06,
+      "loss": 0.2321,
+      "step": 7575
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.357499641081576e-06,
+      "loss": 0.2342,
+      "step": 7576
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3550287512135646e-06,
+      "loss": 0.2291,
+      "step": 7577
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.352558367069866e-06,
+      "loss": 0.2556,
+      "step": 7578
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350088488871796e-06,
+      "loss": 0.22,
+      "step": 7579
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347619116840632e-06,
+      "loss": 0.2223,
+      "step": 7580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.345150251197593e-06,
+      "loss": 0.2187,
+      "step": 7581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342681892163868e-06,
+      "loss": 0.2266,
+      "step": 7582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340214039960591e-06,
+      "loss": 0.2279,
+      "step": 7583
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337746694808854e-06,
+      "loss": 0.2074,
+      "step": 7584
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.335279856929705e-06,
+      "loss": 0.2223,
+      "step": 7585
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332813526544139e-06,
+      "loss": 0.203,
+      "step": 7586
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330347703873112e-06,
+      "loss": 0.1898,
+      "step": 7587
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3278823891375345e-06,
+      "loss": 0.2186,
+      "step": 7588
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.32541758255827e-06,
+      "loss": 0.224,
+      "step": 7589
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322953284356137e-06,
+      "loss": 0.2291,
+      "step": 7590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320489494751908e-06,
+      "loss": 0.1927,
+      "step": 7591
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.31802621396631e-06,
+      "loss": 0.2456,
+      "step": 7592
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315563442220025e-06,
+      "loss": 0.2444,
+      "step": 7593
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.313101179733692e-06,
+      "loss": 0.2262,
+      "step": 7594
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310639426727895e-06,
+      "loss": 0.219,
+      "step": 7595
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308178183423181e-06,
+      "loss": 0.199,
+      "step": 7596
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.30571745004005e-06,
+      "loss": 0.2237,
+      "step": 7597
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.303257226798955e-06,
+      "loss": 0.1976,
+      "step": 7598
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300797513920303e-06,
+      "loss": 0.2156,
+      "step": 7599
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298338311624456e-06,
+      "loss": 0.1981,
+      "step": 7600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295879620131732e-06,
+      "loss": 0.2248,
+      "step": 7601
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2934214396624e-06,
+      "loss": 0.2405,
+      "step": 7602
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290963770436687e-06,
+      "loss": 0.2653,
+      "step": 7603
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288506612674766e-06,
+      "loss": 0.2108,
+      "step": 7604
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.286049966596775e-06,
+      "loss": 0.2348,
+      "step": 7605
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283593832422794e-06,
+      "loss": 0.2048,
+      "step": 7606
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281138210372875e-06,
+      "loss": 0.2,
+      "step": 7607
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278683100667012e-06,
+      "loss": 0.2107,
+      "step": 7608
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.276228503525147e-06,
+      "loss": 0.1888,
+      "step": 7609
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273774419167187e-06,
+      "loss": 0.1996,
+      "step": 7610
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271320847812991e-06,
+      "loss": 0.2151,
+      "step": 7611
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268867789682373e-06,
+      "loss": 0.206,
+      "step": 7612
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266415244995091e-06,
+      "loss": 0.2164,
+      "step": 7613
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263963213970866e-06,
+      "loss": 0.2336,
+      "step": 7614
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.261511696829379e-06,
+      "loss": 0.2192,
+      "step": 7615
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.259060693790254e-06,
+      "loss": 0.2269,
+      "step": 7616
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256610205073074e-06,
+      "loss": 0.2171,
+      "step": 7617
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.25416023089737e-06,
+      "loss": 0.231,
+      "step": 7618
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.251710771482635e-06,
+      "loss": 0.2489,
+      "step": 7619
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2492618270483095e-06,
+      "loss": 0.2596,
+      "step": 7620
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2468133978137945e-06,
+      "loss": 0.228,
+      "step": 7621
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244365483998438e-06,
+      "loss": 0.2177,
+      "step": 7622
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.241918085821547e-06,
+      "loss": 0.2124,
+      "step": 7623
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.239471203502379e-06,
+      "loss": 0.2235,
+      "step": 7624
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2370248372601455e-06,
+      "loss": 0.202,
+      "step": 7625
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234578987314017e-06,
+      "loss": 0.2258,
+      "step": 7626
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232133653883108e-06,
+      "loss": 0.2304,
+      "step": 7627
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.229688837186494e-06,
+      "loss": 0.2342,
+      "step": 7628
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227244537443202e-06,
+      "loss": 0.2391,
+      "step": 7629
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.224800754872215e-06,
+      "loss": 0.2418,
+      "step": 7630
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222357489692465e-06,
+      "loss": 0.2316,
+      "step": 7631
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.219914742122844e-06,
+      "loss": 0.2161,
+      "step": 7632
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.21747251238219e-06,
+      "loss": 0.2183,
+      "step": 7633
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.215030800689301e-06,
+      "loss": 0.2333,
+      "step": 7634
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.21258960726293e-06,
+      "loss": 0.2546,
+      "step": 7635
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2101489323217715e-06,
+      "loss": 0.2013,
+      "step": 7636
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.207708776084486e-06,
+      "loss": 0.2109,
+      "step": 7637
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205269138769682e-06,
+      "loss": 0.2189,
+      "step": 7638
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202830020595925e-06,
+      "loss": 0.2389,
+      "step": 7639
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200391421781731e-06,
+      "loss": 0.237,
+      "step": 7640
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1979533425455695e-06,
+      "loss": 0.2458,
+      "step": 7641
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1955157831058665e-06,
+      "loss": 0.2626,
+      "step": 7642
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193078743680996e-06,
+      "loss": 0.2271,
+      "step": 7643
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190642224489295e-06,
+      "loss": 0.2045,
+      "step": 7644
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18820622574904e-06,
+      "loss": 0.2025,
+      "step": 7645
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185770747678466e-06,
+      "loss": 0.2404,
+      "step": 7646
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1833357904957725e-06,
+      "loss": 0.2195,
+      "step": 7647
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1809013544191014e-06,
+      "loss": 0.2285,
+      "step": 7648
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1784674396665514e-06,
+      "loss": 0.2016,
+      "step": 7649
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176034046456167e-06,
+      "loss": 0.1986,
+      "step": 7650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173601175005955e-06,
+      "loss": 0.1976,
+      "step": 7651
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.171168825533875e-06,
+      "loss": 0.2311,
+      "step": 7652
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168736998257835e-06,
+      "loss": 0.2162,
+      "step": 7653
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1663056933956995e-06,
+      "loss": 0.2086,
+      "step": 7654
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163874911165285e-06,
+      "loss": 0.2198,
+      "step": 7655
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161444651784362e-06,
+      "loss": 0.2006,
+      "step": 7656
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.159014915470654e-06,
+      "loss": 0.2251,
+      "step": 7657
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15658570244184e-06,
+      "loss": 0.2135,
+      "step": 7658
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154157012915543e-06,
+      "loss": 0.1984,
+      "step": 7659
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15172884710935e-06,
+      "loss": 0.2385,
+      "step": 7660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149301205240795e-06,
+      "loss": 0.225,
+      "step": 7661
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146874087527366e-06,
+      "loss": 0.1859,
+      "step": 7662
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144447494186507e-06,
+      "loss": 0.2172,
+      "step": 7663
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142021425435612e-06,
+      "loss": 0.2253,
+      "step": 7664
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.139595881492028e-06,
+      "loss": 0.2275,
+      "step": 7665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.137170862573056e-06,
+      "loss": 0.2254,
+      "step": 7666
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.134746368895955e-06,
+      "loss": 0.189,
+      "step": 7667
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132322400677923e-06,
+      "loss": 0.2022,
+      "step": 7668
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.129898958136124e-06,
+      "loss": 0.208,
+      "step": 7669
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127476041487668e-06,
+      "loss": 0.223,
+      "step": 7670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125053650949623e-06,
+      "loss": 0.2307,
+      "step": 7671
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1226317867390075e-06,
+      "loss": 0.2259,
+      "step": 7672
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12021044907279e-06,
+      "loss": 0.1984,
+      "step": 7673
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117789638167897e-06,
+      "loss": 0.2145,
+      "step": 7674
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1153693542412045e-06,
+      "loss": 0.2362,
+      "step": 7675
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112949597509545e-06,
+      "loss": 0.1877,
+      "step": 7676
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110530368189695e-06,
+      "loss": 0.2348,
+      "step": 7677
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108111666498387e-06,
+      "loss": 0.2303,
+      "step": 7678
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105693492652317e-06,
+      "loss": 0.2175,
+      "step": 7679
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1032758468681275e-06,
+      "loss": 0.1876,
+      "step": 7680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100858729362403e-06,
+      "loss": 0.2143,
+      "step": 7681
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098442140351692e-06,
+      "loss": 0.2396,
+      "step": 7682
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096026080052494e-06,
+      "loss": 0.2101,
+      "step": 7683
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093610548681261e-06,
+      "loss": 0.197,
+      "step": 7684
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091195546454398e-06,
+      "loss": 0.2316,
+      "step": 7685
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0887810735882525e-06,
+      "loss": 0.2327,
+      "step": 7686
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086367130299144e-06,
+      "loss": 0.2245,
+      "step": 7687
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.083953716803329e-06,
+      "loss": 0.2166,
+      "step": 7688
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081540833317023e-06,
+      "loss": 0.2202,
+      "step": 7689
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079128480056397e-06,
+      "loss": 0.2057,
+      "step": 7690
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076716657237561e-06,
+      "loss": 0.2257,
+      "step": 7691
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0743053650765895e-06,
+      "loss": 0.2137,
+      "step": 7692
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071894603789508e-06,
+      "loss": 0.2121,
+      "step": 7693
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0694843735922926e-06,
+      "loss": 0.2145,
+      "step": 7694
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.067074674700873e-06,
+      "loss": 0.2183,
+      "step": 7695
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.064665507331128e-06,
+      "loss": 0.2338,
+      "step": 7696
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062256871698893e-06,
+      "loss": 0.2372,
+      "step": 7697
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059848768019954e-06,
+      "loss": 0.2264,
+      "step": 7698
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057441196510051e-06,
+      "loss": 0.2213,
+      "step": 7699
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.055034157384871e-06,
+      "loss": 0.2023,
+      "step": 7700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.052627650860057e-06,
+      "loss": 0.2299,
+      "step": 7701
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.050221677151207e-06,
+      "loss": 0.2312,
+      "step": 7702
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.047816236473866e-06,
+      "loss": 0.2302,
+      "step": 7703
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045411329043536e-06,
+      "loss": 0.2152,
+      "step": 7704
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043006955075667e-06,
+      "loss": 0.2213,
+      "step": 7705
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.040603114785665e-06,
+      "loss": 0.221,
+      "step": 7706
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038199808388884e-06,
+      "loss": 0.2253,
+      "step": 7707
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.03579703610064e-06,
+      "loss": 0.2115,
+      "step": 7708
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.033394798136182e-06,
+      "loss": 0.2209,
+      "step": 7709
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.030993094710729e-06,
+      "loss": 0.1992,
+      "step": 7710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028591926039442e-06,
+      "loss": 0.2555,
+      "step": 7711
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026191292337448e-06,
+      "loss": 0.2489,
+      "step": 7712
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023791193819807e-06,
+      "loss": 0.2357,
+      "step": 7713
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.021391630701542e-06,
+      "loss": 0.2151,
+      "step": 7714
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.018992603197627e-06,
+      "loss": 0.2289,
+      "step": 7715
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.016594111522987e-06,
+      "loss": 0.2335,
+      "step": 7716
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014196155892503e-06,
+      "loss": 0.2073,
+      "step": 7717
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011798736520992e-06,
+      "loss": 0.1957,
+      "step": 7718
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0094018536232485e-06,
+      "loss": 0.2222,
+      "step": 7719
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007005507414001e-06,
+      "loss": 0.2226,
+      "step": 7720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004609698107935e-06,
+      "loss": 0.2034,
+      "step": 7721
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.002214425919685e-06,
+      "loss": 0.21,
+      "step": 7722
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999819691063841e-06,
+      "loss": 0.2051,
+      "step": 7723
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997425493754943e-06,
+      "loss": 0.214,
+      "step": 7724
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995031834207486e-06,
+      "loss": 0.2161,
+      "step": 7725
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99263871263591e-06,
+      "loss": 0.2159,
+      "step": 7726
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990246129254616e-06,
+      "loss": 0.1989,
+      "step": 7727
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987854084277948e-06,
+      "loss": 0.2525,
+      "step": 7728
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985462577920208e-06,
+      "loss": 0.226,
+      "step": 7729
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983071610395646e-06,
+      "loss": 0.232,
+      "step": 7730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98068118191847e-06,
+      "loss": 0.2312,
+      "step": 7731
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.978291292702826e-06,
+      "loss": 0.2298,
+      "step": 7732
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975901942962826e-06,
+      "loss": 0.2366,
+      "step": 7733
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9735131329125255e-06,
+      "loss": 0.1993,
+      "step": 7734
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971124862765937e-06,
+      "loss": 0.2393,
+      "step": 7735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9687371327370205e-06,
+      "loss": 0.2196,
+      "step": 7736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96634994303969e-06,
+      "loss": 0.2273,
+      "step": 7737
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.963963293887809e-06,
+      "loss": 0.2284,
+      "step": 7738
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.961577185495196e-06,
+      "loss": 0.2142,
+      "step": 7739
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95919161807562e-06,
+      "loss": 0.2279,
+      "step": 7740
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.956806591842793e-06,
+      "loss": 0.2148,
+      "step": 7741
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954422107010392e-06,
+      "loss": 0.243,
+      "step": 7742
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.952038163792033e-06,
+      "loss": 0.209,
+      "step": 7743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949654762401302e-06,
+      "loss": 0.2156,
+      "step": 7744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947271903051713e-06,
+      "loss": 0.2069,
+      "step": 7745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944889585956746e-06,
+      "loss": 0.2201,
+      "step": 7746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94250781132983e-06,
+      "loss": 0.1994,
+      "step": 7747
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940126579384344e-06,
+      "loss": 0.202,
+      "step": 7748
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937745890333623e-06,
+      "loss": 0.2058,
+      "step": 7749
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935365744390942e-06,
+      "loss": 0.2602,
+      "step": 7750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932986141769534e-06,
+      "loss": 0.223,
+      "step": 7751
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930607082682592e-06,
+      "loss": 0.2223,
+      "step": 7752
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9282285673432515e-06,
+      "loss": 0.2155,
+      "step": 7753
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925850595964593e-06,
+      "loss": 0.2288,
+      "step": 7754
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92347316875966e-06,
+      "loss": 0.2075,
+      "step": 7755
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92109628594144e-06,
+      "loss": 0.2198,
+      "step": 7756
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9187199477228764e-06,
+      "loss": 0.1982,
+      "step": 7757
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9163441543168614e-06,
+      "loss": 0.1979,
+      "step": 7758
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9139689059362385e-06,
+      "loss": 0.2119,
+      "step": 7759
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911594202793803e-06,
+      "loss": 0.2143,
+      "step": 7760
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.909220045102299e-06,
+      "loss": 0.2282,
+      "step": 7761
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906846433074429e-06,
+      "loss": 0.2342,
+      "step": 7762
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.904473366922835e-06,
+      "loss": 0.2217,
+      "step": 7763
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.902100846860117e-06,
+      "loss": 0.2111,
+      "step": 7764
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899728873098829e-06,
+      "loss": 0.205,
+      "step": 7765
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8973574458514695e-06,
+      "loss": 0.2311,
+      "step": 7766
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.894986565330492e-06,
+      "loss": 0.2077,
+      "step": 7767
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892616231748301e-06,
+      "loss": 0.2288,
+      "step": 7768
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.89024644531725e-06,
+      "loss": 0.2046,
+      "step": 7769
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887877206249644e-06,
+      "loss": 0.2047,
+      "step": 7770
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88550851475774e-06,
+      "loss": 0.2151,
+      "step": 7771
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8831403710537495e-06,
+      "loss": 0.2148,
+      "step": 7772
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.880772775349825e-06,
+      "loss": 0.2138,
+      "step": 7773
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.878405727858076e-06,
+      "loss": 0.1993,
+      "step": 7774
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.876039228790565e-06,
+      "loss": 0.2154,
+      "step": 7775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873673278359303e-06,
+      "loss": 0.2175,
+      "step": 7776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8713078767762505e-06,
+      "loss": 0.2578,
+      "step": 7777
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868943024253322e-06,
+      "loss": 0.2158,
+      "step": 7778
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86657872100238e-06,
+      "loss": 0.2255,
+      "step": 7779
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864214967235238e-06,
+      "loss": 0.1889,
+      "step": 7780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.861851763163665e-06,
+      "loss": 0.2469,
+      "step": 7781
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859489108999372e-06,
+      "loss": 0.217,
+      "step": 7782
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8571270049540225e-06,
+      "loss": 0.2043,
+      "step": 7783
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854765451239243e-06,
+      "loss": 0.1864,
+      "step": 7784
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8524044480666e-06,
+      "loss": 0.2536,
+      "step": 7785
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850043995647606e-06,
+      "loss": 0.1899,
+      "step": 7786
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.847684094193733e-06,
+      "loss": 0.2143,
+      "step": 7787
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845324743916402e-06,
+      "loss": 0.2167,
+      "step": 7788
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.842965945026983e-06,
+      "loss": 0.2141,
+      "step": 7789
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840607697736797e-06,
+      "loss": 0.2056,
+      "step": 7790
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.838250002257116e-06,
+      "loss": 0.2679,
+      "step": 7791
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835892858799162e-06,
+      "loss": 0.2199,
+      "step": 7792
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833536267574108e-06,
+      "loss": 0.2004,
+      "step": 7793
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8311802287930826e-06,
+      "loss": 0.202,
+      "step": 7794
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.82882474266715e-06,
+      "loss": 0.2122,
+      "step": 7795
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.82646980940734e-06,
+      "loss": 0.2182,
+      "step": 7796
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.824115429224625e-06,
+      "loss": 0.2372,
+      "step": 7797
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821761602329933e-06,
+      "loss": 0.2155,
+      "step": 7798
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819408328934139e-06,
+      "loss": 0.2313,
+      "step": 7799
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817055609248069e-06,
+      "loss": 0.2381,
+      "step": 7800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8147034434825e-06,
+      "loss": 0.2013,
+      "step": 7801
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8123518318481565e-06,
+      "loss": 0.2008,
+      "step": 7802
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8100007745557222e-06,
+      "loss": 0.2065,
+      "step": 7803
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.807650271815817e-06,
+      "loss": 0.2058,
+      "step": 7804
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.805300323839022e-06,
+      "loss": 0.2388,
+      "step": 7805
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8029509308358647e-06,
+      "loss": 0.2212,
+      "step": 7806
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8006020930168254e-06,
+      "loss": 0.216,
+      "step": 7807
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7982538105923306e-06,
+      "loss": 0.2216,
+      "step": 7808
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7959060837727614e-06,
+      "loss": 0.2151,
+      "step": 7809
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7935589127684457e-06,
+      "loss": 0.2163,
+      "step": 7810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7912122977896637e-06,
+      "loss": 0.2178,
+      "step": 7811
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7888662390466447e-06,
+      "loss": 0.2283,
+      "step": 7812
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7865207367495716e-06,
+      "loss": 0.2523,
+      "step": 7813
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7841757911085685e-06,
+      "loss": 0.2057,
+      "step": 7814
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.781831402333714e-06,
+      "loss": 0.2273,
+      "step": 7815
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7794875706350477e-06,
+      "loss": 0.231,
+      "step": 7816
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777144296222547e-06,
+      "loss": 0.2091,
+      "step": 7817
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7748015793061365e-06,
+      "loss": 0.2281,
+      "step": 7818
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.772459420095701e-06,
+      "loss": 0.2216,
+      "step": 7819
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.770117818801069e-06,
+      "loss": 0.1955,
+      "step": 7820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.767776775632023e-06,
+      "loss": 0.2097,
+      "step": 7821
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7654362907982956e-06,
+      "loss": 0.2028,
+      "step": 7822
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7630963645095576e-06,
+      "loss": 0.212,
+      "step": 7823
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7607569969754486e-06,
+      "loss": 0.2342,
+      "step": 7824
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.758418188405547e-06,
+      "loss": 0.2355,
+      "step": 7825
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.756079939009386e-06,
+      "loss": 0.1974,
+      "step": 7826
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7537422489964382e-06,
+      "loss": 0.2074,
+      "step": 7827
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.751405118576138e-06,
+      "loss": 0.2512,
+      "step": 7828
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.749068547957865e-06,
+      "loss": 0.2175,
+      "step": 7829
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.746732537350949e-06,
+      "loss": 0.2191,
+      "step": 7830
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7443970869646684e-06,
+      "loss": 0.2253,
+      "step": 7831
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.742062197008255e-06,
+      "loss": 0.2434,
+      "step": 7832
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7397278676908866e-06,
+      "loss": 0.2425,
+      "step": 7833
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7373940992216917e-06,
+      "loss": 0.2363,
+      "step": 7834
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7350608918097543e-06,
+      "loss": 0.2209,
+      "step": 7835
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7327282456640947e-06,
+      "loss": 0.2017,
+      "step": 7836
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7303961609936933e-06,
+      "loss": 0.2156,
+      "step": 7837
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7280646380074803e-06,
+      "loss": 0.2167,
+      "step": 7838
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7257336769143326e-06,
+      "loss": 0.2014,
+      "step": 7839
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7234032779230767e-06,
+      "loss": 0.2214,
+      "step": 7840
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.72107344124249e-06,
+      "loss": 0.2233,
+      "step": 7841
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7187441670812985e-06,
+      "loss": 0.2333,
+      "step": 7842
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716415455648179e-06,
+      "loss": 0.2082,
+      "step": 7843
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7140873071517603e-06,
+      "loss": 0.2275,
+      "step": 7844
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7117597218006105e-06,
+      "loss": 0.2056,
+      "step": 7845
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7094326998032593e-06,
+      "loss": 0.2294,
+      "step": 7846
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.707106241368179e-06,
+      "loss": 0.2248,
+      "step": 7847
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7047803467037914e-06,
+      "loss": 0.1851,
+      "step": 7848
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702455016018478e-06,
+      "loss": 0.2152,
+      "step": 7849
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.700130249520555e-06,
+      "loss": 0.2207,
+      "step": 7850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.697806047418294e-06,
+      "loss": 0.2134,
+      "step": 7851
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6954824099199206e-06,
+      "loss": 0.216,
+      "step": 7852
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6931593372336026e-06,
+      "loss": 0.2179,
+      "step": 7853
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6908368295674667e-06,
+      "loss": 0.2103,
+      "step": 7854
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6885148871295696e-06,
+      "loss": 0.1917,
+      "step": 7855
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6861935101279434e-06,
+      "loss": 0.2192,
+      "step": 7856
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.683872698770552e-06,
+      "loss": 0.2179,
+      "step": 7857
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.681552453265318e-06,
+      "loss": 0.2095,
+      "step": 7858
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6792327738200995e-06,
+      "loss": 0.2163,
+      "step": 7859
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.676913660642719e-06,
+      "loss": 0.212,
+      "step": 7860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6745951139409396e-06,
+      "loss": 0.1968,
+      "step": 7861
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6722771339224794e-06,
+      "loss": 0.2172,
+      "step": 7862
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.669959720795e-06,
+      "loss": 0.2476,
+      "step": 7863
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6676428747661165e-06,
+      "loss": 0.2158,
+      "step": 7864
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6653265960433903e-06,
+      "loss": 0.1901,
+      "step": 7865
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6630108848343347e-06,
+      "loss": 0.2223,
+      "step": 7866
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.660695741346414e-06,
+      "loss": 0.2106,
+      "step": 7867
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6583811657870306e-06,
+      "loss": 0.234,
+      "step": 7868
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6560671583635467e-06,
+      "loss": 0.1983,
+      "step": 7869
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.653753719283273e-06,
+      "loss": 0.2099,
+      "step": 7870
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6514408487534636e-06,
+      "loss": 0.239,
+      "step": 7871
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6491285469813287e-06,
+      "loss": 0.2012,
+      "step": 7872
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6468168141740214e-06,
+      "loss": 0.1906,
+      "step": 7873
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.644505650538647e-06,
+      "loss": 0.2169,
+      "step": 7874
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6421950562822584e-06,
+      "loss": 0.235,
+      "step": 7875
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639885031611863e-06,
+      "loss": 0.2051,
+      "step": 7876
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6375755767344047e-06,
+      "loss": 0.2099,
+      "step": 7877
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.635266691856787e-06,
+      "loss": 0.1932,
+      "step": 7878
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.63295837718586e-06,
+      "loss": 0.21,
+      "step": 7879
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6306506329284196e-06,
+      "loss": 0.1975,
+      "step": 7880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6283434592912216e-06,
+      "loss": 0.2056,
+      "step": 7881
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6260368564809525e-06,
+      "loss": 0.2381,
+      "step": 7882
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6237308247042614e-06,
+      "loss": 0.2547,
+      "step": 7883
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.621425364167741e-06,
+      "loss": 0.1787,
+      "step": 7884
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6191204750779372e-06,
+      "loss": 0.2139,
+      "step": 7885
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6168161576413364e-06,
+      "loss": 0.2067,
+      "step": 7886
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6145124120643817e-06,
+      "loss": 0.2105,
+      "step": 7887
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.612209238553457e-06,
+      "loss": 0.2062,
+      "step": 7888
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6099066373149093e-06,
+      "loss": 0.2105,
+      "step": 7889
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607604608555023e-06,
+      "loss": 0.217,
+      "step": 7890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6053031524800284e-06,
+      "loss": 0.2345,
+      "step": 7891
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.603002269296111e-06,
+      "loss": 0.2242,
+      "step": 7892
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.600701959209405e-06,
+      "loss": 0.2078,
+      "step": 7893
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5984022224259906e-06,
+      "loss": 0.2147,
+      "step": 7894
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.596103059151899e-06,
+      "loss": 0.2311,
+      "step": 7895
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5938044695931063e-06,
+      "loss": 0.2081,
+      "step": 7896
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5915064539555424e-06,
+      "loss": 0.1954,
+      "step": 7897
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5892090124450806e-06,
+      "loss": 0.1949,
+      "step": 7898
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5869121452675504e-06,
+      "loss": 0.1913,
+      "step": 7899
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.584615852628717e-06,
+      "loss": 0.2283,
+      "step": 7900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.582320134734304e-06,
+      "loss": 0.2036,
+      "step": 7901
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.580024991789984e-06,
+      "loss": 0.2067,
+      "step": 7902
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5777304240013734e-06,
+      "loss": 0.2035,
+      "step": 7903
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5754364315740387e-06,
+      "loss": 0.2269,
+      "step": 7904
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.573143014713496e-06,
+      "loss": 0.2275,
+      "step": 7905
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.570850173625209e-06,
+      "loss": 0.2063,
+      "step": 7906
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5685579085145893e-06,
+      "loss": 0.203,
+      "step": 7907
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566266219587001e-06,
+      "loss": 0.2318,
+      "step": 7908
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5639751070477478e-06,
+      "loss": 0.2246,
+      "step": 7909
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5616845711020876e-06,
+      "loss": 0.2275,
+      "step": 7910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.559394611955227e-06,
+      "loss": 0.2285,
+      "step": 7911
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.557105229812321e-06,
+      "loss": 0.2148,
+      "step": 7912
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.554816424878471e-06,
+      "loss": 0.1957,
+      "step": 7913
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5525281973587277e-06,
+      "loss": 0.2166,
+      "step": 7914
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.550240547458089e-06,
+      "loss": 0.2062,
+      "step": 7915
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5479534753815037e-06,
+      "loss": 0.2111,
+      "step": 7916
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5456669813338684e-06,
+      "loss": 0.2297,
+      "step": 7917
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.543381065520022e-06,
+      "loss": 0.2371,
+      "step": 7918
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.541095728144758e-06,
+      "loss": 0.2514,
+      "step": 7919
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5388109694128136e-06,
+      "loss": 0.2051,
+      "step": 7920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5365267895288823e-06,
+      "loss": 0.2265,
+      "step": 7921
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5342431886976027e-06,
+      "loss": 0.2063,
+      "step": 7922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.53196016712355e-06,
+      "loss": 0.2084,
+      "step": 7923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.529677725011261e-06,
+      "loss": 0.2239,
+      "step": 7924
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527395862565217e-06,
+      "loss": 0.1881,
+      "step": 7925
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.525114579989849e-06,
+      "loss": 0.2005,
+      "step": 7926
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5228338774895223e-06,
+      "loss": 0.1915,
+      "step": 7927
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.520553755268574e-06,
+      "loss": 0.2281,
+      "step": 7928
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.518274213531272e-06,
+      "loss": 0.2039,
+      "step": 7929
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5159952524818377e-06,
+      "loss": 0.2207,
+      "step": 7930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.513716872324441e-06,
+      "loss": 0.2443,
+      "step": 7931
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.511439073263194e-06,
+      "loss": 0.2079,
+      "step": 7932
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5091618555021656e-06,
+      "loss": 0.2244,
+      "step": 7933
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5068852192453652e-06,
+      "loss": 0.2122,
+      "step": 7934
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504609164696754e-06,
+      "loss": 0.2248,
+      "step": 7935
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.502333692060241e-06,
+      "loss": 0.2161,
+      "step": 7936
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.500058801539683e-06,
+      "loss": 0.2065,
+      "step": 7937
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.497784493338883e-06,
+      "loss": 0.2357,
+      "step": 7938
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4955107676615916e-06,
+      "loss": 0.237,
+      "step": 7939
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493237624711514e-06,
+      "loss": 0.2157,
+      "step": 7940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49096506469229e-06,
+      "loss": 0.2235,
+      "step": 7941
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4886930878075178e-06,
+      "loss": 0.2432,
+      "step": 7942
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48642169426074e-06,
+      "loss": 0.2087,
+      "step": 7943
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4841508842554484e-06,
+      "loss": 0.2171,
+      "step": 7944
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.481880657995079e-06,
+      "loss": 0.206,
+      "step": 7945
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479611015683022e-06,
+      "loss": 0.226,
+      "step": 7946
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4773419575226077e-06,
+      "loss": 0.2025,
+      "step": 7947
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4750734837171185e-06,
+      "loss": 0.2043,
+      "step": 7948
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.472805594469787e-06,
+      "loss": 0.1951,
+      "step": 7949
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.470538289983784e-06,
+      "loss": 0.2252,
+      "step": 7950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.468271570462235e-06,
+      "loss": 0.2103,
+      "step": 7951
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4660054361082107e-06,
+      "loss": 0.1932,
+      "step": 7952
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463739887124736e-06,
+      "loss": 0.2112,
+      "step": 7953
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4614749237147793e-06,
+      "loss": 0.2115,
+      "step": 7954
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4592105460812463e-06,
+      "loss": 0.209,
+      "step": 7955
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4569467544270053e-06,
+      "loss": 0.2345,
+      "step": 7956
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4546835489548647e-06,
+      "loss": 0.2173,
+      "step": 7957
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.452420929867584e-06,
+      "loss": 0.2204,
+      "step": 7958
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4501588973678612e-06,
+      "loss": 0.1944,
+      "step": 7959
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.44789745165835e-06,
+      "loss": 0.2045,
+      "step": 7960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.445636592941655e-06,
+      "loss": 0.1907,
+      "step": 7961
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4433763214203197e-06,
+      "loss": 0.2051,
+      "step": 7962
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4411166372968428e-06,
+      "loss": 0.2082,
+      "step": 7963
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.438857540773659e-06,
+      "loss": 0.255,
+      "step": 7964
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4365990320531595e-06,
+      "loss": 0.1864,
+      "step": 7965
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4343411113376824e-06,
+      "loss": 0.2077,
+      "step": 7966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.432083778829509e-06,
+      "loss": 0.2209,
+      "step": 7967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4298270347308727e-06,
+      "loss": 0.2022,
+      "step": 7968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.42757087924395e-06,
+      "loss": 0.2401,
+      "step": 7969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4253153125708673e-06,
+      "loss": 0.2346,
+      "step": 7970
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423060334913697e-06,
+      "loss": 0.1921,
+      "step": 7971
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4208059464744626e-06,
+      "loss": 0.2503,
+      "step": 7972
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.418552147455125e-06,
+      "loss": 0.2029,
+      "step": 7973
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4162989380576008e-06,
+      "loss": 0.24,
+      "step": 7974
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4140463184837523e-06,
+      "loss": 0.1976,
+      "step": 7975
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4117942889353896e-06,
+      "loss": 0.2114,
+      "step": 7976
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4095428496142658e-06,
+      "loss": 0.2183,
+      "step": 7977
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4072920007220856e-06,
+      "loss": 0.2163,
+      "step": 7978
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.405041742460499e-06,
+      "loss": 0.2112,
+      "step": 7979
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4027920750311018e-06,
+      "loss": 0.2138,
+      "step": 7980
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4005429986354443e-06,
+      "loss": 0.2316,
+      "step": 7981
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.398294513475009e-06,
+      "loss": 0.1969,
+      "step": 7982
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3960466197512377e-06,
+      "loss": 0.2361,
+      "step": 7983
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.393799317665516e-06,
+      "loss": 0.2324,
+      "step": 7984
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.391552607419174e-06,
+      "loss": 0.2405,
+      "step": 7985
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3893064892134998e-06,
+      "loss": 0.1879,
+      "step": 7986
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3870609632497096e-06,
+      "loss": 0.2147,
+      "step": 7987
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3848160297289813e-06,
+      "loss": 0.223,
+      "step": 7988
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382571688852433e-06,
+      "loss": 0.2293,
+      "step": 7989
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3803279408211376e-06,
+      "loss": 0.2294,
+      "step": 7990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3780847858361e-06,
+      "loss": 0.2007,
+      "step": 7991
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3758422240982814e-06,
+      "loss": 0.2336,
+      "step": 7992
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.373600255808598e-06,
+      "loss": 0.1963,
+      "step": 7993
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3713588811678987e-06,
+      "loss": 0.2253,
+      "step": 7994
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3691181003769903e-06,
+      "loss": 0.2066,
+      "step": 7995
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3668779136366127e-06,
+      "loss": 0.1954,
+      "step": 7996
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3646383211474633e-06,
+      "loss": 0.2262,
+      "step": 7997
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.362399323110186e-06,
+      "loss": 0.223,
+      "step": 7998
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3601609197253683e-06,
+      "loss": 0.1945,
+      "step": 7999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.357923111193545e-06,
+      "loss": 0.2145,
+      "step": 8000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3556858977151985e-06,
+      "loss": 0.2145,
+      "step": 8001
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3534492794907557e-06,
+      "loss": 0.2135,
+      "step": 8002
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3512132567205947e-06,
+      "loss": 0.2058,
+      "step": 8003
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3489778296050377e-06,
+      "loss": 0.2409,
+      "step": 8004
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3467429983443477e-06,
+      "loss": 0.2056,
+      "step": 8005
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.344508763138744e-06,
+      "loss": 0.2073,
+      "step": 8006
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.342275124188388e-06,
+      "loss": 0.19,
+      "step": 8007
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340042081693388e-06,
+      "loss": 0.2343,
+      "step": 8008
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.337809635853797e-06,
+      "loss": 0.1885,
+      "step": 8009
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335577786869618e-06,
+      "loss": 0.2292,
+      "step": 8010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.333346534940799e-06,
+      "loss": 0.2347,
+      "step": 8011
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.331115880267234e-06,
+      "loss": 0.2339,
+      "step": 8012
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3288858230487674e-06,
+      "loss": 0.2309,
+      "step": 8013
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3266563634851814e-06,
+      "loss": 0.2037,
+      "step": 8014
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3244275017762096e-06,
+      "loss": 0.2258,
+      "step": 8015
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3221992381215362e-06,
+      "loss": 0.2324,
+      "step": 8016
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3199715727207816e-06,
+      "loss": 0.206,
+      "step": 8017
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3177445057735312e-06,
+      "loss": 0.2255,
+      "step": 8018
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3155180374792927e-06,
+      "loss": 0.2144,
+      "step": 8019
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3132921680375363e-06,
+      "loss": 0.2273,
+      "step": 8020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3110668976476745e-06,
+      "loss": 0.2349,
+      "step": 8021
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308842226509068e-06,
+      "loss": 0.2063,
+      "step": 8022
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.306618154821016e-06,
+      "loss": 0.207,
+      "step": 8023
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.30439468278277e-06,
+      "loss": 0.2275,
+      "step": 8024
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.302171810593533e-06,
+      "loss": 0.191,
+      "step": 8025
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299949538452446e-06,
+      "loss": 0.2304,
+      "step": 8026
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.297727866558602e-06,
+      "loss": 0.2119,
+      "step": 8027
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.295506795111031e-06,
+      "loss": 0.2153,
+      "step": 8028
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2932863243087188e-06,
+      "loss": 0.2288,
+      "step": 8029
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.291066454350593e-06,
+      "loss": 0.2176,
+      "step": 8030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2888471854355318e-06,
+      "loss": 0.2114,
+      "step": 8031
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.286628517762347e-06,
+      "loss": 0.2262,
+      "step": 8032
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.284410451529816e-06,
+      "loss": 0.241,
+      "step": 8033
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2821929869366463e-06,
+      "loss": 0.1852,
+      "step": 8034
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2799761241815e-06,
+      "loss": 0.2265,
+      "step": 8035
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2777598634629835e-06,
+      "loss": 0.2358,
+      "step": 8036
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.275544204979643e-06,
+      "loss": 0.1966,
+      "step": 8037
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2733291489299777e-06,
+      "loss": 0.2162,
+      "step": 8038
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.271114695512434e-06,
+      "loss": 0.2045,
+      "step": 8039
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2689008449253977e-06,
+      "loss": 0.2151,
+      "step": 8040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.266687597367207e-06,
+      "loss": 0.1919,
+      "step": 8041
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.264474953036142e-06,
+      "loss": 0.2264,
+      "step": 8042
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2622629121304306e-06,
+      "loss": 0.2234,
+      "step": 8043
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2600514748482458e-06,
+      "loss": 0.2314,
+      "step": 8044
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.257840641387712e-06,
+      "loss": 0.2168,
+      "step": 8045
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.255630411946884e-06,
+      "loss": 0.2001,
+      "step": 8046
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2534207867237798e-06,
+      "loss": 0.2272,
+      "step": 8047
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2512117659163547e-06,
+      "loss": 0.2215,
+      "step": 8048
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2490033497225114e-06,
+      "loss": 0.2095,
+      "step": 8049
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2467955383400982e-06,
+      "loss": 0.1964,
+      "step": 8050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2445883319669113e-06,
+      "loss": 0.2146,
+      "step": 8051
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24238173080069e-06,
+      "loss": 0.2153,
+      "step": 8052
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2401757350391195e-06,
+      "loss": 0.2217,
+      "step": 8053
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237970344879835e-06,
+      "loss": 0.2287,
+      "step": 8054
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2357655605204086e-06,
+      "loss": 0.2198,
+      "step": 8055
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2335613821583657e-06,
+      "loss": 0.2443,
+      "step": 8056
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2313578099911724e-06,
+      "loss": 0.1977,
+      "step": 8057
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.22915484421625e-06,
+      "loss": 0.1973,
+      "step": 8058
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2269524850309596e-06,
+      "loss": 0.1939,
+      "step": 8059
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.224750732632599e-06,
+      "loss": 0.233,
+      "step": 8060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2225495872184244e-06,
+      "loss": 0.2205,
+      "step": 8061
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2203490489856316e-06,
+      "loss": 0.2302,
+      "step": 8062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2181491181313683e-06,
+      "loss": 0.2266,
+      "step": 8063
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.215949794852712e-06,
+      "loss": 0.2215,
+      "step": 8064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2137510793467077e-06,
+      "loss": 0.2274,
+      "step": 8065
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.21155297181033e-06,
+      "loss": 0.2051,
+      "step": 8066
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.209355472440505e-06,
+      "loss": 0.2034,
+      "step": 8067
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2071585814341065e-06,
+      "loss": 0.2094,
+      "step": 8068
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2049622989879446e-06,
+      "loss": 0.2183,
+      "step": 8069
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2027666252987833e-06,
+      "loss": 0.243,
+      "step": 8070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2005715605633303e-06,
+      "loss": 0.2177,
+      "step": 8071
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1983771049782376e-06,
+      "loss": 0.2283,
+      "step": 8072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196183258740102e-06,
+      "loss": 0.2176,
+      "step": 8073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.19399002204547e-06,
+      "loss": 0.2343,
+      "step": 8074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1917973950908266e-06,
+      "loss": 0.2308,
+      "step": 8075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1896053780726087e-06,
+      "loss": 0.2143,
+      "step": 8076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.187413971187198e-06,
+      "loss": 0.2317,
+      "step": 8077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1852231746309117e-06,
+      "loss": 0.2089,
+      "step": 8078
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1830329886000242e-06,
+      "loss": 0.2027,
+      "step": 8079
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1808434132907516e-06,
+      "loss": 0.2044,
+      "step": 8080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.178654448899253e-06,
+      "loss": 0.2205,
+      "step": 8081
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1764660956216366e-06,
+      "loss": 0.1845,
+      "step": 8082
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1742783536539513e-06,
+      "loss": 0.2118,
+      "step": 8083
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172091223192194e-06,
+      "loss": 0.2263,
+      "step": 8084
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1699047044323075e-06,
+      "loss": 0.195,
+      "step": 8085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1677187975701807e-06,
+      "loss": 0.2069,
+      "step": 8086
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.16553350280164e-06,
+      "loss": 0.1933,
+      "step": 8087
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163348820322466e-06,
+      "loss": 0.2081,
+      "step": 8088
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.161164750328377e-06,
+      "loss": 0.1909,
+      "step": 8089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.15898129301505e-06,
+      "loss": 0.2488,
+      "step": 8090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.156798448578089e-06,
+      "loss": 0.2131,
+      "step": 8091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1546162172130547e-06,
+      "loss": 0.1867,
+      "step": 8092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.152434599115448e-06,
+      "loss": 0.205,
+      "step": 8093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.15025359448072e-06,
+      "loss": 0.2238,
+      "step": 8094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.148073203504264e-06,
+      "loss": 0.1875,
+      "step": 8095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.145893426381412e-06,
+      "loss": 0.2294,
+      "step": 8096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.143714263307448e-06,
+      "loss": 0.2034,
+      "step": 8097
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.141535714477606e-06,
+      "loss": 0.2279,
+      "step": 8098
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1393577800870544e-06,
+      "loss": 0.1913,
+      "step": 8099
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.137180460330915e-06,
+      "loss": 0.216,
+      "step": 8100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1350037554042446e-06,
+      "loss": 0.2361,
+      "step": 8101
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.132827665502054e-06,
+      "loss": 0.2457,
+      "step": 8102
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1306521908192955e-06,
+      "loss": 0.2217,
+      "step": 8103
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1284773315508656e-06,
+      "loss": 0.1937,
+      "step": 8104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.126303087891608e-06,
+      "loss": 0.2445,
+      "step": 8105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1241294600363093e-06,
+      "loss": 0.223,
+      "step": 8106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1219564481797018e-06,
+      "loss": 0.2108,
+      "step": 8107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1197840525164613e-06,
+      "loss": 0.2012,
+      "step": 8108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.117612273241213e-06,
+      "loss": 0.2013,
+      "step": 8109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1154411105485183e-06,
+      "loss": 0.2561,
+      "step": 8110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1132705646328885e-06,
+      "loss": 0.2197,
+      "step": 8111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.111100635688783e-06,
+      "loss": 0.2074,
+      "step": 8112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1089313239106013e-06,
+      "loss": 0.1873,
+      "step": 8113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1067626294926868e-06,
+      "loss": 0.2119,
+      "step": 8114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1045945526293307e-06,
+      "loss": 0.2022,
+      "step": 8115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1024270935147695e-06,
+      "loss": 0.2355,
+      "step": 8116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1002602523431792e-06,
+      "loss": 0.1917,
+      "step": 8117
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098094029308689e-06,
+      "loss": 0.2312,
+      "step": 8118
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.095928424605361e-06,
+      "loss": 0.2019,
+      "step": 8119
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.093763438427212e-06,
+      "loss": 0.2343,
+      "step": 8120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091599070968199e-06,
+      "loss": 0.1926,
+      "step": 8121
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.089435322422224e-06,
+      "loss": 0.2126,
+      "step": 8122
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0872721929831338e-06,
+      "loss": 0.2361,
+      "step": 8123
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0851096828447223e-06,
+      "loss": 0.2356,
+      "step": 8124
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.082947792200722e-06,
+      "loss": 0.2248,
+      "step": 8125
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.080786521244816e-06,
+      "loss": 0.2187,
+      "step": 8126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078625870170632e-06,
+      "loss": 0.2327,
+      "step": 8127
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0764658391717304e-06,
+      "loss": 0.235,
+      "step": 8128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.074306428441629e-06,
+      "loss": 0.2188,
+      "step": 8129
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0721476381737904e-06,
+      "loss": 0.2304,
+      "step": 8130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.069989468561616e-06,
+      "loss": 0.2352,
+      "step": 8131
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0678319197984496e-06,
+      "loss": 0.2171,
+      "step": 8132
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.065674992077584e-06,
+      "loss": 0.2219,
+      "step": 8133
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.063518685592255e-06,
+      "loss": 0.227,
+      "step": 8134
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0613630005356434e-06,
+      "loss": 0.2264,
+      "step": 8135
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.059207937100874e-06,
+      "loss": 0.1876,
+      "step": 8136
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.057053495481014e-06,
+      "loss": 0.198,
+      "step": 8137
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0548996758690786e-06,
+      "loss": 0.2109,
+      "step": 8138
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052746478458024e-06,
+      "loss": 0.2475,
+      "step": 8139
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0505939034407516e-06,
+      "loss": 0.2241,
+      "step": 8140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0484419510101116e-06,
+      "loss": 0.2257,
+      "step": 8141
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0462906213588863e-06,
+      "loss": 0.2219,
+      "step": 8142
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0441399146798146e-06,
+      "loss": 0.2073,
+      "step": 8143
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.041989831165575e-06,
+      "loss": 0.2395,
+      "step": 8144
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0398403710087885e-06,
+      "loss": 0.1995,
+      "step": 8145
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037691534402023e-06,
+      "loss": 0.2117,
+      "step": 8146
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0355433215377904e-06,
+      "loss": 0.2293,
+      "step": 8147
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.033395732608544e-06,
+      "loss": 0.2239,
+      "step": 8148
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031248767806685e-06,
+      "loss": 0.2026,
+      "step": 8149
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.029102427324557e-06,
+      "loss": 0.2328,
+      "step": 8150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.026956711354444e-06,
+      "loss": 0.2319,
+      "step": 8151
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.024811620088579e-06,
+      "loss": 0.2375,
+      "step": 8152
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0226671537191386e-06,
+      "loss": 0.2255,
+      "step": 8153
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0205233124382404e-06,
+      "loss": 0.2123,
+      "step": 8154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0183800964379495e-06,
+      "loss": 0.2158,
+      "step": 8155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.016237505910272e-06,
+      "loss": 0.1938,
+      "step": 8156
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0140955410471606e-06,
+      "loss": 0.2088,
+      "step": 8157
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.01195420204051e-06,
+      "loss": 0.2253,
+      "step": 8158
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009813489082162e-06,
+      "loss": 0.2352,
+      "step": 8159
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0076734023638954e-06,
+      "loss": 0.206,
+      "step": 8160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0055339420774344e-06,
+      "loss": 0.1729,
+      "step": 8161
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0033951084144587e-06,
+      "loss": 0.2093,
+      "step": 8162
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0012569015665827e-06,
+      "loss": 0.2105,
+      "step": 8163
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999119321725358e-06,
+      "loss": 0.1805,
+      "step": 8164
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9969823690822907e-06,
+      "loss": 0.2184,
+      "step": 8165
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.994846043828827e-06,
+      "loss": 0.2042,
+      "step": 8166
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.992710346156358e-06,
+      "loss": 0.2401,
+      "step": 8167
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990575276256219e-06,
+      "loss": 0.1942,
+      "step": 8168
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9884408343196793e-06,
+      "loss": 0.2509,
+      "step": 8169
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9863070205379686e-06,
+      "loss": 0.2422,
+      "step": 8170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.98417383510225e-06,
+      "loss": 0.1931,
+      "step": 8171
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9820412782036356e-06,
+      "loss": 0.2346,
+      "step": 8172
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97990935003317e-06,
+      "loss": 0.2302,
+      "step": 8173
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9777780507818545e-06,
+      "loss": 0.2223,
+      "step": 8174
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9756473806406282e-06,
+      "loss": 0.2429,
+      "step": 8175
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.973517339800375e-06,
+      "loss": 0.2007,
+      "step": 8176
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97138792845192e-06,
+      "loss": 0.21,
+      "step": 8177
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9692591467860355e-06,
+      "loss": 0.2268,
+      "step": 8178
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.967130994993437e-06,
+      "loss": 0.2099,
+      "step": 8179
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9650034732647793e-06,
+      "loss": 0.233,
+      "step": 8180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.962876581790667e-06,
+      "loss": 0.2144,
+      "step": 8181
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.960750320761645e-06,
+      "loss": 0.2263,
+      "step": 8182
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958624690368198e-06,
+      "loss": 0.2014,
+      "step": 8183
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9564996908007603e-06,
+      "loss": 0.2139,
+      "step": 8184
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9543753222497075e-06,
+      "loss": 0.1798,
+      "step": 8185
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9522515849053577e-06,
+      "loss": 0.1961,
+      "step": 8186
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9501284789579745e-06,
+      "loss": 0.244,
+      "step": 8187
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.948006004597762e-06,
+      "loss": 0.2205,
+      "step": 8188
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9458841620148713e-06,
+      "loss": 0.2335,
+      "step": 8189
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9437629513993947e-06,
+      "loss": 0.2184,
+      "step": 8190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9416423729413703e-06,
+      "loss": 0.2307,
+      "step": 8191
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9395224268307723e-06,
+      "loss": 0.2074,
+      "step": 8192
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9374031132575253e-06,
+      "loss": 0.1961,
+      "step": 8193
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9352844324114928e-06,
+      "loss": 0.2159,
+      "step": 8194
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9331663844824943e-06,
+      "loss": 0.232,
+      "step": 8195
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9310489696602715e-06,
+      "loss": 0.2107,
+      "step": 8196
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.2053,
+      "step": 8197
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.926816040094894e-06,
+      "loss": 0.2165,
+      "step": 8198
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.92470052573096e-06,
+      "loss": 0.2083,
+      "step": 8199
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.922585645232252e-06,
+      "loss": 0.2055,
+      "step": 8200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9204713987882307e-06,
+      "loss": 0.2064,
+      "step": 8201
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9183577865883163e-06,
+      "loss": 0.2266,
+      "step": 8202
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9162448088218596e-06,
+      "loss": 0.2076,
+      "step": 8203
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9141324656781656e-06,
+      "loss": 0.1921,
+      "step": 8204
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9120207573464674e-06,
+      "loss": 0.2222,
+      "step": 8205
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9099096840159537e-06,
+      "loss": 0.2113,
+      "step": 8206
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.907799245875751e-06,
+      "loss": 0.2195,
+      "step": 8207
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9056894431149317e-06,
+      "loss": 0.2123,
+      "step": 8208
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9035802759225084e-06,
+      "loss": 0.1995,
+      "step": 8209
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9014717444874396e-06,
+      "loss": 0.2497,
+      "step": 8210
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8993638489986255e-06,
+      "loss": 0.2287,
+      "step": 8211
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.897256589644907e-06,
+      "loss": 0.1916,
+      "step": 8212
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895149966615076e-06,
+      "loss": 0.2209,
+      "step": 8213
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.893043980097854e-06,
+      "loss": 0.2223,
+      "step": 8214
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8909386302819153e-06,
+      "loss": 0.2202,
+      "step": 8215
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.888833917355878e-06,
+      "loss": 0.2128,
+      "step": 8216
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.886729841508297e-06,
+      "loss": 0.2373,
+      "step": 8217
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.884626402927674e-06,
+      "loss": 0.2302,
+      "step": 8218
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882523601802454e-06,
+      "loss": 0.2035,
+      "step": 8219
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8804214383210226e-06,
+      "loss": 0.2071,
+      "step": 8220
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.878319912671711e-06,
+      "loss": 0.2133,
+      "step": 8221
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.876219025042789e-06,
+      "loss": 0.204,
+      "step": 8222
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8741187756224775e-06,
+      "loss": 0.2204,
+      "step": 8223
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.872019164598926e-06,
+      "loss": 0.2219,
+      "step": 8224
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8699201921602415e-06,
+      "loss": 0.2129,
+      "step": 8225
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8678218584944616e-06,
+      "loss": 0.2195,
+      "step": 8226
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.865724163789585e-06,
+      "loss": 0.2121,
+      "step": 8227
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8636271082335287e-06,
+      "loss": 0.2076,
+      "step": 8228
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8615306920141696e-06,
+      "loss": 0.2308,
+      "step": 8229
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8594349153193223e-06,
+      "loss": 0.2154,
+      "step": 8230
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.857339778336744e-06,
+      "loss": 0.2132,
+      "step": 8231
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8552452812541387e-06,
+      "loss": 0.2326,
+      "step": 8232
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853151424259141e-06,
+      "loss": 0.2368,
+      "step": 8233
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.851058207539338e-06,
+      "loss": 0.2251,
+      "step": 8234
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8489656312822633e-06,
+      "loss": 0.2003,
+      "step": 8235
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8468736956753874e-06,
+      "loss": 0.2023,
+      "step": 8236
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8447824009061185e-06,
+      "loss": 0.2409,
+      "step": 8237
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8426917471618144e-06,
+      "loss": 0.2164,
+      "step": 8238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.840601734629774e-06,
+      "loss": 0.2033,
+      "step": 8239
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.838512363497239e-06,
+      "loss": 0.216,
+      "step": 8240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8364236339513918e-06,
+      "loss": 0.1947,
+      "step": 8241
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8343355461793585e-06,
+      "loss": 0.2411,
+      "step": 8242
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8322481003682088e-06,
+      "loss": 0.2202,
+      "step": 8243
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.830161296704953e-06,
+      "loss": 0.2258,
+      "step": 8244
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.828075135376548e-06,
+      "loss": 0.2068,
+      "step": 8245
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.825989616569883e-06,
+      "loss": 0.2075,
+      "step": 8246
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8239047404718e-06,
+      "loss": 0.2095,
+      "step": 8247
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8218205072690805e-06,
+      "loss": 0.1995,
+      "step": 8248
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8197369171484466e-06,
+      "loss": 0.2439,
+      "step": 8249
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8176539702965643e-06,
+      "loss": 0.2196,
+      "step": 8250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.815571666900042e-06,
+      "loss": 0.2225,
+      "step": 8251
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8134900071454297e-06,
+      "loss": 0.2144,
+      "step": 8252
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8114089912192197e-06,
+      "loss": 0.2516,
+      "step": 8253
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8093286193078517e-06,
+      "loss": 0.2139,
+      "step": 8254
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.807248891597695e-06,
+      "loss": 0.2374,
+      "step": 8255
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8051698082750735e-06,
+      "loss": 0.2079,
+      "step": 8256
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8030913695262487e-06,
+      "loss": 0.2226,
+      "step": 8257
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8010135755374235e-06,
+      "loss": 0.2126,
+      "step": 8258
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.798936426494747e-06,
+      "loss": 0.2198,
+      "step": 8259
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7968599225843053e-06,
+      "loss": 0.2288,
+      "step": 8260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7947840639921308e-06,
+      "loss": 0.2162,
+      "step": 8261
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7927088509041955e-06,
+      "loss": 0.2145,
+      "step": 8262
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7906342835064147e-06,
+      "loss": 0.1985,
+      "step": 8263
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.788560361984649e-06,
+      "loss": 0.2217,
+      "step": 8264
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786487086524692e-06,
+      "loss": 0.2033,
+      "step": 8265
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7844144573122843e-06,
+      "loss": 0.2016,
+      "step": 8266
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.782342474533117e-06,
+      "loss": 0.2196,
+      "step": 8267
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.780271138372815e-06,
+      "loss": 0.2069,
+      "step": 8268
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7782004490169413e-06,
+      "loss": 0.2135,
+      "step": 8269
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7761304066510066e-06,
+      "loss": 0.2055,
+      "step": 8270
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.774061011460464e-06,
+      "loss": 0.1947,
+      "step": 8271
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7719922636307074e-06,
+      "loss": 0.2117,
+      "step": 8272
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.769924163347073e-06,
+      "loss": 0.2168,
+      "step": 8273
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7678567107948384e-06,
+      "loss": 0.2302,
+      "step": 8274
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7657899061592243e-06,
+      "loss": 0.2223,
+      "step": 8275
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.763723749625391e-06,
+      "loss": 0.2221,
+      "step": 8276
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7616582413784465e-06,
+      "loss": 0.2226,
+      "step": 8277
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7595933816034303e-06,
+      "loss": 0.2161,
+      "step": 8278
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7575291704853325e-06,
+      "loss": 0.2191,
+      "step": 8279
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7554656082090823e-06,
+      "loss": 0.2357,
+      "step": 8280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753402694959553e-06,
+      "loss": 0.1972,
+      "step": 8281
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7513404309215563e-06,
+      "loss": 0.2185,
+      "step": 8282
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7492788162798477e-06,
+      "loss": 0.21,
+      "step": 8283
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.747217851219124e-06,
+      "loss": 0.2213,
+      "step": 8284
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7451575359240234e-06,
+      "loss": 0.2131,
+      "step": 8285
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.74309787057913e-06,
+      "loss": 0.2263,
+      "step": 8286
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.741038855368959e-06,
+      "loss": 0.2094,
+      "step": 8287
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7389804904779804e-06,
+      "loss": 0.2405,
+      "step": 8288
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7369227760905957e-06,
+      "loss": 0.2153,
+      "step": 8289
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.734865712391156e-06,
+      "loss": 0.1999,
+      "step": 8290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.732809299563949e-06,
+      "loss": 0.2046,
+      "step": 8291
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.730753537793205e-06,
+      "loss": 0.1994,
+      "step": 8292
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7286984272630966e-06,
+      "loss": 0.2165,
+      "step": 8293
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7266439681577397e-06,
+      "loss": 0.2101,
+      "step": 8294
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7245901606611914e-06,
+      "loss": 0.2244,
+      "step": 8295
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.722537004957444e-06,
+      "loss": 0.2122,
+      "step": 8296
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720484501230439e-06,
+      "loss": 0.2083,
+      "step": 8297
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7184326496640546e-06,
+      "loss": 0.2005,
+      "step": 8298
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7163814504421192e-06,
+      "loss": 0.2323,
+      "step": 8299
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7143309037483955e-06,
+      "loss": 0.1895,
+      "step": 8300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712281009766584e-06,
+      "loss": 0.1867,
+      "step": 8301
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.710231768680334e-06,
+      "loss": 0.1999,
+      "step": 8302
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.708183180673233e-06,
+      "loss": 0.1999,
+      "step": 8303
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7061352459288128e-06,
+      "loss": 0.2135,
+      "step": 8304
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704087964630546e-06,
+      "loss": 0.2166,
+      "step": 8305
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7020413369618383e-06,
+      "loss": 0.2067,
+      "step": 8306
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6999953631060504e-06,
+      "loss": 0.2478,
+      "step": 8307
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6979500432464777e-06,
+      "loss": 0.2444,
+      "step": 8308
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.695905377566358e-06,
+      "loss": 0.2342,
+      "step": 8309
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.693861366248864e-06,
+      "loss": 0.1936,
+      "step": 8310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6918180094771207e-06,
+      "loss": 0.2272,
+      "step": 8311
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.689775307434187e-06,
+      "loss": 0.2147,
+      "step": 8312
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.687733260303066e-06,
+      "loss": 0.2154,
+      "step": 8313
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.685691868266702e-06,
+      "loss": 0.2067,
+      "step": 8314
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6836511315079806e-06,
+      "loss": 0.2104,
+      "step": 8315
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6816110502097257e-06,
+      "loss": 0.2107,
+      "step": 8316
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.679571624554709e-06,
+      "loss": 0.202,
+      "step": 8317
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6775328547256397e-06,
+      "loss": 0.2034,
+      "step": 8318
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6754947409051636e-06,
+      "loss": 0.2377,
+      "step": 8319
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.673457283275873e-06,
+      "loss": 0.2168,
+      "step": 8320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6714204820203037e-06,
+      "loss": 0.2209,
+      "step": 8321
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6693843373209273e-06,
+      "loss": 0.2019,
+      "step": 8322
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6673488493601583e-06,
+      "loss": 0.1896,
+      "step": 8323
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.665314018320355e-06,
+      "loss": 0.2494,
+      "step": 8324
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.663279844383815e-06,
+      "loss": 0.2225,
+      "step": 8325
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.661246327732775e-06,
+      "loss": 0.2015,
+      "step": 8326
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6592134685494186e-06,
+      "loss": 0.2098,
+      "step": 8327
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6571812670158603e-06,
+      "loss": 0.2145,
+      "step": 8328
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6551497233141644e-06,
+      "loss": 0.2191,
+      "step": 8329
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6531188376263352e-06,
+      "loss": 0.1912,
+      "step": 8330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6510886101343126e-06,
+      "loss": 0.2339,
+      "step": 8331
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.64905904101999e-06,
+      "loss": 0.2168,
+      "step": 8332
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.647030130465186e-06,
+      "loss": 0.2335,
+      "step": 8333
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6450018786516694e-06,
+      "loss": 0.2399,
+      "step": 8334
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6429742857611486e-06,
+      "loss": 0.2243,
+      "step": 8335
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6409473519752758e-06,
+      "loss": 0.1897,
+      "step": 8336
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6389210774756335e-06,
+      "loss": 0.2173,
+      "step": 8337
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.636895462443755e-06,
+      "loss": 0.2022,
+      "step": 8338
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6348705070611158e-06,
+      "loss": 0.2217,
+      "step": 8339
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6328462115091257e-06,
+      "loss": 0.2155,
+      "step": 8340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.630822575969143e-06,
+      "loss": 0.2424,
+      "step": 8341
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6287996006224536e-06,
+      "loss": 0.2444,
+      "step": 8342
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.626777285650298e-06,
+      "loss": 0.2144,
+      "step": 8343
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6247556312338518e-06,
+      "loss": 0.2047,
+      "step": 8344
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6227346375542307e-06,
+      "loss": 0.2013,
+      "step": 8345
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.620714304792494e-06,
+      "loss": 0.2379,
+      "step": 8346
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.618694633129639e-06,
+      "loss": 0.1913,
+      "step": 8347
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6166756227466063e-06,
+      "loss": 0.2068,
+      "step": 8348
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6146572738242736e-06,
+      "loss": 0.1916,
+      "step": 8349
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6126395865434674e-06,
+      "loss": 0.219,
+      "step": 8350
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6106225610849413e-06,
+      "loss": 0.2071,
+      "step": 8351
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6086061976294008e-06,
+      "loss": 0.2523,
+      "step": 8352
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6065904963574907e-06,
+      "loss": 0.2366,
+      "step": 8353
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6045754574497917e-06,
+      "loss": 0.2255,
+      "step": 8354
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6025610810868307e-06,
+      "loss": 0.2497,
+      "step": 8355
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.60054736744907e-06,
+      "loss": 0.2262,
+      "step": 8356
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5985343167169174e-06,
+      "loss": 0.2509,
+      "step": 8357
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.596521929070718e-06,
+      "loss": 0.2019,
+      "step": 8358
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.594510204690761e-06,
+      "loss": 0.2172,
+      "step": 8359
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5924991437572688e-06,
+      "loss": 0.1904,
+      "step": 8360
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5904887464504115e-06,
+      "loss": 0.1945,
+      "step": 8361
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.588479012950297e-06,
+      "loss": 0.1949,
+      "step": 8362
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5864699434369733e-06,
+      "loss": 0.1766,
+      "step": 8363
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.584461538090437e-06,
+      "loss": 0.191,
+      "step": 8364
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5824537970906095e-06,
+      "loss": 0.2212,
+      "step": 8365
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5804467206173646e-06,
+      "loss": 0.2261,
+      "step": 8366
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.578440308850513e-06,
+      "loss": 0.2279,
+      "step": 8367
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.576434561969808e-06,
+      "loss": 0.2152,
+      "step": 8368
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5744294801549372e-06,
+      "loss": 0.2092,
+      "step": 8369
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5724250635855353e-06,
+      "loss": 0.2177,
+      "step": 8370
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5704213124411714e-06,
+      "loss": 0.2247,
+      "step": 8371
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5684182269013647e-06,
+      "loss": 0.2476,
+      "step": 8372
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.566415807145568e-06,
+      "loss": 0.2128,
+      "step": 8373
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5644140533531704e-06,
+      "loss": 0.2265,
+      "step": 8374
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.562412965703508e-06,
+      "loss": 0.224,
+      "step": 8375
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.560412544375854e-06,
+      "loss": 0.1985,
+      "step": 8376
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5584127895494292e-06,
+      "loss": 0.2103,
+      "step": 8377
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5564137014033764e-06,
+      "loss": 0.2192,
+      "step": 8378
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.554415280116802e-06,
+      "loss": 0.2342,
+      "step": 8379
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5524175258687367e-06,
+      "loss": 0.2157,
+      "step": 8380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.550420438838158e-06,
+      "loss": 0.1971,
+      "step": 8381
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5484240192039846e-06,
+      "loss": 0.2421,
+      "step": 8382
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5464282671450646e-06,
+      "loss": 0.2369,
+      "step": 8383
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5444331828402004e-06,
+      "loss": 0.2125,
+      "step": 8384
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.542438766468126e-06,
+      "loss": 0.2199,
+      "step": 8385
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.540445018207519e-06,
+      "loss": 0.2018,
+      "step": 8386
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5384519382369967e-06,
+      "loss": 0.2253,
+      "step": 8387
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5364595267351155e-06,
+      "loss": 0.2029,
+      "step": 8388
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.534467783880373e-06,
+      "loss": 0.2372,
+      "step": 8389
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.532476709851207e-06,
+      "loss": 0.2346,
+      "step": 8390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5304863048259952e-06,
+      "loss": 0.22,
+      "step": 8391
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.528496568983052e-06,
+      "loss": 0.2191,
+      "step": 8392
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5265075025006356e-06,
+      "loss": 0.2382,
+      "step": 8393
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5245191055569452e-06,
+      "loss": 0.2144,
+      "step": 8394
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5225313783301164e-06,
+      "loss": 0.2014,
+      "step": 8395
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5205443209982284e-06,
+      "loss": 0.2138,
+      "step": 8396
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5185579337392964e-06,
+      "loss": 0.2328,
+      "step": 8397
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.51657221673128e-06,
+      "loss": 0.2042,
+      "step": 8398
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5145871701520753e-06,
+      "loss": 0.2256,
+      "step": 8399
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5126027941795228e-06,
+      "loss": 0.1935,
+      "step": 8400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510619088991394e-06,
+      "loss": 0.2411,
+      "step": 8401
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5086360547654088e-06,
+      "loss": 0.2249,
+      "step": 8402
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5066536916792216e-06,
+      "loss": 0.2035,
+      "step": 8403
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.504671999910434e-06,
+      "loss": 0.2149,
+      "step": 8404
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502690979636584e-06,
+      "loss": 0.2261,
+      "step": 8405
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5007106310351427e-06,
+      "loss": 0.1844,
+      "step": 8406
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498730954283527e-06,
+      "loss": 0.2254,
+      "step": 8407
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496751949559095e-06,
+      "loss": 0.2176,
+      "step": 8408
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4947736170391466e-06,
+      "loss": 0.201,
+      "step": 8409
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.492795956900906e-06,
+      "loss": 0.2039,
+      "step": 8410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.49081896932156e-06,
+      "loss": 0.2137,
+      "step": 8411
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4888426544782197e-06,
+      "loss": 0.2215,
+      "step": 8412
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4868670125479397e-06,
+      "loss": 0.1959,
+      "step": 8413
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4848920437077195e-06,
+      "loss": 0.1877,
+      "step": 8414
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4829177481344855e-06,
+      "loss": 0.2016,
+      "step": 8415
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.480944126005116e-06,
+      "loss": 0.2147,
+      "step": 8416
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4789711774964243e-06,
+      "loss": 0.2076,
+      "step": 8417
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.476998902785165e-06,
+      "loss": 0.2371,
+      "step": 8418
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4750273020480286e-06,
+      "loss": 0.1948,
+      "step": 8419
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.473056375461651e-06,
+      "loss": 0.2458,
+      "step": 8420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.471086123202602e-06,
+      "loss": 0.1967,
+      "step": 8421
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4691165454473943e-06,
+      "loss": 0.2354,
+      "step": 8422
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4671476423724826e-06,
+      "loss": 0.2187,
+      "step": 8423
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4651794141542517e-06,
+      "loss": 0.2451,
+      "step": 8424
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4632118609690347e-06,
+      "loss": 0.2073,
+      "step": 8425
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461244982993103e-06,
+      "loss": 0.2563,
+      "step": 8426
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.459278780402665e-06,
+      "loss": 0.1937,
+      "step": 8427
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4573132533738696e-06,
+      "loss": 0.2178,
+      "step": 8428
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.455348402082807e-06,
+      "loss": 0.2286,
+      "step": 8429
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.453384226705503e-06,
+      "loss": 0.1999,
+      "step": 8430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4514207274179268e-06,
+      "loss": 0.1923,
+      "step": 8431
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4494579043959864e-06,
+      "loss": 0.1954,
+      "step": 8432
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4474957578155246e-06,
+      "loss": 0.2056,
+      "step": 8433
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.445534287852328e-06,
+      "loss": 0.1974,
+      "step": 8434
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.44357349468212e-06,
+      "loss": 0.2322,
+      "step": 8435
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4416133784805706e-06,
+      "loss": 0.2036,
+      "step": 8436
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439653939423283e-06,
+      "loss": 0.1984,
+      "step": 8437
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.437695177685795e-06,
+      "loss": 0.1792,
+      "step": 8438
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.435737093443592e-06,
+      "loss": 0.2405,
+      "step": 8439
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4337796868720963e-06,
+      "loss": 0.2148,
+      "step": 8440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4318229581466703e-06,
+      "loss": 0.1924,
+      "step": 8441
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4298669074426096e-06,
+      "loss": 0.2282,
+      "step": 8442
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4279115349351546e-06,
+      "loss": 0.2138,
+      "step": 8443
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4259568407994884e-06,
+      "loss": 0.1986,
+      "step": 8444
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424002825210726e-06,
+      "loss": 0.2069,
+      "step": 8445
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.422049488343928e-06,
+      "loss": 0.1803,
+      "step": 8446
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4200968303740867e-06,
+      "loss": 0.1951,
+      "step": 8447
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4181448514761383e-06,
+      "loss": 0.2522,
+      "step": 8448
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4161935518249604e-06,
+      "loss": 0.2062,
+      "step": 8449
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.414242931595363e-06,
+      "loss": 0.1943,
+      "step": 8450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.412292990962102e-06,
+      "loss": 0.219,
+      "step": 8451
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.410343730099869e-06,
+      "loss": 0.2303,
+      "step": 8452
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4083951491832947e-06,
+      "loss": 0.2173,
+      "step": 8453
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4064472483869507e-06,
+      "loss": 0.2069,
+      "step": 8454
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4045000278853483e-06,
+      "loss": 0.216,
+      "step": 8455
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.40255348785293e-06,
+      "loss": 0.2043,
+      "step": 8456
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.400607628464087e-06,
+      "loss": 0.1912,
+      "step": 8457
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3986624498931455e-06,
+      "loss": 0.2002,
+      "step": 8458
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.39671795231437e-06,
+      "loss": 0.1894,
+      "step": 8459
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3947741359019683e-06,
+      "loss": 0.202,
+      "step": 8460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3928310008300804e-06,
+      "loss": 0.224,
+      "step": 8461
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3908885472727907e-06,
+      "loss": 0.2285,
+      "step": 8462
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3889467754041195e-06,
+      "loss": 0.1925,
+      "step": 8463
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3870056853980317e-06,
+      "loss": 0.1852,
+      "step": 8464
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.385065277428419e-06,
+      "loss": 0.1892,
+      "step": 8465
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.383125551669123e-06,
+      "loss": 0.2106,
+      "step": 8466
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3811865082939225e-06,
+      "loss": 0.2155,
+      "step": 8467
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379248147476527e-06,
+      "loss": 0.1981,
+      "step": 8468
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3773104693906036e-06,
+      "loss": 0.2191,
+      "step": 8469
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3753734742097345e-06,
+      "loss": 0.2072,
+      "step": 8470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.373437162107457e-06,
+      "loss": 0.2086,
+      "step": 8471
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371501533257241e-06,
+      "loss": 0.2148,
+      "step": 8472
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3695665878324993e-06,
+      "loss": 0.2194,
+      "step": 8473
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.367632326006576e-06,
+      "loss": 0.2147,
+      "step": 8474
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365698747952758e-06,
+      "loss": 0.2006,
+      "step": 8475
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.363765853844278e-06,
+      "loss": 0.2067,
+      "step": 8476
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3618336438542977e-06,
+      "loss": 0.2256,
+      "step": 8477
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.359902118155922e-06,
+      "loss": 0.2212,
+      "step": 8478
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3579712769221907e-06,
+      "loss": 0.2012,
+      "step": 8479
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3560411203260857e-06,
+      "loss": 0.2031,
+      "step": 8480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.354111648540528e-06,
+      "loss": 0.2066,
+      "step": 8481
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.352182861738375e-06,
+      "loss": 0.2014,
+      "step": 8482
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.350254760092424e-06,
+      "loss": 0.2302,
+      "step": 8483
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3483273437754106e-06,
+      "loss": 0.2268,
+      "step": 8484
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346400612960009e-06,
+      "loss": 0.2291,
+      "step": 8485
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.344474567818833e-06,
+      "loss": 0.2264,
+      "step": 8486
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342549208524436e-06,
+      "loss": 0.2432,
+      "step": 8487
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340624535249302e-06,
+      "loss": 0.2262,
+      "step": 8488
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.338700548165863e-06,
+      "loss": 0.2341,
+      "step": 8489
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3367772474464867e-06,
+      "loss": 0.2407,
+      "step": 8490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3348546332634767e-06,
+      "loss": 0.2019,
+      "step": 8491
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.332932705789078e-06,
+      "loss": 0.2318,
+      "step": 8492
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331011465195475e-06,
+      "loss": 0.2186,
+      "step": 8493
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3290909116547857e-06,
+      "loss": 0.2184,
+      "step": 8494
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3271710453390718e-06,
+      "loss": 0.2243,
+      "step": 8495
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3252518664203317e-06,
+      "loss": 0.2482,
+      "step": 8496
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.323333375070499e-06,
+      "loss": 0.2194,
+      "step": 8497
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3214155714614485e-06,
+      "loss": 0.2081,
+      "step": 8498
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.319498455764995e-06,
+      "loss": 0.186,
+      "step": 8499
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3175820281528883e-06,
+      "loss": 0.1897,
+      "step": 8500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3156662887968206e-06,
+      "loss": 0.1995,
+      "step": 8501
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.313751237868418e-06,
+      "loss": 0.1971,
+      "step": 8502
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3118368755392484e-06,
+      "loss": 0.2303,
+      "step": 8503
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.309923201980815e-06,
+      "loss": 0.1955,
+      "step": 8504
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308010217364566e-06,
+      "loss": 0.2243,
+      "step": 8505
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306097921861874e-06,
+      "loss": 0.191,
+      "step": 8506
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3041863156440613e-06,
+      "loss": 0.1997,
+      "step": 8507
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3022753988823897e-06,
+      "loss": 0.2044,
+      "step": 8508
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3003651717480546e-06,
+      "loss": 0.2338,
+      "step": 8509
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.29845563441219e-06,
+      "loss": 0.2163,
+      "step": 8510
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296546787045866e-06,
+      "loss": 0.2159,
+      "step": 8511
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294638629820094e-06,
+      "loss": 0.2064,
+      "step": 8512
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2927311629058236e-06,
+      "loss": 0.2247,
+      "step": 8513
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2908243864739445e-06,
+      "loss": 0.2196,
+      "step": 8514
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2889183006952744e-06,
+      "loss": 0.1923,
+      "step": 8515
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2870129057405833e-06,
+      "loss": 0.2001,
+      "step": 8516
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2851082017805704e-06,
+      "loss": 0.2235,
+      "step": 8517
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283204188985876e-06,
+      "loss": 0.2032,
+      "step": 8518
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2813008675270797e-06,
+      "loss": 0.2152,
+      "step": 8519
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2793982375746915e-06,
+      "loss": 0.1945,
+      "step": 8520
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.277496299299169e-06,
+      "loss": 0.2283,
+      "step": 8521
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2755950528709025e-06,
+      "loss": 0.23,
+      "step": 8522
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2736944984602227e-06,
+      "loss": 0.2362,
+      "step": 8523
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2717946362373967e-06,
+      "loss": 0.2131,
+      "step": 8524
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.2249,
+      "step": 8525
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2679969890360676e-06,
+      "loss": 0.2119,
+      "step": 8526
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2660992043977915e-06,
+      "loss": 0.2282,
+      "step": 8527
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264202112627821e-06,
+      "loss": 0.1808,
+      "step": 8528
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2623057138961115e-06,
+      "loss": 0.1947,
+      "step": 8529
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26041000837256e-06,
+      "loss": 0.1935,
+      "step": 8530
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.258514996226999e-06,
+      "loss": 0.203,
+      "step": 8531
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.256620677629201e-06,
+      "loss": 0.1933,
+      "step": 8532
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.254727052748874e-06,
+      "loss": 0.2262,
+      "step": 8533
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.252834121755666e-06,
+      "loss": 0.2458,
+      "step": 8534
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25094188481916e-06,
+      "loss": 0.2081,
+      "step": 8535
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2490503421088817e-06,
+      "loss": 0.229,
+      "step": 8536
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247159493794291e-06,
+      "loss": 0.2018,
+      "step": 8537
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2452693400447823e-06,
+      "loss": 0.1849,
+      "step": 8538
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2433798810296946e-06,
+      "loss": 0.2125,
+      "step": 8539
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2414911169182985e-06,
+      "loss": 0.217,
+      "step": 8540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2396030478798125e-06,
+      "loss": 0.2253,
+      "step": 8541
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2377156740833795e-06,
+      "loss": 0.1993,
+      "step": 8542
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2358289956980882e-06,
+      "loss": 0.2047,
+      "step": 8543
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.233943012892963e-06,
+      "loss": 0.2254,
+      "step": 8544
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.232057725836967e-06,
+      "loss": 0.2306,
+      "step": 8545
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2301731346990017e-06,
+      "loss": 0.2115,
+      "step": 8546
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2282892396478984e-06,
+      "loss": 0.2241,
+      "step": 8547
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226406040852438e-06,
+      "loss": 0.2095,
+      "step": 8548
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2245235384813333e-06,
+      "loss": 0.223,
+      "step": 8549
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2226417327032333e-06,
+      "loss": 0.2227,
+      "step": 8550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2207606236867285e-06,
+      "loss": 0.2125,
+      "step": 8551
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2188802116003405e-06,
+      "loss": 0.2284,
+      "step": 8552
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2170004966125334e-06,
+      "loss": 0.2221,
+      "step": 8553
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2151214788917097e-06,
+      "loss": 0.2276,
+      "step": 8554
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2132431586062074e-06,
+      "loss": 0.2054,
+      "step": 8555
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2113655359243016e-06,
+      "loss": 0.2256,
+      "step": 8556
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2094886110142065e-06,
+      "loss": 0.2123,
+      "step": 8557
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2076123840440723e-06,
+      "loss": 0.2238,
+      "step": 8558
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.205736855181987e-06,
+      "loss": 0.2164,
+      "step": 8559
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2038620245959797e-06,
+      "loss": 0.193,
+      "step": 8560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2019878924540084e-06,
+      "loss": 0.2085,
+      "step": 8561
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2001144589239754e-06,
+      "loss": 0.1918,
+      "step": 8562
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.19824172417372e-06,
+      "loss": 0.214,
+      "step": 8563
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.196369688371016e-06,
+      "loss": 0.217,
+      "step": 8564
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1944983516835784e-06,
+      "loss": 0.2315,
+      "step": 8565
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1926277142790554e-06,
+      "loss": 0.2086,
+      "step": 8566
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1907577763250353e-06,
+      "loss": 0.2356,
+      "step": 8567
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1888885379890433e-06,
+      "loss": 0.2027,
+      "step": 8568
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1870199994385443e-06,
+      "loss": 0.2208,
+      "step": 8569
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.185152160840932e-06,
+      "loss": 0.203,
+      "step": 8570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.183285022363546e-06,
+      "loss": 0.228,
+      "step": 8571
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1814185841736566e-06,
+      "loss": 0.1719,
+      "step": 8572
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179552846438486e-06,
+      "loss": 0.2038,
+      "step": 8573
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1776878093251718e-06,
+      "loss": 0.2253,
+      "step": 8574
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.175823473000804e-06,
+      "loss": 0.2328,
+      "step": 8575
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1739598376324056e-06,
+      "loss": 0.2168,
+      "step": 8576
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1720969033869367e-06,
+      "loss": 0.2199,
+      "step": 8577
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.170234670431296e-06,
+      "loss": 0.2122,
+      "step": 8578
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.168373138932315e-06,
+      "loss": 0.2062,
+      "step": 8579
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1665123090567643e-06,
+      "loss": 0.205,
+      "step": 8580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1646521809713583e-06,
+      "loss": 0.1924,
+      "step": 8581
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.162792754842743e-06,
+      "loss": 0.2191,
+      "step": 8582
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160934030837496e-06,
+      "loss": 0.2149,
+      "step": 8583
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1590760091221408e-06,
+      "loss": 0.2027,
+      "step": 8584
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1572186898631344e-06,
+      "loss": 0.197,
+      "step": 8585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.15536207322687e-06,
+      "loss": 0.2223,
+      "step": 8586
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1535061593796815e-06,
+      "loss": 0.2156,
+      "step": 8587
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1516509484878347e-06,
+      "loss": 0.2149,
+      "step": 8588
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1497964407175376e-06,
+      "loss": 0.2043,
+      "step": 8589
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.147942636234931e-06,
+      "loss": 0.208,
+      "step": 8590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.146089535206095e-06,
+      "loss": 0.2269,
+      "step": 8591
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1442371377970483e-06,
+      "loss": 0.2283,
+      "step": 8592
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.142385444173739e-06,
+      "loss": 0.2242,
+      "step": 8593
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1405344545020613e-06,
+      "loss": 0.2142,
+      "step": 8594
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1386841689478423e-06,
+      "loss": 0.2225,
+      "step": 8595
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1368345876768448e-06,
+      "loss": 0.2144,
+      "step": 8596
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.13498571085477e-06,
+      "loss": 0.1784,
+      "step": 8597
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1331375386472585e-06,
+      "loss": 0.2277,
+      "step": 8598
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1312900712198825e-06,
+      "loss": 0.2096,
+      "step": 8599
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1294433087381548e-06,
+      "loss": 0.2134,
+      "step": 8600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.127597251367527e-06,
+      "loss": 0.1898,
+      "step": 8601
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1257518992733783e-06,
+      "loss": 0.1955,
+      "step": 8602
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1239072526210346e-06,
+      "loss": 0.2207,
+      "step": 8603
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1220633115757537e-06,
+      "loss": 0.2021,
+      "step": 8604
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.120220076302734e-06,
+      "loss": 0.2118,
+      "step": 8605
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1183775469671054e-06,
+      "loss": 0.2259,
+      "step": 8606
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116535723733938e-06,
+      "loss": 0.2318,
+      "step": 8607
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1146946067682385e-06,
+      "loss": 0.2132,
+      "step": 8608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.112854196234949e-06,
+      "loss": 0.2296,
+      "step": 8609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.111014492298953e-06,
+      "loss": 0.2238,
+      "step": 8610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10917549512506e-06,
+      "loss": 0.2188,
+      "step": 8611
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1073372048780238e-06,
+      "loss": 0.2149,
+      "step": 8612
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.105499621722539e-06,
+      "loss": 0.236,
+      "step": 8613
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.103662745823232e-06,
+      "loss": 0.2042,
+      "step": 8614
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10182657734466e-06,
+      "loss": 0.2052,
+      "step": 8615
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.099991116451325e-06,
+      "loss": 0.2111,
+      "step": 8616
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.098156363307665e-06,
+      "loss": 0.2253,
+      "step": 8617
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0963223180780503e-06,
+      "loss": 0.2185,
+      "step": 8618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.094488980926792e-06,
+      "loss": 0.2042,
+      "step": 8619
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092656352018134e-06,
+      "loss": 0.218,
+      "step": 8620
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090824431516261e-06,
+      "loss": 0.2206,
+      "step": 8621
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0889932195852903e-06,
+      "loss": 0.232,
+      "step": 8622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0871627163892806e-06,
+      "loss": 0.2213,
+      "step": 8623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085332922092218e-06,
+      "loss": 0.1997,
+      "step": 8624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.083503836858034e-06,
+      "loss": 0.1968,
+      "step": 8625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.081675460850594e-06,
+      "loss": 0.2312,
+      "step": 8626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0798477942336993e-06,
+      "loss": 0.2002,
+      "step": 8627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0780208371710875e-06,
+      "loss": 0.2292,
+      "step": 8628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0761945898264334e-06,
+      "loss": 0.2206,
+      "step": 8629
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0743690523633466e-06,
+      "loss": 0.2075,
+      "step": 8630
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0725442249453743e-06,
+      "loss": 0.217,
+      "step": 8631
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0707201077360006e-06,
+      "loss": 0.199,
+      "step": 8632
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0688967008986495e-06,
+      "loss": 0.2156,
+      "step": 8633
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0670740045966684e-06,
+      "loss": 0.2184,
+      "step": 8634
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0652520189933547e-06,
+      "loss": 0.2332,
+      "step": 8635
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0634307442519374e-06,
+      "loss": 0.2256,
+      "step": 8636
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0616101805355814e-06,
+      "loss": 0.2418,
+      "step": 8637
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0597903280073885e-06,
+      "loss": 0.2261,
+      "step": 8638
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0579711868303954e-06,
+      "loss": 0.2331,
+      "step": 8639
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0561527571675778e-06,
+      "loss": 0.213,
+      "step": 8640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.054335039181844e-06,
+      "loss": 0.187,
+      "step": 8641
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.052518033036045e-06,
+      "loss": 0.2025,
+      "step": 8642
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0507017388929572e-06,
+      "loss": 0.2043,
+      "step": 8643
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0488861569153006e-06,
+      "loss": 0.2199,
+      "step": 8644
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047071287265735e-06,
+      "loss": 0.2165,
+      "step": 8645
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0452571301068514e-06,
+      "loss": 0.1843,
+      "step": 8646
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043443685601173e-06,
+      "loss": 0.2112,
+      "step": 8647
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0416309539111656e-06,
+      "loss": 0.2174,
+      "step": 8648
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039818935199229e-06,
+      "loss": 0.2303,
+      "step": 8649
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0380076296276995e-06,
+      "loss": 0.1981,
+      "step": 8650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0361970373588525e-06,
+      "loss": 0.2211,
+      "step": 8651
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.034387158554887e-06,
+      "loss": 0.2283,
+      "step": 8652
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0325779933779545e-06,
+      "loss": 0.2092,
+      "step": 8653
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0307695419901354e-06,
+      "loss": 0.2021,
+      "step": 8654
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0289618045534466e-06,
+      "loss": 0.1964,
+      "step": 8655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0271547812298352e-06,
+      "loss": 0.2139,
+      "step": 8656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0253484721811935e-06,
+      "loss": 0.2059,
+      "step": 8657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0235428775693457e-06,
+      "loss": 0.2258,
+      "step": 8658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0217379975560514e-06,
+      "loss": 0.2192,
+      "step": 8659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0199338323030072e-06,
+      "loss": 0.2103,
+      "step": 8660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0181303819718456e-06,
+      "loss": 0.2194,
+      "step": 8661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0163276467241367e-06,
+      "loss": 0.2055,
+      "step": 8662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.014525626721382e-06,
+      "loss": 0.1996,
+      "step": 8663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012724322125026e-06,
+      "loss": 0.1932,
+      "step": 8664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0109237330964395e-06,
+      "loss": 0.2254,
+      "step": 8665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0091238597969354e-06,
+      "loss": 0.2237,
+      "step": 8666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.007324702387764e-06,
+      "loss": 0.2441,
+      "step": 8667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0055262610301086e-06,
+      "loss": 0.2306,
+      "step": 8668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0037285358850866e-06,
+      "loss": 0.2067,
+      "step": 8669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.001931527113756e-06,
+      "loss": 0.2071,
+      "step": 8670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0001352348771075e-06,
+      "loss": 0.2158,
+      "step": 8671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.998339659336067e-06,
+      "loss": 0.2028,
+      "step": 8672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9965448006514985e-06,
+      "loss": 0.2069,
+      "step": 8673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9947506589842035e-06,
+      "loss": 0.2165,
+      "step": 8674
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.99295723449491e-06,
+      "loss": 0.2127,
+      "step": 8675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9911645273442916e-06,
+      "loss": 0.1993,
+      "step": 8676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9893725376929506e-06,
+      "loss": 0.2145,
+      "step": 8677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.987581265701438e-06,
+      "loss": 0.2083,
+      "step": 8678
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985790711530221e-06,
+      "loss": 0.2362,
+      "step": 8679
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9840008753397177e-06,
+      "loss": 0.2166,
+      "step": 8680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9822117572902765e-06,
+      "loss": 0.2493,
+      "step": 8681
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9804233575421795e-06,
+      "loss": 0.2224,
+      "step": 8682
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9786356762556513e-06,
+      "loss": 0.1998,
+      "step": 8683
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.976848713590839e-06,
+      "loss": 0.2206,
+      "step": 8684
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9750624697078426e-06,
+      "loss": 0.2134,
+      "step": 8685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9732769447666843e-06,
+      "loss": 0.2077,
+      "step": 8686
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.971492138927331e-06,
+      "loss": 0.2074,
+      "step": 8687
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9697080523496746e-06,
+      "loss": 0.227,
+      "step": 8688
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967924685193552e-06,
+      "loss": 0.1854,
+      "step": 8689
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9661420376187313e-06,
+      "loss": 0.2218,
+      "step": 8690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9643601097849186e-06,
+      "loss": 0.2165,
+      "step": 8691
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9625789018517527e-06,
+      "loss": 0.1995,
+      "step": 8692
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9607984139788105e-06,
+      "loss": 0.2177,
+      "step": 8693
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.959018646325601e-06,
+      "loss": 0.1895,
+      "step": 8694
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9572395990515736e-06,
+      "loss": 0.2056,
+      "step": 8695
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9554612723161113e-06,
+      "loss": 0.2078,
+      "step": 8696
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9536836662785276e-06,
+      "loss": 0.2232,
+      "step": 8697
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9519067810980773e-06,
+      "loss": 0.2272,
+      "step": 8698
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9501306169339486e-06,
+      "loss": 0.2336,
+      "step": 8699
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9483551739452657e-06,
+      "loss": 0.2041,
+      "step": 8700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.946580452291087e-06,
+      "loss": 0.2173,
+      "step": 8701
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944806452130409e-06,
+      "loss": 0.2242,
+      "step": 8702
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9430331736221608e-06,
+      "loss": 0.2194,
+      "step": 8703
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9412606169252067e-06,
+      "loss": 0.2107,
+      "step": 8704
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.939488782198351e-06,
+      "loss": 0.1878,
+      "step": 8705
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.937717669600323e-06,
+      "loss": 0.2283,
+      "step": 8706
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9359472792897992e-06,
+      "loss": 0.2034,
+      "step": 8707
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.934177611425384e-06,
+      "loss": 0.2257,
+      "step": 8708
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.932408666165617e-06,
+      "loss": 0.2088,
+      "step": 8709
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.930640443668984e-06,
+      "loss": 0.2026,
+      "step": 8710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.928872944093888e-06,
+      "loss": 0.2445,
+      "step": 8711
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9271061675986814e-06,
+      "loss": 0.2094,
+      "step": 8712
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.925340114341644e-06,
+      "loss": 0.2165,
+      "step": 8713
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923574784480997e-06,
+      "loss": 0.2142,
+      "step": 8714
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921810178174893e-06,
+      "loss": 0.2259,
+      "step": 8715
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.920046295581418e-06,
+      "loss": 0.1963,
+      "step": 8716
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918283136858595e-06,
+      "loss": 0.2282,
+      "step": 8717
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916520702164386e-06,
+      "loss": 0.2083,
+      "step": 8718
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9147589916566867e-06,
+      "loss": 0.213,
+      "step": 8719
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9129980054933196e-06,
+      "loss": 0.2231,
+      "step": 8720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.911237743832053e-06,
+      "loss": 0.2285,
+      "step": 8721
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.909478206830584e-06,
+      "loss": 0.2202,
+      "step": 8722
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9077193946465477e-06,
+      "loss": 0.2198,
+      "step": 8723
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.905961307437515e-06,
+      "loss": 0.2127,
+      "step": 8724
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9042039453609874e-06,
+      "loss": 0.2084,
+      "step": 8725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9024473085744055e-06,
+      "loss": 0.2312,
+      "step": 8726
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.900691397235145e-06,
+      "loss": 0.2202,
+      "step": 8727
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.898936211500515e-06,
+      "loss": 0.227,
+      "step": 8728
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.897181751527758e-06,
+      "loss": 0.2444,
+      "step": 8729
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8954280174740536e-06,
+      "loss": 0.1978,
+      "step": 8730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8936750094965172e-06,
+      "loss": 0.1922,
+      "step": 8731
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8919227277521968e-06,
+      "loss": 0.2051,
+      "step": 8732
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8901711723980788e-06,
+      "loss": 0.2245,
+      "step": 8733
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8884203435910808e-06,
+      "loss": 0.2373,
+      "step": 8734
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8866702414880579e-06,
+      "loss": 0.2247,
+      "step": 8735
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8849208662457974e-06,
+      "loss": 0.1912,
+      "step": 8736
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8831722180210276e-06,
+      "loss": 0.2201,
+      "step": 8737
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8814242969704011e-06,
+      "loss": 0.2065,
+      "step": 8738
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8796771032505134e-06,
+      "loss": 0.1812,
+      "step": 8739
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8779306370178941e-06,
+      "loss": 0.2217,
+      "step": 8740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8761848984290065e-06,
+      "loss": 0.1911,
+      "step": 8741
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.874439887640248e-06,
+      "loss": 0.2068,
+      "step": 8742
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8726956048079515e-06,
+      "loss": 0.2077,
+      "step": 8743
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8709520500883837e-06,
+      "loss": 0.2135,
+      "step": 8744
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.869209223637749e-06,
+      "loss": 0.1913,
+      "step": 8745
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8674671256121856e-06,
+      "loss": 0.2152,
+      "step": 8746
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.865725756167761e-06,
+      "loss": 0.2238,
+      "step": 8747
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.863985115460485e-06,
+      "loss": 0.1914,
+      "step": 8748
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8622452036462935e-06,
+      "loss": 0.2134,
+      "step": 8749
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.860506020881071e-06,
+      "loss": 0.2166,
+      "step": 8750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8587675673206273e-06,
+      "loss": 0.2234,
+      "step": 8751
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8570298431207013e-06,
+      "loss": 0.2107,
+      "step": 8752
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.855292848436977e-06,
+      "loss": 0.2032,
+      "step": 8753
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.853556583425068e-06,
+      "loss": 0.2142,
+      "step": 8754
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8518210482405253e-06,
+      "loss": 0.2188,
+      "step": 8755
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.850086243038831e-06,
+      "loss": 0.2083,
+      "step": 8756
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8483521679754046e-06,
+      "loss": 0.1985,
+      "step": 8757
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8466188232055993e-06,
+      "loss": 0.235,
+      "step": 8758
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8448862088847018e-06,
+      "loss": 0.2108,
+      "step": 8759
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8431543251679385e-06,
+      "loss": 0.2122,
+      "step": 8760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8414231722104602e-06,
+      "loss": 0.2278,
+      "step": 8761
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.83969275016736e-06,
+      "loss": 0.2176,
+      "step": 8762
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8379630591936647e-06,
+      "loss": 0.2265,
+      "step": 8763
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8362340994443361e-06,
+      "loss": 0.195,
+      "step": 8764
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8345058710742657e-06,
+      "loss": 0.1927,
+      "step": 8765
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8327783742382854e-06,
+      "loss": 0.2128,
+      "step": 8766
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.831051609091158e-06,
+      "loss": 0.2032,
+      "step": 8767
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8293255757875828e-06,
+      "loss": 0.2172,
+      "step": 8768
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.827600274482193e-06,
+      "loss": 0.2153,
+      "step": 8769
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8258757053295528e-06,
+      "loss": 0.2169,
+      "step": 8770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8241518684841642e-06,
+      "loss": 0.1699,
+      "step": 8771
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8224287641004645e-06,
+      "loss": 0.2201,
+      "step": 8772
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.820706392332824e-06,
+      "loss": 0.2002,
+      "step": 8773
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8189847533355476e-06,
+      "loss": 0.2039,
+      "step": 8774
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8172638472628722e-06,
+      "loss": 0.2073,
+      "step": 8775
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8155436742689737e-06,
+      "loss": 0.2201,
+      "step": 8776
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8138242345079582e-06,
+      "loss": 0.203,
+      "step": 8777
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8121055281338707e-06,
+      "loss": 0.2043,
+      "step": 8778
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8103875553006822e-06,
+      "loss": 0.2002,
+      "step": 8779
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8086703161623053e-06,
+      "loss": 0.2242,
+      "step": 8780
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8069538108725837e-06,
+      "loss": 0.2021,
+      "step": 8781
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.805238039585301e-06,
+      "loss": 0.2264,
+      "step": 8782
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8035230024541706e-06,
+      "loss": 0.2279,
+      "step": 8783
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8018086996328333e-06,
+      "loss": 0.2174,
+      "step": 8784
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.800095131274876e-06,
+      "loss": 0.2079,
+      "step": 8785
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7983822975338127e-06,
+      "loss": 0.1987,
+      "step": 8786
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7966701985630975e-06,
+      "loss": 0.1933,
+      "step": 8787
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7949588345161083e-06,
+      "loss": 0.2259,
+      "step": 8788
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7932482055461642e-06,
+      "loss": 0.2089,
+      "step": 8789
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7915383118065233e-06,
+      "loss": 0.1978,
+      "step": 8790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7898291534503687e-06,
+      "loss": 0.1739,
+      "step": 8791
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7881207306308258e-06,
+      "loss": 0.2054,
+      "step": 8792
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.786413043500942e-06,
+      "loss": 0.2083,
+      "step": 8793
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7847060922137106e-06,
+      "loss": 0.1944,
+      "step": 8794
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7829998769220548e-06,
+      "loss": 0.2122,
+      "step": 8795
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.781294397778831e-06,
+      "loss": 0.2126,
+      "step": 8796
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7795896549368308e-06,
+      "loss": 0.2571,
+      "step": 8797
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7778856485487782e-06,
+      "loss": 0.2458,
+      "step": 8798
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.776182378767336e-06,
+      "loss": 0.2023,
+      "step": 8799
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.774479845745094e-06,
+      "loss": 0.1838,
+      "step": 8800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7727780496345836e-06,
+      "loss": 0.2166,
+      "step": 8801
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7710769905882608e-06,
+      "loss": 0.2084,
+      "step": 8802
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7693766687585235e-06,
+      "loss": 0.2208,
+      "step": 8803
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7676770842977009e-06,
+      "loss": 0.1944,
+      "step": 8804
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.765978237358056e-06,
+      "loss": 0.2064,
+      "step": 8805
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7642801280917855e-06,
+      "loss": 0.2356,
+      "step": 8806
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7625827566510213e-06,
+      "loss": 0.2255,
+      "step": 8807
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7608861231878282e-06,
+      "loss": 0.2301,
+      "step": 8808
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.759190227854205e-06,
+      "loss": 0.1937,
+      "step": 8809
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7574950708020854e-06,
+      "loss": 0.2209,
+      "step": 8810
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7558006521833327e-06,
+      "loss": 0.2224,
+      "step": 8811
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7541069721497494e-06,
+      "loss": 0.2259,
+      "step": 8812
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7524140308530691e-06,
+      "loss": 0.2242,
+      "step": 8813
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7507218284449568e-06,
+      "loss": 0.1885,
+      "step": 8814
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7490303650770235e-06,
+      "loss": 0.212,
+      "step": 8815
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.747339640900797e-06,
+      "loss": 0.2065,
+      "step": 8816
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7456496560677482e-06,
+      "loss": 0.2189,
+      "step": 8817
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7439604107292807e-06,
+      "loss": 0.2228,
+      "step": 8818
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7422719050367331e-06,
+      "loss": 0.247,
+      "step": 8819
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7405841391413724e-06,
+      "loss": 0.2017,
+      "step": 8820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7388971131944032e-06,
+      "loss": 0.2106,
+      "step": 8821
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7372108273469668e-06,
+      "loss": 0.2029,
+      "step": 8822
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7355252817501333e-06,
+      "loss": 0.2127,
+      "step": 8823
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.733840476554911e-06,
+      "loss": 0.2088,
+      "step": 8824
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7321564119122336e-06,
+      "loss": 0.263,
+      "step": 8825
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7304730879729769e-06,
+      "loss": 0.2176,
+      "step": 8826
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7287905048879471e-06,
+      "loss": 0.1945,
+      "step": 8827
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7271086628078859e-06,
+      "loss": 0.2124,
+      "step": 8828
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7254275618834604e-06,
+      "loss": 0.1903,
+      "step": 8829
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7237472022652857e-06,
+      "loss": 0.2295,
+      "step": 8830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7220675841038991e-06,
+      "loss": 0.2319,
+      "step": 8831
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7203887075497749e-06,
+      "loss": 0.2135,
+      "step": 8832
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7187105727533248e-06,
+      "loss": 0.1876,
+      "step": 8833
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7170331798648842e-06,
+      "loss": 0.211,
+      "step": 8834
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7153565290347308e-06,
+      "loss": 0.226,
+      "step": 8835
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.713680620413074e-06,
+      "loss": 0.2246,
+      "step": 8836
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7120054541500552e-06,
+      "loss": 0.2186,
+      "step": 8837
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.710331030395751e-06,
+      "loss": 0.1967,
+      "step": 8838
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.708657349300168e-06,
+      "loss": 0.2272,
+      "step": 8839
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7069844110132516e-06,
+      "loss": 0.21,
+      "step": 8840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7053122156848756e-06,
+      "loss": 0.2545,
+      "step": 8841
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7036407634648532e-06,
+      "loss": 0.1997,
+      "step": 8842
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7019700545029215e-06,
+      "loss": 0.2174,
+      "step": 8843
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7003000889487599e-06,
+      "loss": 0.21,
+      "step": 8844
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6986308669519768e-06,
+      "loss": 0.1904,
+      "step": 8845
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6969623886621146e-06,
+      "loss": 0.2181,
+      "step": 8846
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6952946542286563e-06,
+      "loss": 0.2063,
+      "step": 8847
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6936276638010042e-06,
+      "loss": 0.2067,
+      "step": 8848
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6919614175285036e-06,
+      "loss": 0.2049,
+      "step": 8849
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6902959155604316e-06,
+      "loss": 0.2061,
+      "step": 8850
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6886311580459991e-06,
+      "loss": 0.2096,
+      "step": 8851
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.686967145134346e-06,
+      "loss": 0.1957,
+      "step": 8852
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.2196,
+      "step": 8853
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6836413537156182e-06,
+      "loss": 0.2066,
+      "step": 8854
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6819795755064993e-06,
+      "loss": 0.197,
+      "step": 8855
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6803185424960688e-06,
+      "loss": 0.2236,
+      "step": 8856
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.678658254833131e-06,
+      "loss": 0.2126,
+      "step": 8857
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6769987126664322e-06,
+      "loss": 0.2139,
+      "step": 8858
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.675339916144647e-06,
+      "loss": 0.2215,
+      "step": 8859
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6736818654163867e-06,
+      "loss": 0.2003,
+      "step": 8860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6720245606301876e-06,
+      "loss": 0.2314,
+      "step": 8861
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6703680019345304e-06,
+      "loss": 0.194,
+      "step": 8862
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6687121894778236e-06,
+      "loss": 0.2302,
+      "step": 8863
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6670571234084066e-06,
+      "loss": 0.2192,
+      "step": 8864
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6654028038745584e-06,
+      "loss": 0.2091,
+      "step": 8865
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6637492310244818e-06,
+      "loss": 0.1965,
+      "step": 8866
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6620964050063192e-06,
+      "loss": 0.2219,
+      "step": 8867
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6604443259681457e-06,
+      "loss": 0.2013,
+      "step": 8868
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6587929940579683e-06,
+      "loss": 0.2188,
+      "step": 8869
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6571424094237265e-06,
+      "loss": 0.1912,
+      "step": 8870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6554925722132952e-06,
+      "loss": 0.2021,
+      "step": 8871
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.65384348257448e-06,
+      "loss": 0.2105,
+      "step": 8872
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6521951406550207e-06,
+      "loss": 0.2073,
+      "step": 8873
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6505475466025911e-06,
+      "loss": 0.217,
+      "step": 8874
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6489007005647929e-06,
+      "loss": 0.2229,
+      "step": 8875
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6472546026891656e-06,
+      "loss": 0.22,
+      "step": 8876
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6456092531231816e-06,
+      "loss": 0.2015,
+      "step": 8877
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6439646520142428e-06,
+      "loss": 0.1987,
+      "step": 8878
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6423207995096946e-06,
+      "loss": 0.2266,
+      "step": 8879
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6406776957567982e-06,
+      "loss": 0.1946,
+      "step": 8880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6390353409027592e-06,
+      "loss": 0.201,
+      "step": 8881
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6373937350947155e-06,
+      "loss": 0.2013,
+      "step": 8882
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6357528784797373e-06,
+      "loss": 0.2252,
+      "step": 8883
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.634112771204821e-06,
+      "loss": 0.2104,
+      "step": 8884
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6324734134169052e-06,
+      "loss": 0.2176,
+      "step": 8885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.630834805262853e-06,
+      "loss": 0.2091,
+      "step": 8886
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.629196946889472e-06,
+      "loss": 0.227,
+      "step": 8887
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6275598384434932e-06,
+      "loss": 0.2161,
+      "step": 8888
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6259234800715796e-06,
+      "loss": 0.209,
+      "step": 8889
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6242878719203314e-06,
+      "loss": 0.1992,
+      "step": 8890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6226530141362806e-06,
+      "loss": 0.2169,
+      "step": 8891
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6210189068658932e-06,
+      "loss": 0.2253,
+      "step": 8892
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.619385550255561e-06,
+      "loss": 0.2213,
+      "step": 8893
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6177529444516193e-06,
+      "loss": 0.233,
+      "step": 8894
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6161210896003287e-06,
+      "loss": 0.2349,
+      "step": 8895
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6144899858478856e-06,
+      "loss": 0.2025,
+      "step": 8896
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.61285963334042e-06,
+      "loss": 0.2172,
+      "step": 8897
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.611230032223987e-06,
+      "loss": 0.1969,
+      "step": 8898
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6096011826445835e-06,
+      "loss": 0.1943,
+      "step": 8899
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6079730847481356e-06,
+      "loss": 0.2022,
+      "step": 8900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6063457386805004e-06,
+      "loss": 0.2397,
+      "step": 8901
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.604719144587472e-06,
+      "loss": 0.2241,
+      "step": 8902
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6030933026147733e-06,
+      "loss": 0.2272,
+      "step": 8903
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6014682129080595e-06,
+      "loss": 0.2232,
+      "step": 8904
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.599843875612922e-06,
+      "loss": 0.1853,
+      "step": 8905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5982202908748846e-06,
+      "loss": 0.2449,
+      "step": 8906
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5965974588393963e-06,
+      "loss": 0.2008,
+      "step": 8907
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5949753796518475e-06,
+      "loss": 0.2287,
+      "step": 8908
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5933540534575566e-06,
+      "loss": 0.2046,
+      "step": 8909
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.591733480401777e-06,
+      "loss": 0.2115,
+      "step": 8910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590113660629692e-06,
+      "loss": 0.2074,
+      "step": 8911
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5884945942864193e-06,
+      "loss": 0.2093,
+      "step": 8912
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.586876281517009e-06,
+      "loss": 0.2014,
+      "step": 8913
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.585258722466444e-06,
+      "loss": 0.2176,
+      "step": 8914
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5836419172796391e-06,
+      "loss": 0.2087,
+      "step": 8915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5820258661014376e-06,
+      "loss": 0.2143,
+      "step": 8916
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5804105690766224e-06,
+      "loss": 0.2366,
+      "step": 8917
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5787960263499025e-06,
+      "loss": 0.2148,
+      "step": 8918
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5771822380659263e-06,
+      "loss": 0.1991,
+      "step": 8919
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5755692043692717e-06,
+      "loss": 0.2039,
+      "step": 8920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.573956925404443e-06,
+      "loss": 0.2125,
+      "step": 8921
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5723454013158834e-06,
+      "loss": 0.2096,
+      "step": 8922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5707346322479667e-06,
+      "loss": 0.2055,
+      "step": 8923
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5691246183450025e-06,
+      "loss": 0.2178,
+      "step": 8924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.567515359751225e-06,
+      "loss": 0.1996,
+      "step": 8925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5659068566108048e-06,
+      "loss": 0.1856,
+      "step": 8926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5642991090678495e-06,
+      "loss": 0.1893,
+      "step": 8927
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5626921172663933e-06,
+      "loss": 0.21,
+      "step": 8928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.561085881350405e-06,
+      "loss": 0.2198,
+      "step": 8929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.559480401463782e-06,
+      "loss": 0.1964,
+      "step": 8930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5578756777503567e-06,
+      "loss": 0.2013,
+      "step": 8931
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.556271710353897e-06,
+      "loss": 0.2186,
+      "step": 8932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5546684994180973e-06,
+      "loss": 0.2153,
+      "step": 8933
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5530660450865875e-06,
+      "loss": 0.215,
+      "step": 8934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.551464347502929e-06,
+      "loss": 0.2148,
+      "step": 8935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5498634068106167e-06,
+      "loss": 0.2261,
+      "step": 8936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5482632231530736e-06,
+      "loss": 0.2115,
+      "step": 8937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546663796673663e-06,
+      "loss": 0.1994,
+      "step": 8938
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5450651275156693e-06,
+      "loss": 0.2175,
+      "step": 8939
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5434672158223162e-06,
+      "loss": 0.2167,
+      "step": 8940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.54187006173676e-06,
+      "loss": 0.204,
+      "step": 8941
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5402736654020857e-06,
+      "loss": 0.1977,
+      "step": 8942
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5386780269613122e-06,
+      "loss": 0.2526,
+      "step": 8943
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.537083146557391e-06,
+      "loss": 0.228,
+      "step": 8944
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5354890243332043e-06,
+      "loss": 0.187,
+      "step": 8945
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.533895660431568e-06,
+      "loss": 0.2228,
+      "step": 8946
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.532303054995231e-06,
+      "loss": 0.2369,
+      "step": 8947
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5307112081668664e-06,
+      "loss": 0.2181,
+      "step": 8948
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5291201200890903e-06,
+      "loss": 0.2403,
+      "step": 8949
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5275297909044428e-06,
+      "loss": 0.2013,
+      "step": 8950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5259402207554018e-06,
+      "loss": 0.2171,
+      "step": 8951
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5243514097843726e-06,
+      "loss": 0.2024,
+      "step": 8952
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5227633581336954e-06,
+      "loss": 0.2109,
+      "step": 8953
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.521176065945641e-06,
+      "loss": 0.1875,
+      "step": 8954
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5195895333624122e-06,
+      "loss": 0.2089,
+      "step": 8955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5180037605261466e-06,
+      "loss": 0.2006,
+      "step": 8956
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.516418747578906e-06,
+      "loss": 0.2237,
+      "step": 8957
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5148344946626903e-06,
+      "loss": 0.1772,
+      "step": 8958
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5132510019194346e-06,
+      "loss": 0.2253,
+      "step": 8959
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511668269490998e-06,
+      "loss": 0.184,
+      "step": 8960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5100862975191788e-06,
+      "loss": 0.2074,
+      "step": 8961
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5085050861456995e-06,
+      "loss": 0.2202,
+      "step": 8962
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.506924635512218e-06,
+      "loss": 0.2079,
+      "step": 8963
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5053449457603263e-06,
+      "loss": 0.2078,
+      "step": 8964
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.503766017031547e-06,
+      "loss": 0.2036,
+      "step": 8965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5021878494673325e-06,
+      "loss": 0.218,
+      "step": 8966
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5006104432090695e-06,
+      "loss": 0.2165,
+      "step": 8967
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.499033798398074e-06,
+      "loss": 0.1983,
+      "step": 8968
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4974579151755963e-06,
+      "loss": 0.2033,
+      "step": 8969
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4958827936828203e-06,
+      "loss": 0.2043,
+      "step": 8970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4943084340608527e-06,
+      "loss": 0.1903,
+      "step": 8971
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.49273483645074e-06,
+      "loss": 0.2025,
+      "step": 8972
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4911620009934602e-06,
+      "loss": 0.215,
+      "step": 8973
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4895899278299198e-06,
+      "loss": 0.1926,
+      "step": 8974
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4880186171009592e-06,
+      "loss": 0.2253,
+      "step": 8975
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.486448068947348e-06,
+      "loss": 0.2086,
+      "step": 8976
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.484878283509792e-06,
+      "loss": 0.1975,
+      "step": 8977
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4833092609289245e-06,
+      "loss": 0.221,
+      "step": 8978
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4817410013453136e-06,
+      "loss": 0.2127,
+      "step": 8979
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4801735048994525e-06,
+      "loss": 0.224,
+      "step": 8980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4786067717317742e-06,
+      "loss": 0.224,
+      "step": 8981
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4770408019826398e-06,
+      "loss": 0.2277,
+      "step": 8982
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4754755957923417e-06,
+      "loss": 0.2031,
+      "step": 8983
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4739111533011042e-06,
+      "loss": 0.2174,
+      "step": 8984
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.472347474649083e-06,
+      "loss": 0.2179,
+      "step": 8985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.470784559976366e-06,
+      "loss": 0.2309,
+      "step": 8986
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4692224094229734e-06,
+      "loss": 0.2062,
+      "step": 8987
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.467661023128857e-06,
+      "loss": 0.2252,
+      "step": 8988
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4661004012338943e-06,
+      "loss": 0.2221,
+      "step": 8989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4645405438778992e-06,
+      "loss": 0.2133,
+      "step": 8990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4629814512006225e-06,
+      "loss": 0.2208,
+      "step": 8991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4614231233417397e-06,
+      "loss": 0.1914,
+      "step": 8992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4598655604408552e-06,
+      "loss": 0.2051,
+      "step": 8993
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4583087626375103e-06,
+      "loss": 0.2117,
+      "step": 8994
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.456752730071176e-06,
+      "loss": 0.2131,
+      "step": 8995
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.455197462881256e-06,
+      "loss": 0.2167,
+      "step": 8996
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4536429612070846e-06,
+      "loss": 0.2102,
+      "step": 8997
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4520892251879225e-06,
+      "loss": 0.2332,
+      "step": 8998
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.450536254962973e-06,
+      "loss": 0.1804,
+      "step": 8999
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4489840506713615e-06,
+      "loss": 0.2214,
+      "step": 9000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4474326124521464e-06,
+      "loss": 0.2332,
+      "step": 9001
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4458819404443235e-06,
+      "loss": 0.2074,
+      "step": 9002
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4443320347868096e-06,
+      "loss": 0.2141,
+      "step": 9003
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4427828956184586e-06,
+      "loss": 0.2068,
+      "step": 9004
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4412345230780588e-06,
+      "loss": 0.1896,
+      "step": 9005
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4396869173043237e-06,
+      "loss": 0.2053,
+      "step": 9006
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4381400784359034e-06,
+      "loss": 0.2144,
+      "step": 9007
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4365940066113748e-06,
+      "loss": 0.169,
+      "step": 9008
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4350487019692494e-06,
+      "loss": 0.2185,
+      "step": 9009
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4335041646479675e-06,
+      "loss": 0.2467,
+      "step": 9010
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.431960394785905e-06,
+      "loss": 0.2032,
+      "step": 9011
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4304173925213616e-06,
+      "loss": 0.208,
+      "step": 9012
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.428875157992573e-06,
+      "loss": 0.2218,
+      "step": 9013
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4273336913377078e-06,
+      "loss": 0.2147,
+      "step": 9014
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4257929926948632e-06,
+      "loss": 0.2162,
+      "step": 9015
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4242530622020668e-06,
+      "loss": 0.1941,
+      "step": 9016
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4227138999972801e-06,
+      "loss": 0.2247,
+      "step": 9017
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.421175506218393e-06,
+      "loss": 0.2249,
+      "step": 9018
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4196378810032285e-06,
+      "loss": 0.1941,
+      "step": 9019
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4181010244895422e-06,
+      "loss": 0.2036,
+      "step": 9020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4165649368150136e-06,
+      "loss": 0.2393,
+      "step": 9021
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4150296181172618e-06,
+      "loss": 0.202,
+      "step": 9022
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.413495068533831e-06,
+      "loss": 0.2392,
+      "step": 9023
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.411961288202206e-06,
+      "loss": 0.1718,
+      "step": 9024
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4104282772597889e-06,
+      "loss": 0.2347,
+      "step": 9025
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4088960358439219e-06,
+      "loss": 0.2094,
+      "step": 9026
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4073645640918754e-06,
+      "loss": 0.1861,
+      "step": 9027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4058338621408541e-06,
+      "loss": 0.1972,
+      "step": 9028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4043039301279904e-06,
+      "loss": 0.1863,
+      "step": 9029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4027747681903447e-06,
+      "loss": 0.213,
+      "step": 9030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4012463764649164e-06,
+      "loss": 0.2094,
+      "step": 9031
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3997187550886315e-06,
+      "loss": 0.209,
+      "step": 9032
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3981919041983483e-06,
+      "loss": 0.2066,
+      "step": 9033
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3966658239308506e-06,
+      "loss": 0.1828,
+      "step": 9034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3951405144228613e-06,
+      "loss": 0.2418,
+      "step": 9035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.393615975811029e-06,
+      "loss": 0.2194,
+      "step": 9036
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3920922082319355e-06,
+      "loss": 0.2142,
+      "step": 9037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3905692118220915e-06,
+      "loss": 0.2338,
+      "step": 9038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3890469867179412e-06,
+      "loss": 0.2262,
+      "step": 9039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.387525533055859e-06,
+      "loss": 0.2074,
+      "step": 9040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.386004850972148e-06,
+      "loss": 0.1898,
+      "step": 9041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3844849406030447e-06,
+      "loss": 0.2322,
+      "step": 9042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.382965802084718e-06,
+      "loss": 0.1966,
+      "step": 9043
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3814474355532604e-06,
+      "loss": 0.2058,
+      "step": 9044
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3799298411447014e-06,
+      "loss": 0.1713,
+      "step": 9045
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3784130189950007e-06,
+      "loss": 0.2297,
+      "step": 9046
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3768969692400492e-06,
+      "loss": 0.205,
+      "step": 9047
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3753816920156649e-06,
+      "loss": 0.2022,
+      "step": 9048
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3738671874576015e-06,
+      "loss": 0.1879,
+      "step": 9049
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3723534557015404e-06,
+      "loss": 0.1993,
+      "step": 9050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3708404968830935e-06,
+      "loss": 0.2101,
+      "step": 9051
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3693283111378087e-06,
+      "loss": 0.2335,
+      "step": 9052
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3678168986011543e-06,
+      "loss": 0.1797,
+      "step": 9053
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3663062594085374e-06,
+      "loss": 0.211,
+      "step": 9054
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3647963936952935e-06,
+      "loss": 0.1882,
+      "step": 9055
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3632873015966942e-06,
+      "loss": 0.231,
+      "step": 9056
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3617789832479311e-06,
+      "loss": 0.1831,
+      "step": 9057
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3602714387841332e-06,
+      "loss": 0.2149,
+      "step": 9058
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3587646683403598e-06,
+      "loss": 0.2167,
+      "step": 9059
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3572586720516e-06,
+      "loss": 0.2072,
+      "step": 9060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3557534500527768e-06,
+      "loss": 0.1984,
+      "step": 9061
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3542490024787347e-06,
+      "loss": 0.2016,
+      "step": 9062
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3527453294642556e-06,
+      "loss": 0.2011,
+      "step": 9063
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3512424311440563e-06,
+      "loss": 0.2056,
+      "step": 9064
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3497403076527782e-06,
+      "loss": 0.207,
+      "step": 9065
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3482389591249901e-06,
+      "loss": 0.2118,
+      "step": 9066
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3467383856951976e-06,
+      "loss": 0.2242,
+      "step": 9067
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3452385874978357e-06,
+      "loss": 0.192,
+      "step": 9068
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3437395646672679e-06,
+      "loss": 0.2337,
+      "step": 9069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3422413173377903e-06,
+      "loss": 0.2266,
+      "step": 9070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3407438456436274e-06,
+      "loss": 0.2065,
+      "step": 9071
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3392471497189362e-06,
+      "loss": 0.2335,
+      "step": 9072
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3377512296978035e-06,
+      "loss": 0.2252,
+      "step": 9073
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3362560857142481e-06,
+      "loss": 0.2477,
+      "step": 9074
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3347617179022132e-06,
+      "loss": 0.2088,
+      "step": 9075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3332681263955794e-06,
+      "loss": 0.2016,
+      "step": 9076
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3317753113281562e-06,
+      "loss": 0.1958,
+      "step": 9077
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3302832728336813e-06,
+      "loss": 0.2158,
+      "step": 9078
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3287920110458251e-06,
+      "loss": 0.2354,
+      "step": 9079
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327301526098186e-06,
+      "loss": 0.1725,
+      "step": 9080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3258118181242963e-06,
+      "loss": 0.2217,
+      "step": 9081
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3243228872576141e-06,
+      "loss": 0.2254,
+      "step": 9082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.322834733631533e-06,
+      "loss": 0.2054,
+      "step": 9083
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3213473573793756e-06,
+      "loss": 0.1864,
+      "step": 9084
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3198607586343891e-06,
+      "loss": 0.2048,
+      "step": 9085
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3183749375297573e-06,
+      "loss": 0.2208,
+      "step": 9086
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.316889894198593e-06,
+      "loss": 0.1973,
+      "step": 9087
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3154056287739404e-06,
+      "loss": 0.2056,
+      "step": 9088
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3139221413887703e-06,
+      "loss": 0.2132,
+      "step": 9089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312439432175987e-06,
+      "loss": 0.2003,
+      "step": 9090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.310957501268425e-06,
+      "loss": 0.2139,
+      "step": 9091
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.309476348798847e-06,
+      "loss": 0.2118,
+      "step": 9092
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3079959748999494e-06,
+      "loss": 0.1983,
+      "step": 9093
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3065163797043523e-06,
+      "loss": 0.1978,
+      "step": 9094
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3050375633446122e-06,
+      "loss": 0.2358,
+      "step": 9095
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3035595259532152e-06,
+      "loss": 0.2166,
+      "step": 9096
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3020822676625788e-06,
+      "loss": 0.2244,
+      "step": 9097
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3006057886050438e-06,
+      "loss": 0.1964,
+      "step": 9098
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2991300889128867e-06,
+      "loss": 0.2122,
+      "step": 9099
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2976551687183137e-06,
+      "loss": 0.197,
+      "step": 9100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2961810281534603e-06,
+      "loss": 0.2355,
+      "step": 9101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2947076673503934e-06,
+      "loss": 0.2109,
+      "step": 9102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.293235086441109e-06,
+      "loss": 0.1896,
+      "step": 9103
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2917632855575323e-06,
+      "loss": 0.2182,
+      "step": 9104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2902922648315209e-06,
+      "loss": 0.2085,
+      "step": 9105
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2888220243948623e-06,
+      "loss": 0.1809,
+      "step": 9106
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2873525643792694e-06,
+      "loss": 0.239,
+      "step": 9107
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2858838849163912e-06,
+      "loss": 0.2126,
+      "step": 9108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2844159861378048e-06,
+      "loss": 0.2273,
+      "step": 9109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2829488681750157e-06,
+      "loss": 0.2161,
+      "step": 9110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2814825311594613e-06,
+      "loss": 0.2292,
+      "step": 9111
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2800169752225089e-06,
+      "loss": 0.1903,
+      "step": 9112
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2785522004954554e-06,
+      "loss": 0.197,
+      "step": 9113
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.277088207109527e-06,
+      "loss": 0.2182,
+      "step": 9114
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2756249951958843e-06,
+      "loss": 0.2112,
+      "step": 9115
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2741625648856083e-06,
+      "loss": 0.2148,
+      "step": 9116
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.272700916309718e-06,
+      "loss": 0.2108,
+      "step": 9117
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2712400495991618e-06,
+      "loss": 0.2097,
+      "step": 9118
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.269779964884814e-06,
+      "loss": 0.2502,
+      "step": 9119
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.268320662297484e-06,
+      "loss": 0.2386,
+      "step": 9120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2668621419679085e-06,
+      "loss": 0.2081,
+      "step": 9121
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.265404404026751e-06,
+      "loss": 0.2271,
+      "step": 9122
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.263947448604611e-06,
+      "loss": 0.2036,
+      "step": 9123
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262491275832014e-06,
+      "loss": 0.2086,
+      "step": 9124
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2610358858394189e-06,
+      "loss": 0.2383,
+      "step": 9125
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.259581278757206e-06,
+      "loss": 0.235,
+      "step": 9126
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.258127454715693e-06,
+      "loss": 0.2015,
+      "step": 9127
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2566744138451282e-06,
+      "loss": 0.2175,
+      "step": 9128
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2552221562756895e-06,
+      "loss": 0.2184,
+      "step": 9129
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2537706821374761e-06,
+      "loss": 0.2036,
+      "step": 9130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2523199915605277e-06,
+      "loss": 0.2101,
+      "step": 9131
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.250870084674808e-06,
+      "loss": 0.2105,
+      "step": 9132
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2494209616102116e-06,
+      "loss": 0.2174,
+      "step": 9133
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.247972622496566e-06,
+      "loss": 0.2227,
+      "step": 9134
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2465250674636197e-06,
+      "loss": 0.217,
+      "step": 9135
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2450782966410625e-06,
+      "loss": 0.2089,
+      "step": 9136
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.243632310158507e-06,
+      "loss": 0.1815,
+      "step": 9137
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2421871081454984e-06,
+      "loss": 0.2325,
+      "step": 9138
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2407426907315056e-06,
+      "loss": 0.224,
+      "step": 9139
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2392990580459351e-06,
+      "loss": 0.2118,
+      "step": 9140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.237856210218118e-06,
+      "loss": 0.1964,
+      "step": 9141
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2364141473773183e-06,
+      "loss": 0.1992,
+      "step": 9142
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2349728696527286e-06,
+      "loss": 0.2065,
+      "step": 9143
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2335323771734687e-06,
+      "loss": 0.209,
+      "step": 9144
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.232092670068592e-06,
+      "loss": 0.2152,
+      "step": 9145
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2306537484670778e-06,
+      "loss": 0.1977,
+      "step": 9146
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2292156124978416e-06,
+      "loss": 0.209,
+      "step": 9147
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2277782622897173e-06,
+      "loss": 0.2313,
+      "step": 9148
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2263416979714782e-06,
+      "loss": 0.213,
+      "step": 9149
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2249059196718228e-06,
+      "loss": 0.2134,
+      "step": 9150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2234709275193812e-06,
+      "loss": 0.2063,
+      "step": 9151
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.222036721642712e-06,
+      "loss": 0.2497,
+      "step": 9152
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2206033021703035e-06,
+      "loss": 0.2096,
+      "step": 9153
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2191706692305737e-06,
+      "loss": 0.1944,
+      "step": 9154
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2177388229518694e-06,
+      "loss": 0.1988,
+      "step": 9155
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2163077634624698e-06,
+      "loss": 0.2419,
+      "step": 9156
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2148774908905782e-06,
+      "loss": 0.2133,
+      "step": 9157
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2134480053643305e-06,
+      "loss": 0.1869,
+      "step": 9158
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2120193070117948e-06,
+      "loss": 0.2229,
+      "step": 9159
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2105913959609605e-06,
+      "loss": 0.2144,
+      "step": 9160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2091642723397623e-06,
+      "loss": 0.1881,
+      "step": 9161
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2077379362760445e-06,
+      "loss": 0.2135,
+      "step": 9162
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2063123878975935e-06,
+      "loss": 0.2133,
+      "step": 9163
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.204887627332122e-06,
+      "loss": 0.2126,
+      "step": 9164
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.203463654707272e-06,
+      "loss": 0.219,
+      "step": 9165
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2020404701506183e-06,
+      "loss": 0.212,
+      "step": 9166
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2006180737896533e-06,
+      "loss": 0.1864,
+      "step": 9167
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.199196465751815e-06,
+      "loss": 0.2058,
+      "step": 9168
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1977756461644618e-06,
+      "loss": 0.1872,
+      "step": 9169
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1963556151548827e-06,
+      "loss": 0.2176,
+      "step": 9170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1949363728502938e-06,
+      "loss": 0.2085,
+      "step": 9171
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1935179193778434e-06,
+      "loss": 0.195,
+      "step": 9172
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.19210025486461e-06,
+      "loss": 0.2344,
+      "step": 9173
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1906833794375993e-06,
+      "loss": 0.1815,
+      "step": 9174
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.189267293223747e-06,
+      "loss": 0.2123,
+      "step": 9175
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.187851996349918e-06,
+      "loss": 0.2148,
+      "step": 9176
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1864374889429065e-06,
+      "loss": 0.2024,
+      "step": 9177
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1850237711294366e-06,
+      "loss": 0.2119,
+      "step": 9178
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1836108430361626e-06,
+      "loss": 0.2041,
+      "step": 9179
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1821987047896632e-06,
+      "loss": 0.1995,
+      "step": 9180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1807873565164507e-06,
+      "loss": 0.2385,
+      "step": 9181
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.179376798342966e-06,
+      "loss": 0.2264,
+      "step": 9182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1779670303955792e-06,
+      "loss": 0.2309,
+      "step": 9183
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1765580528005893e-06,
+      "loss": 0.2303,
+      "step": 9184
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1751498656842242e-06,
+      "loss": 0.1841,
+      "step": 9185
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1737424691726407e-06,
+      "loss": 0.2165,
+      "step": 9186
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1723358633919257e-06,
+      "loss": 0.2101,
+      "step": 9187
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1709300484680973e-06,
+      "loss": 0.2059,
+      "step": 9188
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1695250245270962e-06,
+      "loss": 0.2262,
+      "step": 9189
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1681207916947979e-06,
+      "loss": 0.1968,
+      "step": 9190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1667173500970065e-06,
+      "loss": 0.1902,
+      "step": 9191
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.16531469985945e-06,
+      "loss": 0.2276,
+      "step": 9192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.163912841107797e-06,
+      "loss": 0.2282,
+      "step": 9193
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1625117739676328e-06,
+      "loss": 0.1969,
+      "step": 9194
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1611114985644767e-06,
+      "loss": 0.2068,
+      "step": 9195
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1597120150237796e-06,
+      "loss": 0.2382,
+      "step": 9196
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1583133234709198e-06,
+      "loss": 0.2411,
+      "step": 9197
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1569154240311998e-06,
+      "loss": 0.2103,
+      "step": 9198
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1555183168298578e-06,
+      "loss": 0.2051,
+      "step": 9199
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1541220019920562e-06,
+      "loss": 0.2328,
+      "step": 9200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1527264796428927e-06,
+      "loss": 0.2163,
+      "step": 9201
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1513317499073895e-06,
+      "loss": 0.2187,
+      "step": 9202
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1499378129104953e-06,
+      "loss": 0.2409,
+      "step": 9203
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1485446687770919e-06,
+      "loss": 0.2223,
+      "step": 9204
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1471523176319898e-06,
+      "loss": 0.1962,
+      "step": 9205
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1457607595999276e-06,
+      "loss": 0.1895,
+      "step": 9206
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.144369994805572e-06,
+      "loss": 0.1898,
+      "step": 9207
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1429800233735199e-06,
+      "loss": 0.2083,
+      "step": 9208
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1415908454282964e-06,
+      "loss": 0.1856,
+      "step": 9209
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1402024610943563e-06,
+      "loss": 0.2083,
+      "step": 9210
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1388148704960845e-06,
+      "loss": 0.2298,
+      "step": 9211
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1374280737577904e-06,
+      "loss": 0.2276,
+      "step": 9212
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1360420710037145e-06,
+      "loss": 0.2065,
+      "step": 9213
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1346568623580278e-06,
+      "loss": 0.2226,
+      "step": 9214
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1332724479448309e-06,
+      "loss": 0.2228,
+      "step": 9215
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1318888278881479e-06,
+      "loss": 0.2315,
+      "step": 9216
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1305060023119385e-06,
+      "loss": 0.2048,
+      "step": 9217
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1291239713400847e-06,
+      "loss": 0.2263,
+      "step": 9218
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1277427350964043e-06,
+      "loss": 0.2601,
+      "step": 9219
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.126362293704638e-06,
+      "loss": 0.2365,
+      "step": 9220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1249826472884574e-06,
+      "loss": 0.2353,
+      "step": 9221
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.123603795971462e-06,
+      "loss": 0.2359,
+      "step": 9222
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.122225739877182e-06,
+      "loss": 0.2021,
+      "step": 9223
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1208484791290752e-06,
+      "loss": 0.1969,
+      "step": 9224
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.119472013850529e-06,
+      "loss": 0.196,
+      "step": 9225
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1180963441648574e-06,
+      "loss": 0.2122,
+      "step": 9226
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1167214701953067e-06,
+      "loss": 0.2258,
+      "step": 9227
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.115347392065047e-06,
+      "loss": 0.1831,
+      "step": 9228
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1139741098971835e-06,
+      "loss": 0.2213,
+      "step": 9229
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.112601623814743e-06,
+      "loss": 0.2467,
+      "step": 9230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.111229933940685e-06,
+      "loss": 0.223,
+      "step": 9231
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1098590403978948e-06,
+      "loss": 0.2045,
+      "step": 9232
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.108488943309195e-06,
+      "loss": 0.2034,
+      "step": 9233
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1071196427973285e-06,
+      "loss": 0.2305,
+      "step": 9234
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1057511389849651e-06,
+      "loss": 0.2165,
+      "step": 9235
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.10438343199471e-06,
+      "loss": 0.2012,
+      "step": 9236
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.103016521949093e-06,
+      "loss": 0.1892,
+      "step": 9237
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1016504089705749e-06,
+      "loss": 0.2097,
+      "step": 9238
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100285093181539e-06,
+      "loss": 0.2209,
+      "step": 9239
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0989205747043075e-06,
+      "loss": 0.2273,
+      "step": 9240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0975568536611225e-06,
+      "loss": 0.211,
+      "step": 9241
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0961939301741597e-06,
+      "loss": 0.2126,
+      "step": 9242
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0948318043655214e-06,
+      "loss": 0.2167,
+      "step": 9243
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0934704763572345e-06,
+      "loss": 0.2042,
+      "step": 9244
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0921099462712625e-06,
+      "loss": 0.204,
+      "step": 9245
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.09075021422949e-06,
+      "loss": 0.1992,
+      "step": 9246
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0893912803537354e-06,
+      "loss": 0.1987,
+      "step": 9247
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0880331447657434e-06,
+      "loss": 0.1923,
+      "step": 9248
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0866758075871865e-06,
+      "loss": 0.215,
+      "step": 9249
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0853192689396664e-06,
+      "loss": 0.185,
+      "step": 9250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0839635289447136e-06,
+      "loss": 0.1991,
+      "step": 9251
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0826085877237892e-06,
+      "loss": 0.1984,
+      "step": 9252
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0812544453982765e-06,
+      "loss": 0.2411,
+      "step": 9253
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0799011020894922e-06,
+      "loss": 0.2265,
+      "step": 9254
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0785485579186805e-06,
+      "loss": 0.2177,
+      "step": 9255
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0771968130070143e-06,
+      "loss": 0.2094,
+      "step": 9256
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0758458674755933e-06,
+      "loss": 0.2047,
+      "step": 9257
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0744957214454465e-06,
+      "loss": 0.2027,
+      "step": 9258
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0731463750375336e-06,
+      "loss": 0.2111,
+      "step": 9259
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0717978283727381e-06,
+      "loss": 0.1979,
+      "step": 9260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0704500815718777e-06,
+      "loss": 0.2409,
+      "step": 9261
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0691031347556902e-06,
+      "loss": 0.1947,
+      "step": 9262
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0677569880448479e-06,
+      "loss": 0.2171,
+      "step": 9263
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0664116415599502e-06,
+      "loss": 0.2485,
+      "step": 9264
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0650670954215259e-06,
+      "loss": 0.202,
+      "step": 9265
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0637233497500332e-06,
+      "loss": 0.2136,
+      "step": 9266
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0623804046658503e-06,
+      "loss": 0.2278,
+      "step": 9267
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0610382602892933e-06,
+      "loss": 0.2159,
+      "step": 9268
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0596969167406012e-06,
+      "loss": 0.215,
+      "step": 9269
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0583563741399472e-06,
+      "loss": 0.2086,
+      "step": 9270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0570166326074215e-06,
+      "loss": 0.1877,
+      "step": 9271
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0556776922630519e-06,
+      "loss": 0.2324,
+      "step": 9272
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0543395532267942e-06,
+      "loss": 0.2311,
+      "step": 9273
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0530022156185293e-06,
+      "loss": 0.2184,
+      "step": 9274
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0516656795580703e-06,
+      "loss": 0.1975,
+      "step": 9275
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0503299451651484e-06,
+      "loss": 0.1978,
+      "step": 9276
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0489950125594351e-06,
+      "loss": 0.198,
+      "step": 9277
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0476608818605228e-06,
+      "loss": 0.2209,
+      "step": 9278
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0463275531879358e-06,
+      "loss": 0.2018,
+      "step": 9279
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.044995026661123e-06,
+      "loss": 0.2091,
+      "step": 9280
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0436633023994657e-06,
+      "loss": 0.2249,
+      "step": 9281
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0423323805222696e-06,
+      "loss": 0.2027,
+      "step": 9282
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0410022611487702e-06,
+      "loss": 0.1854,
+      "step": 9283
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0396729443981325e-06,
+      "loss": 0.2107,
+      "step": 9284
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0383444303894453e-06,
+      "loss": 0.1932,
+      "step": 9285
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0370167192417292e-06,
+      "loss": 0.2113,
+      "step": 9286
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.035689811073931e-06,
+      "loss": 0.2059,
+      "step": 9287
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0343637060049272e-06,
+      "loss": 0.1968,
+      "step": 9288
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0330384041535225e-06,
+      "loss": 0.2129,
+      "step": 9289
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0317139056384462e-06,
+      "loss": 0.2045,
+      "step": 9290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0303902105783615e-06,
+      "loss": 0.1993,
+      "step": 9291
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0290673190918533e-06,
+      "loss": 0.2173,
+      "step": 9292
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.02774523129744e-06,
+      "loss": 0.2266,
+      "step": 9293
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0264239473135618e-06,
+      "loss": 0.1917,
+      "step": 9294
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0251034672585925e-06,
+      "loss": 0.2011,
+      "step": 9295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0237837912508332e-06,
+      "loss": 0.2182,
+      "step": 9296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0224649194085067e-06,
+      "loss": 0.1997,
+      "step": 9297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.021146851849777e-06,
+      "loss": 0.1883,
+      "step": 9298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0198295886927213e-06,
+      "loss": 0.1943,
+      "step": 9299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.018513130055353e-06,
+      "loss": 0.2194,
+      "step": 9300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0171974760556115e-06,
+      "loss": 0.2263,
+      "step": 9301
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0158826268113652e-06,
+      "loss": 0.2142,
+      "step": 9302
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0145685824404084e-06,
+      "loss": 0.2089,
+      "step": 9303
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0132553430604608e-06,
+      "loss": 0.2427,
+      "step": 9304
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0119429087891808e-06,
+      "loss": 0.246,
+      "step": 9305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.010631279744142e-06,
+      "loss": 0.1933,
+      "step": 9306
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.009320456042856e-06,
+      "loss": 0.2097,
+      "step": 9307
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0080104378027522e-06,
+      "loss": 0.1916,
+      "step": 9308
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0067012251411957e-06,
+      "loss": 0.1924,
+      "step": 9309
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0053928181754768e-06,
+      "loss": 0.2164,
+      "step": 9310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0040852170228122e-06,
+      "loss": 0.2051,
+      "step": 9311
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.00277842180035e-06,
+      "loss": 0.2247,
+      "step": 9312
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0014724326251624e-06,
+      "loss": 0.2061,
+      "step": 9313
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0001672496142523e-06,
+      "loss": 0.1914,
+      "step": 9314
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988628728845483e-07,
+      "loss": 0.2133,
+      "step": 9315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.975593025529096e-07,
+      "loss": 0.2233,
+      "step": 9316
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962565387361167e-07,
+      "loss": 0.2002,
+      "step": 9317
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.949545815508843e-07,
+      "loss": 0.2113,
+      "step": 9318
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.936534311138536e-07,
+      "loss": 0.2082,
+      "step": 9319
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.92353087541592e-07,
+      "loss": 0.2108,
+      "step": 9320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.910535509505948e-07,
+      "loss": 0.2031,
+      "step": 9321
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.897548214572872e-07,
+      "loss": 0.1869,
+      "step": 9322
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.884568991780186e-07,
+      "loss": 0.2232,
+      "step": 9323
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.87159784229068e-07,
+      "loss": 0.2071,
+      "step": 9324
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.85863476726645e-07,
+      "loss": 0.2153,
+      "step": 9325
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.845679767868787e-07,
+      "loss": 0.2011,
+      "step": 9326
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.83273284525834e-07,
+      "loss": 0.2258,
+      "step": 9327
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.819794000594995e-07,
+      "loss": 0.2063,
+      "step": 9328
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.80686323503791e-07,
+      "loss": 0.2093,
+      "step": 9329
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.793940549745583e-07,
+      "loss": 0.2268,
+      "step": 9330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.7810259458757e-07,
+      "loss": 0.2237,
+      "step": 9331
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.768119424585254e-07,
+      "loss": 0.2021,
+      "step": 9332
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.755220987030523e-07,
+      "loss": 0.2018,
+      "step": 9333
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.742330634367091e-07,
+      "loss": 0.2329,
+      "step": 9334
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.72944836774975e-07,
+      "loss": 0.2109,
+      "step": 9335
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.716574188332583e-07,
+      "loss": 0.1997,
+      "step": 9336
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.703708097269027e-07,
+      "loss": 0.2105,
+      "step": 9337
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.69085009571169e-07,
+      "loss": 0.1948,
+      "step": 9338
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.678000184812553e-07,
+      "loss": 0.2058,
+      "step": 9339
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66515836572276e-07,
+      "loss": 0.2547,
+      "step": 9340
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.652324639592803e-07,
+      "loss": 0.2027,
+      "step": 9341
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.63949900757245e-07,
+      "loss": 0.1893,
+      "step": 9342
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.626681470810751e-07,
+      "loss": 0.2175,
+      "step": 9343
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.61387203045594e-07,
+      "loss": 0.2159,
+      "step": 9344
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.601070687655667e-07,
+      "loss": 0.2265,
+      "step": 9345
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.588277443556747e-07,
+      "loss": 0.2362,
+      "step": 9346
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.575492299305323e-07,
+      "loss": 0.2162,
+      "step": 9347
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.562715256046805e-07,
+      "loss": 0.2063,
+      "step": 9348
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54994631492584e-07,
+      "loss": 0.2046,
+      "step": 9349
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.537185477086397e-07,
+      "loss": 0.2281,
+      "step": 9350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.524432743671685e-07,
+      "loss": 0.2455,
+      "step": 9351
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.511688115824214e-07,
+      "loss": 0.2099,
+      "step": 9352
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.498951594685757e-07,
+      "loss": 0.2197,
+      "step": 9353
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.486223181397347e-07,
+      "loss": 0.2263,
+      "step": 9354
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.473502877099317e-07,
+      "loss": 0.1777,
+      "step": 9355
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.460790682931253e-07,
+      "loss": 0.2254,
+      "step": 9356
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.448086600032047e-07,
+      "loss": 0.2002,
+      "step": 9357
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.435390629539798e-07,
+      "loss": 0.2266,
+      "step": 9358
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.422702772591918e-07,
+      "loss": 0.199,
+      "step": 9359
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.410023030325122e-07,
+      "loss": 0.2001,
+      "step": 9360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.397351403875354e-07,
+      "loss": 0.2075,
+      "step": 9361
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.384687894377842e-07,
+      "loss": 0.2194,
+      "step": 9362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.372032502967099e-07,
+      "loss": 0.2095,
+      "step": 9363
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.359385230776896e-07,
+      "loss": 0.2079,
+      "step": 9364
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.346746078940294e-07,
+      "loss": 0.2212,
+      "step": 9365
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.334115048589621e-07,
+      "loss": 0.2073,
+      "step": 9366
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.321492140856425e-07,
+      "loss": 0.185,
+      "step": 9367
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.308877356871615e-07,
+      "loss": 0.2175,
+      "step": 9368
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.296270697765308e-07,
+      "loss": 0.2039,
+      "step": 9369
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.283672164666935e-07,
+      "loss": 0.1856,
+      "step": 9370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.271081758705203e-07,
+      "loss": 0.2196,
+      "step": 9371
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.25849948100801e-07,
+      "loss": 0.21,
+      "step": 9372
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.245925332702599e-07,
+      "loss": 0.2001,
+      "step": 9373
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.233359314915491e-07,
+      "loss": 0.2261,
+      "step": 9374
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.220801428772464e-07,
+      "loss": 0.2041,
+      "step": 9375
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.208251675398494e-07,
+      "loss": 0.2195,
+      "step": 9376
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.195710055917961e-07,
+      "loss": 0.1992,
+      "step": 9377
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.183176571454422e-07,
+      "loss": 0.2122,
+      "step": 9378
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.170651223130745e-07,
+      "loss": 0.216,
+      "step": 9379
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.158134012069075e-07,
+      "loss": 0.2132,
+      "step": 9380
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.145624939390763e-07,
+      "loss": 0.2091,
+      "step": 9381
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.133124006216487e-07,
+      "loss": 0.2127,
+      "step": 9382
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.120631213666209e-07,
+      "loss": 0.211,
+      "step": 9383
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.108146562859133e-07,
+      "loss": 0.2045,
+      "step": 9384
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.09567005491373e-07,
+      "loss": 0.2026,
+      "step": 9385
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.083201690947763e-07,
+      "loss": 0.2053,
+      "step": 9386
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070741472078248e-07,
+      "loss": 0.2301,
+      "step": 9387
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.058289399421471e-07,
+      "loss": 0.2199,
+      "step": 9388
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.045845474093018e-07,
+      "loss": 0.2309,
+      "step": 9389
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.033409697207685e-07,
+      "loss": 0.1756,
+      "step": 9390
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.020982069879592e-07,
+      "loss": 0.2315,
+      "step": 9391
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.008562593222114e-07,
+      "loss": 0.2186,
+      "step": 9392
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.996151268347886e-07,
+      "loss": 0.2076,
+      "step": 9393
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.983748096368827e-07,
+      "loss": 0.209,
+      "step": 9394
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.971353078396117e-07,
+      "loss": 0.229,
+      "step": 9395
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.958966215540199e-07,
+      "loss": 0.2184,
+      "step": 9396
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.946587508910798e-07,
+      "loss": 0.2051,
+      "step": 9397
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.934216959616926e-07,
+      "loss": 0.2024,
+      "step": 9398
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.921854568766797e-07,
+      "loss": 0.2113,
+      "step": 9399
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.909500337467969e-07,
+      "loss": 0.2018,
+      "step": 9400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.897154266827202e-07,
+      "loss": 0.241,
+      "step": 9401
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.884816357950632e-07,
+      "loss": 0.1993,
+      "step": 9402
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.872486611943532e-07,
+      "loss": 0.2092,
+      "step": 9403
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.860165029910528e-07,
+      "loss": 0.2048,
+      "step": 9404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.847851612955483e-07,
+      "loss": 0.2233,
+      "step": 9405
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.835546362181535e-07,
+      "loss": 0.2152,
+      "step": 9406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.823249278691126e-07,
+      "loss": 0.2273,
+      "step": 9407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.810960363585896e-07,
+      "loss": 0.218,
+      "step": 9408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.798679617966777e-07,
+      "loss": 0.2058,
+      "step": 9409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.786407042934019e-07,
+      "loss": 0.2004,
+      "step": 9410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.774142639587091e-07,
+      "loss": 0.2219,
+      "step": 9411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.761886409024755e-07,
+      "loss": 0.2264,
+      "step": 9412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.749638352345002e-07,
+      "loss": 0.2143,
+      "step": 9413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.737398470645131e-07,
+      "loss": 0.2089,
+      "step": 9414
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.725166765021686e-07,
+      "loss": 0.1982,
+      "step": 9415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.712943236570493e-07,
+      "loss": 0.2109,
+      "step": 9416
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.700727886386639e-07,
+      "loss": 0.2278,
+      "step": 9417
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.688520715564486e-07,
+      "loss": 0.2059,
+      "step": 9418
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.676321725197634e-07,
+      "loss": 0.227,
+      "step": 9419
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.664130916378999e-07,
+      "loss": 0.2396,
+      "step": 9420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.651948290200729e-07,
+      "loss": 0.219,
+      "step": 9421
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.639773847754229e-07,
+      "loss": 0.2408,
+      "step": 9422
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.627607590130205e-07,
+      "loss": 0.2204,
+      "step": 9423
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.615449518418606e-07,
+      "loss": 0.2078,
+      "step": 9424
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.603299633708662e-07,
+      "loss": 0.2103,
+      "step": 9425
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.591157937088845e-07,
+      "loss": 0.2085,
+      "step": 9426
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.579024429646932e-07,
+      "loss": 0.2382,
+      "step": 9427
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.56689911246994e-07,
+      "loss": 0.1922,
+      "step": 9428
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.554781986644156e-07,
+      "loss": 0.2177,
+      "step": 9429
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.542673053255157e-07,
+      "loss": 0.2296,
+      "step": 9430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.530572313387708e-07,
+      "loss": 0.2221,
+      "step": 9431
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.518479768125942e-07,
+      "loss": 0.2165,
+      "step": 9432
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.506395418553193e-07,
+      "loss": 0.197,
+      "step": 9433
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.494319265752083e-07,
+      "loss": 0.2161,
+      "step": 9434
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.482251310804501e-07,
+      "loss": 0.2083,
+      "step": 9435
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.470191554791596e-07,
+      "loss": 0.2076,
+      "step": 9436
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.458139998793779e-07,
+      "loss": 0.1997,
+      "step": 9437
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.446096643890722e-07,
+      "loss": 0.2156,
+      "step": 9438
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.434061491161405e-07,
+      "loss": 0.1907,
+      "step": 9439
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.422034541683999e-07,
+      "loss": 0.2032,
+      "step": 9440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.410015796535975e-07,
+      "loss": 0.1835,
+      "step": 9441
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.398005256794117e-07,
+      "loss": 0.215,
+      "step": 9442
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.38600292353442e-07,
+      "loss": 0.2181,
+      "step": 9443
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.374008797832122e-07,
+      "loss": 0.2134,
+      "step": 9444
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.362022880761778e-07,
+      "loss": 0.2081,
+      "step": 9445
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.350045173397192e-07,
+      "loss": 0.2089,
+      "step": 9446
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.33807567681143e-07,
+      "loss": 0.2419,
+      "step": 9447
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.326114392076812e-07,
+      "loss": 0.2044,
+      "step": 9448
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.314161320264936e-07,
+      "loss": 0.2274,
+      "step": 9449
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.302216462446655e-07,
+      "loss": 0.2186,
+      "step": 9450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.290279819692093e-07,
+      "loss": 0.1852,
+      "step": 9451
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.278351393070649e-07,
+      "loss": 0.2412,
+      "step": 9452
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.266431183650969e-07,
+      "loss": 0.2198,
+      "step": 9453
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.254519192500932e-07,
+      "loss": 0.2203,
+      "step": 9454
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.24261542068775e-07,
+      "loss": 0.2337,
+      "step": 9455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.230719869277848e-07,
+      "loss": 0.1968,
+      "step": 9456
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.218832539336941e-07,
+      "loss": 0.2102,
+      "step": 9457
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.206953431929987e-07,
+      "loss": 0.1922,
+      "step": 9458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.195082548121225e-07,
+      "loss": 0.2033,
+      "step": 9459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.183219888974148e-07,
+      "loss": 0.2071,
+      "step": 9460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171365455551506e-07,
+      "loss": 0.2323,
+      "step": 9461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.159519248915349e-07,
+      "loss": 0.2071,
+      "step": 9462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.147681270126917e-07,
+      "loss": 0.2102,
+      "step": 9463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.135851520246773e-07,
+      "loss": 0.2112,
+      "step": 9464
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.124030000334726e-07,
+      "loss": 0.2271,
+      "step": 9465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.112216711449861e-07,
+      "loss": 0.2133,
+      "step": 9466
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.100411654650486e-07,
+      "loss": 0.2415,
+      "step": 9467
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.088614830994223e-07,
+      "loss": 0.2103,
+      "step": 9468
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.076826241537927e-07,
+      "loss": 0.2434,
+      "step": 9469
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.065045887337708e-07,
+      "loss": 0.1979,
+      "step": 9470
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.053273769448977e-07,
+      "loss": 0.1987,
+      "step": 9471
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.041509888926346e-07,
+      "loss": 0.2104,
+      "step": 9472
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.029754246823717e-07,
+      "loss": 0.2047,
+      "step": 9473
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.018006844194293e-07,
+      "loss": 0.2078,
+      "step": 9474
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.006267682090519e-07,
+      "loss": 0.2154,
+      "step": 9475
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.994536761564042e-07,
+      "loss": 0.2054,
+      "step": 9476
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.982814083665825e-07,
+      "loss": 0.206,
+      "step": 9477
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.971099649446112e-07,
+      "loss": 0.1962,
+      "step": 9478
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.959393459954356e-07,
+      "loss": 0.1856,
+      "step": 9479
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.947695516239329e-07,
+      "loss": 0.2219,
+      "step": 9480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.93600581934898e-07,
+      "loss": 0.2112,
+      "step": 9481
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.924324370330605e-07,
+      "loss": 0.2433,
+      "step": 9482
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.912651170230734e-07,
+      "loss": 0.2235,
+      "step": 9483
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.900986220095153e-07,
+      "loss": 0.2006,
+      "step": 9484
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.88932952096888e-07,
+      "loss": 0.2196,
+      "step": 9485
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.877681073896226e-07,
+      "loss": 0.192,
+      "step": 9486
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.866040879920767e-07,
+      "loss": 0.192,
+      "step": 9487
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.854408940085334e-07,
+      "loss": 0.226,
+      "step": 9488
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.842785255432017e-07,
+      "loss": 0.1998,
+      "step": 9489
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.831169827002138e-07,
+      "loss": 0.2035,
+      "step": 9490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.819562655836343e-07,
+      "loss": 0.2127,
+      "step": 9491
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.807963742974467e-07,
+      "loss": 0.2341,
+      "step": 9492
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.796373089455666e-07,
+      "loss": 0.1867,
+      "step": 9493
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.784790696318323e-07,
+      "loss": 0.2177,
+      "step": 9494
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.773216564600072e-07,
+      "loss": 0.2141,
+      "step": 9495
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.76165069533783e-07,
+      "loss": 0.2175,
+      "step": 9496
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.750093089567756e-07,
+      "loss": 0.2099,
+      "step": 9497
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.7385437483253e-07,
+      "loss": 0.2293,
+      "step": 9498
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.727002672645123e-07,
+      "loss": 0.1625,
+      "step": 9499
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.715469863561198e-07,
+      "loss": 0.2265,
+      "step": 9500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.703945322106721e-07,
+      "loss": 0.2016,
+      "step": 9501
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.692429049314154e-07,
+      "loss": 0.2211,
+      "step": 9502
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.680921046215239e-07,
+      "loss": 0.2011,
+      "step": 9503
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.66942131384093e-07,
+      "loss": 0.2089,
+      "step": 9504
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.657929853221491e-07,
+      "loss": 0.2078,
+      "step": 9505
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.646446665386409e-07,
+      "loss": 0.1958,
+      "step": 9506
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.634971751364484e-07,
+      "loss": 0.2032,
+      "step": 9507
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.623505112183694e-07,
+      "loss": 0.2006,
+      "step": 9508
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.1885,
+      "step": 9509
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.600596662453929e-07,
+      "loss": 0.1981,
+      "step": 9510
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.589154853957292e-07,
+      "loss": 0.2014,
+      "step": 9511
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.577721324406495e-07,
+      "loss": 0.2196,
+      "step": 9512
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.566296074825785e-07,
+      "loss": 0.1977,
+      "step": 9513
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.554879106238788e-07,
+      "loss": 0.2199,
+      "step": 9514
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.54347041966832e-07,
+      "loss": 0.2089,
+      "step": 9515
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.532070016136483e-07,
+      "loss": 0.2159,
+      "step": 9516
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.520677896664586e-07,
+      "loss": 0.2134,
+      "step": 9517
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.509294062273254e-07,
+      "loss": 0.2103,
+      "step": 9518
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.497918513982339e-07,
+      "loss": 0.2118,
+      "step": 9519
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.48655125281097e-07,
+      "loss": 0.195,
+      "step": 9520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.47519227977751e-07,
+      "loss": 0.2233,
+      "step": 9521
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.463841595899591e-07,
+      "loss": 0.1927,
+      "step": 9522
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.452499202194119e-07,
+      "loss": 0.2191,
+      "step": 9523
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.441165099677228e-07,
+      "loss": 0.2069,
+      "step": 9524
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.429839289364349e-07,
+      "loss": 0.2253,
+      "step": 9525
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.418521772270093e-07,
+      "loss": 0.2291,
+      "step": 9526
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.407212549408427e-07,
+      "loss": 0.1933,
+      "step": 9527
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.395911621792495e-07,
+      "loss": 0.2062,
+      "step": 9528
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.384618990434745e-07,
+      "loss": 0.2065,
+      "step": 9529
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.373334656346864e-07,
+      "loss": 0.1976,
+      "step": 9530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.362058620539803e-07,
+      "loss": 0.2223,
+      "step": 9531
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.350790884023761e-07,
+      "loss": 0.2283,
+      "step": 9532
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.339531447808191e-07,
+      "loss": 0.2305,
+      "step": 9533
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.328280312901815e-07,
+      "loss": 0.212,
+      "step": 9534
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.317037480312628e-07,
+      "loss": 0.2081,
+      "step": 9535
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.305802951047814e-07,
+      "loss": 0.2137,
+      "step": 9536
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.294576726113866e-07,
+      "loss": 0.2061,
+      "step": 9537
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.283358806516538e-07,
+      "loss": 0.1833,
+      "step": 9538
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.272149193260847e-07,
+      "loss": 0.1972,
+      "step": 9539
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.260947887350999e-07,
+      "loss": 0.2327,
+      "step": 9540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.249754889790539e-07,
+      "loss": 0.2265,
+      "step": 9541
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.238570201582207e-07,
+      "loss": 0.204,
+      "step": 9542
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.227393823728035e-07,
+      "loss": 0.2022,
+      "step": 9543
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.216225757229311e-07,
+      "loss": 0.1814,
+      "step": 9544
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.205066003086525e-07,
+      "loss": 0.2082,
+      "step": 9545
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.193914562299475e-07,
+      "loss": 0.1977,
+      "step": 9546
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.182771435867231e-07,
+      "loss": 0.2126,
+      "step": 9547
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.171636624788081e-07,
+      "loss": 0.2101,
+      "step": 9548
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.160510130059539e-07,
+      "loss": 0.2178,
+      "step": 9549
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.149391952678453e-07,
+      "loss": 0.2197,
+      "step": 9550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.138282093640858e-07,
+      "loss": 0.2032,
+      "step": 9551
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.12718055394207e-07,
+      "loss": 0.2149,
+      "step": 9552
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.116087334576672e-07,
+      "loss": 0.2189,
+      "step": 9553
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.105002436538478e-07,
+      "loss": 0.2033,
+      "step": 9554
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.093925860820583e-07,
+      "loss": 0.2242,
+      "step": 9555
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.082857608415294e-07,
+      "loss": 0.2321,
+      "step": 9556
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.07179768031424e-07,
+      "loss": 0.1987,
+      "step": 9557
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.060746077508219e-07,
+      "loss": 0.2281,
+      "step": 9558
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.049702800987346e-07,
+      "loss": 0.2366,
+      "step": 9559
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.038667851740966e-07,
+      "loss": 0.187,
+      "step": 9560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.027641230757687e-07,
+      "loss": 0.2301,
+      "step": 9561
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.016622939025374e-07,
+      "loss": 0.181,
+      "step": 9562
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.005612977531129e-07,
+      "loss": 0.2129,
+      "step": 9563
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.994611347261316e-07,
+      "loss": 0.1704,
+      "step": 9564
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.983618049201557e-07,
+      "loss": 0.1849,
+      "step": 9565
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.972633084336744e-07,
+      "loss": 0.2112,
+      "step": 9566
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.961656453650978e-07,
+      "loss": 0.2417,
+      "step": 9567
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.950688158127639e-07,
+      "loss": 0.1932,
+      "step": 9568
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.939728198749362e-07,
+      "loss": 0.2412,
+      "step": 9569
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.928776576498053e-07,
+      "loss": 0.2103,
+      "step": 9570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.917833292354825e-07,
+      "loss": 0.1881,
+      "step": 9571
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.906898347300084e-07,
+      "loss": 0.2059,
+      "step": 9572
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.895971742313468e-07,
+      "loss": 0.1899,
+      "step": 9573
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.885053478373882e-07,
+      "loss": 0.2097,
+      "step": 9574
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.874143556459468e-07,
+      "loss": 0.2154,
+      "step": 9575
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.863241977547663e-07,
+      "loss": 0.2232,
+      "step": 9576
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.852348742615067e-07,
+      "loss": 0.2141,
+      "step": 9577
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.841463852637598e-07,
+      "loss": 0.2163,
+      "step": 9578
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.830587308590453e-07,
+      "loss": 0.2327,
+      "step": 9579
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.819719111448043e-07,
+      "loss": 0.1985,
+      "step": 9580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.808859262183998e-07,
+      "loss": 0.2132,
+      "step": 9581
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.798007761771263e-07,
+      "loss": 0.1999,
+      "step": 9582
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.787164611181984e-07,
+      "loss": 0.1872,
+      "step": 9583
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.776329811387605e-07,
+      "loss": 0.1894,
+      "step": 9584
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.765503363358783e-07,
+      "loss": 0.2072,
+      "step": 9585
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.754685268065453e-07,
+      "loss": 0.1967,
+      "step": 9586
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.743875526476785e-07,
+      "loss": 0.1933,
+      "step": 9587
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.733074139561213e-07,
+      "loss": 0.1977,
+      "step": 9588
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.722281108286422e-07,
+      "loss": 0.2083,
+      "step": 9589
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.711496433619324e-07,
+      "loss": 0.2206,
+      "step": 9590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.700720116526116e-07,
+      "loss": 0.2327,
+      "step": 9591
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.689952157972224e-07,
+      "loss": 0.2093,
+      "step": 9592
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.679192558922343e-07,
+      "loss": 0.2017,
+      "step": 9593
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.668441320340402e-07,
+      "loss": 0.2038,
+      "step": 9594
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.657698443189586e-07,
+      "loss": 0.1862,
+      "step": 9595
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.646963928432337e-07,
+      "loss": 0.2282,
+      "step": 9596
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.636237777030341e-07,
+      "loss": 0.2125,
+      "step": 9597
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.625519989944551e-07,
+      "loss": 0.2132,
+      "step": 9598
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.614810568135132e-07,
+      "loss": 0.2009,
+      "step": 9599
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.604109512561541e-07,
+      "loss": 0.2066,
+      "step": 9600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.593416824182463e-07,
+      "loss": 0.2069,
+      "step": 9601
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.582732503955847e-07,
+      "loss": 0.2383,
+      "step": 9602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.572056552838879e-07,
+      "loss": 0.2028,
+      "step": 9603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.561388971787997e-07,
+      "loss": 0.2251,
+      "step": 9604
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.550729761758901e-07,
+      "loss": 0.1988,
+      "step": 9605
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.540078923706538e-07,
+      "loss": 0.216,
+      "step": 9606
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.529436458585103e-07,
+      "loss": 0.2188,
+      "step": 9607
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.518802367348009e-07,
+      "loss": 0.1797,
+      "step": 9608
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.508176650947973e-07,
+      "loss": 0.2046,
+      "step": 9609
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.49755931033691e-07,
+      "loss": 0.2231,
+      "step": 9610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.486950346466047e-07,
+      "loss": 0.2182,
+      "step": 9611
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.476349760285838e-07,
+      "loss": 0.1997,
+      "step": 9612
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.465757552745921e-07,
+      "loss": 0.185,
+      "step": 9613
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.45517372479526e-07,
+      "loss": 0.2318,
+      "step": 9614
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.444598277382041e-07,
+      "loss": 0.2211,
+      "step": 9615
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.434031211453706e-07,
+      "loss": 0.1962,
+      "step": 9616
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.423472527956953e-07,
+      "loss": 0.2108,
+      "step": 9617
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.412922227837681e-07,
+      "loss": 0.1936,
+      "step": 9618
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.402380312041112e-07,
+      "loss": 0.1832,
+      "step": 9619
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.391846781511668e-07,
+      "loss": 0.1879,
+      "step": 9620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.38132163719305e-07,
+      "loss": 0.2028,
+      "step": 9621
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.370804880028158e-07,
+      "loss": 0.1962,
+      "step": 9622
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.360296510959186e-07,
+      "loss": 0.2357,
+      "step": 9623
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.349796530927565e-07,
+      "loss": 0.2075,
+      "step": 9624
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.339304940873969e-07,
+      "loss": 0.2138,
+      "step": 9625
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.328821741738311e-07,
+      "loss": 0.1863,
+      "step": 9626
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.318346934459796e-07,
+      "loss": 0.2308,
+      "step": 9627
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.307880519976817e-07,
+      "loss": 0.2251,
+      "step": 9628
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.297422499227057e-07,
+      "loss": 0.2093,
+      "step": 9629
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.286972873147445e-07,
+      "loss": 0.2068,
+      "step": 9630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.276531642674122e-07,
+      "loss": 0.1989,
+      "step": 9631
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.266098808742515e-07,
+      "loss": 0.2128,
+      "step": 9632
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.255674372287279e-07,
+      "loss": 0.208,
+      "step": 9633
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.245258334242332e-07,
+      "loss": 0.1983,
+      "step": 9634
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.234850695540829e-07,
+      "loss": 0.2141,
+      "step": 9635
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.224451457115177e-07,
+      "loss": 0.2186,
+      "step": 9636
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.214060619897011e-07,
+      "loss": 0.2162,
+      "step": 9637
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.203678184817252e-07,
+      "loss": 0.216,
+      "step": 9638
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.193304152806046e-07,
+      "loss": 0.2181,
+      "step": 9639
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.18293852479277e-07,
+      "loss": 0.2003,
+      "step": 9640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.172581301706071e-07,
+      "loss": 0.2031,
+      "step": 9641
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.162232484473829e-07,
+      "loss": 0.208,
+      "step": 9642
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.151892074023181e-07,
+      "loss": 0.1627,
+      "step": 9643
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.141560071280539e-07,
+      "loss": 0.2066,
+      "step": 9644
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.131236477171498e-07,
+      "loss": 0.2033,
+      "step": 9645
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.120921292620951e-07,
+      "loss": 0.201,
+      "step": 9646
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.110614518552993e-07,
+      "loss": 0.2188,
+      "step": 9647
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.10031615589104e-07,
+      "loss": 0.1939,
+      "step": 9648
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.090026205557653e-07,
+      "loss": 0.2113,
+      "step": 9649
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.079744668474707e-07,
+      "loss": 0.196,
+      "step": 9650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.069471545563333e-07,
+      "loss": 0.1865,
+      "step": 9651
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.059206837743869e-07,
+      "loss": 0.2234,
+      "step": 9652
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.048950545935927e-07,
+      "loss": 0.2321,
+      "step": 9653
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.038702671058327e-07,
+      "loss": 0.2203,
+      "step": 9654
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.028463214029179e-07,
+      "loss": 0.189,
+      "step": 9655
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.018232175765804e-07,
+      "loss": 0.2056,
+      "step": 9656
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008009557184813e-07,
+      "loss": 0.1957,
+      "step": 9657
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.997795359202008e-07,
+      "loss": 0.2067,
+      "step": 9658
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.987589582732478e-07,
+      "loss": 0.2106,
+      "step": 9659
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.977392228690526e-07,
+      "loss": 0.2004,
+      "step": 9660
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.967203297989743e-07,
+      "loss": 0.2091,
+      "step": 9661
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.957022791542943e-07,
+      "loss": 0.2209,
+      "step": 9662
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.946850710262142e-07,
+      "loss": 0.2093,
+      "step": 9663
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.936687055058665e-07,
+      "loss": 0.2034,
+      "step": 9664
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.926531826843062e-07,
+      "loss": 0.1976,
+      "step": 9665
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.916385026525118e-07,
+      "loss": 0.2081,
+      "step": 9666
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.90624665501387e-07,
+      "loss": 0.1866,
+      "step": 9667
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.896116713217603e-07,
+      "loss": 0.2127,
+      "step": 9668
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.885995202043848e-07,
+      "loss": 0.2363,
+      "step": 9669
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875882122399356e-07,
+      "loss": 0.196,
+      "step": 9670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.865777475190171e-07,
+      "loss": 0.2092,
+      "step": 9671
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.855681261321522e-07,
+      "loss": 0.2063,
+      "step": 9672
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.845593481697931e-07,
+      "loss": 0.2162,
+      "step": 9673
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.835514137223142e-07,
+      "loss": 0.2057,
+      "step": 9674
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.825443228800132e-07,
+      "loss": 0.2223,
+      "step": 9675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.815380757331191e-07,
+      "loss": 0.2063,
+      "step": 9676
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.805326723717741e-07,
+      "loss": 0.2227,
+      "step": 9677
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.795281128860542e-07,
+      "loss": 0.2208,
+      "step": 9678
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.785243973659537e-07,
+      "loss": 0.2089,
+      "step": 9679
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.775215259013988e-07,
+      "loss": 0.2106,
+      "step": 9680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.765194985822298e-07,
+      "loss": 0.2147,
+      "step": 9681
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.75518315498218e-07,
+      "loss": 0.1933,
+      "step": 9682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.745179767390585e-07,
+      "loss": 0.2253,
+      "step": 9683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.735184823943707e-07,
+      "loss": 0.2095,
+      "step": 9684
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.725198325536996e-07,
+      "loss": 0.2568,
+      "step": 9685
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.715220273065092e-07,
+      "loss": 0.2469,
+      "step": 9686
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.705250667421925e-07,
+      "loss": 0.198,
+      "step": 9687
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.695289509500657e-07,
+      "loss": 0.2029,
+      "step": 9688
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.685336800193719e-07,
+      "loss": 0.2303,
+      "step": 9689
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.675392540392699e-07,
+      "loss": 0.1837,
+      "step": 9690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.665456730988527e-07,
+      "loss": 0.2206,
+      "step": 9691
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.655529372871348e-07,
+      "loss": 0.2065,
+      "step": 9692
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.645610466930518e-07,
+      "loss": 0.1906,
+      "step": 9693
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.635700014054679e-07,
+      "loss": 0.2149,
+      "step": 9694
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.625798015131656e-07,
+      "loss": 0.1972,
+      "step": 9695
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.615904471048584e-07,
+      "loss": 0.2225,
+      "step": 9696
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.606019382691797e-07,
+      "loss": 0.2059,
+      "step": 9697
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.596142750946887e-07,
+      "loss": 0.2058,
+      "step": 9698
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.586274576698691e-07,
+      "loss": 0.2108,
+      "step": 9699
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.57641486083128e-07,
+      "loss": 0.2205,
+      "step": 9700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.56656360422797e-07,
+      "loss": 0.2011,
+      "step": 9701
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.556720807771332e-07,
+      "loss": 0.2378,
+      "step": 9702
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.546886472343161e-07,
+      "loss": 0.2325,
+      "step": 9703
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.537060598824484e-07,
+      "loss": 0.2325,
+      "step": 9704
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.527243188095599e-07,
+      "loss": 0.2074,
+      "step": 9705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.517434241036035e-07,
+      "loss": 0.2064,
+      "step": 9706
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.507633758524544e-07,
+      "loss": 0.2095,
+      "step": 9707
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.497841741439159e-07,
+      "loss": 0.202,
+      "step": 9708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.488058190657119e-07,
+      "loss": 0.2108,
+      "step": 9709
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.478283107054926e-07,
+      "loss": 0.2134,
+      "step": 9710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.468516491508303e-07,
+      "loss": 0.2165,
+      "step": 9711
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.458758344892246e-07,
+      "loss": 0.2283,
+      "step": 9712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.449008668080947e-07,
+      "loss": 0.2087,
+      "step": 9713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.439267461947884e-07,
+      "loss": 0.2484,
+      "step": 9714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.429534727365715e-07,
+      "loss": 0.1689,
+      "step": 9715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.419810465206443e-07,
+      "loss": 0.213,
+      "step": 9716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.410094676341237e-07,
+      "loss": 0.2189,
+      "step": 9717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.40038736164048e-07,
+      "loss": 0.1804,
+      "step": 9718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.390688521973864e-07,
+      "loss": 0.2079,
+      "step": 9719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.380998158210293e-07,
+      "loss": 0.1946,
+      "step": 9720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.371316271217919e-07,
+      "loss": 0.2042,
+      "step": 9721
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.36164286186408e-07,
+      "loss": 0.2223,
+      "step": 9722
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.351977931015461e-07,
+      "loss": 0.2386,
+      "step": 9723
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.342321479537904e-07,
+      "loss": 0.2205,
+      "step": 9724
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.332673508296515e-07,
+      "loss": 0.2151,
+      "step": 9725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.323034018155659e-07,
+      "loss": 0.2064,
+      "step": 9726
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.313403009978891e-07,
+      "loss": 0.2153,
+      "step": 9727
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.303780484629051e-07,
+      "loss": 0.2187,
+      "step": 9728
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.294166442968208e-07,
+      "loss": 0.1918,
+      "step": 9729
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.284560885857671e-07,
+      "loss": 0.2054,
+      "step": 9730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.274963814157996e-07,
+      "loss": 0.2132,
+      "step": 9731
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.265375228728953e-07,
+      "loss": 0.2179,
+      "step": 9732
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.255795130429575e-07,
+      "loss": 0.2181,
+      "step": 9733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.246223520118133e-07,
+      "loss": 0.2275,
+      "step": 9734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.236660398652138e-07,
+      "loss": 0.2044,
+      "step": 9735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.227105766888319e-07,
+      "loss": 0.2348,
+      "step": 9736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.217559625682666e-07,
+      "loss": 0.2091,
+      "step": 9737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.20802197589041e-07,
+      "loss": 0.2223,
+      "step": 9738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.198492818365996e-07,
+      "loss": 0.2042,
+      "step": 9739
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.188972153963157e-07,
+      "loss": 0.2072,
+      "step": 9740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.179459983534807e-07,
+      "loss": 0.1992,
+      "step": 9741
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.169956307933143e-07,
+      "loss": 0.1845,
+      "step": 9742
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.160461128009586e-07,
+      "loss": 0.1731,
+      "step": 9743
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.150974444614787e-07,
+      "loss": 0.2123,
+      "step": 9744
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.141496258598644e-07,
+      "loss": 0.1982,
+      "step": 9745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.132026570810289e-07,
+      "loss": 0.2295,
+      "step": 9746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.122565382098088e-07,
+      "loss": 0.2059,
+      "step": 9747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.113112693309685e-07,
+      "loss": 0.2104,
+      "step": 9748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.103668505291926e-07,
+      "loss": 0.2031,
+      "step": 9749
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.094232818890876e-07,
+      "loss": 0.2062,
+      "step": 9750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.084805634951884e-07,
+      "loss": 0.2231,
+      "step": 9751
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.075386954319506e-07,
+      "loss": 0.2274,
+      "step": 9752
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065976777837567e-07,
+      "loss": 0.2011,
+      "step": 9753
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.056575106349082e-07,
+      "loss": 0.179,
+      "step": 9754
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.047181940696333e-07,
+      "loss": 0.2079,
+      "step": 9755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.037797281720857e-07,
+      "loss": 0.1968,
+      "step": 9756
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.028421130263416e-07,
+      "loss": 0.2163,
+      "step": 9757
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.019053487164005e-07,
+      "loss": 0.1975,
+      "step": 9758
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.009694353261841e-07,
+      "loss": 0.1816,
+      "step": 9759
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.000343729395396e-07,
+      "loss": 0.2345,
+      "step": 9760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.99100161640238e-07,
+      "loss": 0.2099,
+      "step": 9761
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981668015119734e-07,
+      "loss": 0.2237,
+      "step": 9762
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.972342926383655e-07,
+      "loss": 0.2127,
+      "step": 9763
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.963026351029565e-07,
+      "loss": 0.2257,
+      "step": 9764
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.953718289892106e-07,
+      "loss": 0.2238,
+      "step": 9765
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.94441874380518e-07,
+      "loss": 0.1696,
+      "step": 9766
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.935127713601928e-07,
+      "loss": 0.2285,
+      "step": 9767
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.925845200114698e-07,
+      "loss": 0.1925,
+      "step": 9768
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916571204175113e-07,
+      "loss": 0.2097,
+      "step": 9769
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.907305726614009e-07,
+      "loss": 0.2181,
+      "step": 9770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898048768261476e-07,
+      "loss": 0.2003,
+      "step": 9771
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.888800329946808e-07,
+      "loss": 0.1936,
+      "step": 9772
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.879560412498585e-07,
+      "loss": 0.2126,
+      "step": 9773
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.870329016744579e-07,
+      "loss": 0.1864,
+      "step": 9774
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.861106143511829e-07,
+      "loss": 0.1757,
+      "step": 9775
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.851891793626606e-07,
+      "loss": 0.2027,
+      "step": 9776
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.842685967914384e-07,
+      "loss": 0.2291,
+      "step": 9777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833488667199904e-07,
+      "loss": 0.2168,
+      "step": 9778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.824299892307149e-07,
+      "loss": 0.2048,
+      "step": 9779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.815119644059308e-07,
+      "loss": 0.1857,
+      "step": 9780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805947923278864e-07,
+      "loss": 0.2083,
+      "step": 9781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.796784730787463e-07,
+      "loss": 0.1927,
+      "step": 9782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.787630067406023e-07,
+      "loss": 0.2053,
+      "step": 9783
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778483933954714e-07,
+      "loss": 0.1901,
+      "step": 9784
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769346331252933e-07,
+      "loss": 0.1934,
+      "step": 9785
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7602172601192597e-07,
+      "loss": 0.2408,
+      "step": 9786
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.751096721371573e-07,
+      "loss": 0.1965,
+      "step": 9787
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.741984715826997e-07,
+      "loss": 0.1944,
+      "step": 9788
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7328812443018235e-07,
+      "loss": 0.2079,
+      "step": 9789
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.723786307611655e-07,
+      "loss": 0.1976,
+      "step": 9790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.714699906571263e-07,
+      "loss": 0.2194,
+      "step": 9791
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7056220419946954e-07,
+      "loss": 0.2062,
+      "step": 9792
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6965527146952236e-07,
+      "loss": 0.1806,
+      "step": 9793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6874919254853546e-07,
+      "loss": 0.1858,
+      "step": 9794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.678439675176827e-07,
+      "loss": 0.1977,
+      "step": 9795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6693959645806143e-07,
+      "loss": 0.2165,
+      "step": 9796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.660360794506946e-07,
+      "loss": 0.2201,
+      "step": 9797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6513341657652534e-07,
+      "loss": 0.2084,
+      "step": 9798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.642316079164233e-07,
+      "loss": 0.2474,
+      "step": 9799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6333065355117836e-07,
+      "loss": 0.2104,
+      "step": 9800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.62430553561507e-07,
+      "loss": 0.2035,
+      "step": 9801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.61531308028047e-07,
+      "loss": 0.2005,
+      "step": 9802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.606329170313606e-07,
+      "loss": 0.2118,
+      "step": 9803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5973538065193223e-07,
+      "loss": 0.2016,
+      "step": 9804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.588386989701732e-07,
+      "loss": 0.219,
+      "step": 9805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5794287206641477e-07,
+      "loss": 0.2236,
+      "step": 9806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.570479000209116e-07,
+      "loss": 0.1926,
+      "step": 9807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5615378291384513e-07,
+      "loss": 0.2223,
+      "step": 9808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.552605208253147e-07,
+      "loss": 0.2371,
+      "step": 9809
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.543681138353495e-07,
+      "loss": 0.2199,
+      "step": 9810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.534765620238968e-07,
+      "loss": 0.224,
+      "step": 9811
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.525858654708304e-07,
+      "loss": 0.1878,
+      "step": 9812
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.516960242559454e-07,
+      "loss": 0.2224,
+      "step": 9813
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5080703845896247e-07,
+      "loss": 0.1906,
+      "step": 9814
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.499189081595245e-07,
+      "loss": 0.1969,
+      "step": 9815
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.490316334371969e-07,
+      "loss": 0.1968,
+      "step": 9816
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4814521437147043e-07,
+      "loss": 0.1913,
+      "step": 9817
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4725965104175597e-07,
+      "loss": 0.2246,
+      "step": 9818
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4637494352739006e-07,
+      "loss": 0.2292,
+      "step": 9819
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4549109190763497e-07,
+      "loss": 0.2031,
+      "step": 9820
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4460809626167165e-07,
+      "loss": 0.2107,
+      "step": 9821
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4372595666860806e-07,
+      "loss": 0.2299,
+      "step": 9822
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4284467320746984e-07,
+      "loss": 0.2197,
+      "step": 9823
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.419642459572138e-07,
+      "loss": 0.2154,
+      "step": 9824
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4108467499671345e-07,
+      "loss": 0.2015,
+      "step": 9825
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4020596040477037e-07,
+      "loss": 0.2444,
+      "step": 9826
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.393281022601037e-07,
+      "loss": 0.204,
+      "step": 9827
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3845110064136385e-07,
+      "loss": 0.1996,
+      "step": 9828
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3757495562711693e-07,
+      "loss": 0.2146,
+      "step": 9829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.366996672958568e-07,
+      "loss": 0.2158,
+      "step": 9830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3582523572600065e-07,
+      "loss": 0.2135,
+      "step": 9831
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.349516609958848e-07,
+      "loss": 0.2068,
+      "step": 9832
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3407894318377106e-07,
+      "loss": 0.2237,
+      "step": 9833
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3320708236784693e-07,
+      "loss": 0.2021,
+      "step": 9834
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3233607862622097e-07,
+      "loss": 0.1951,
+      "step": 9835
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.314659320369241e-07,
+      "loss": 0.1998,
+      "step": 9836
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.305966426779118e-07,
+      "loss": 0.2271,
+      "step": 9837
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.297282106270617e-07,
+      "loss": 0.1906,
+      "step": 9838
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.288606359621761e-07,
+      "loss": 0.2105,
+      "step": 9839
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2799391876098163e-07,
+      "loss": 0.199,
+      "step": 9840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2712805910112175e-07,
+      "loss": 0.2108,
+      "step": 9841
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2626305706017113e-07,
+      "loss": 0.2444,
+      "step": 9842
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2539891271562103e-07,
+      "loss": 0.2127,
+      "step": 9843
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2453562614489185e-07,
+      "loss": 0.2302,
+      "step": 9844
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2367319742532164e-07,
+      "loss": 0.2342,
+      "step": 9845
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.228116266341764e-07,
+      "loss": 0.1989,
+      "step": 9846
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2195091384863995e-07,
+      "loss": 0.1902,
+      "step": 9847
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2109105914582503e-07,
+      "loss": 0.2091,
+      "step": 9848
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.202320626027645e-07,
+      "loss": 0.2122,
+      "step": 9849
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1937392429641343e-07,
+      "loss": 0.2183,
+      "step": 9850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1851664430365037e-07,
+      "loss": 0.2066,
+      "step": 9851
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1766022270127717e-07,
+      "loss": 0.1939,
+      "step": 9852
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1680465956602356e-07,
+      "loss": 0.2166,
+      "step": 9853
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.159499549745349e-07,
+      "loss": 0.2078,
+      "step": 9854
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.150961090033823e-07,
+      "loss": 0.2162,
+      "step": 9855
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.142431217290621e-07,
+      "loss": 0.2224,
+      "step": 9856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.133909932279911e-07,
+      "loss": 0.216,
+      "step": 9857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.125397235765127e-07,
+      "loss": 0.2057,
+      "step": 9858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.116893128508859e-07,
+      "loss": 0.2051,
+      "step": 9859
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.108397611273018e-07,
+      "loss": 0.2238,
+      "step": 9860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.099910684818698e-07,
+      "loss": 0.1982,
+      "step": 9861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0914323499062213e-07,
+      "loss": 0.1872,
+      "step": 9862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.082962607295171e-07,
+      "loss": 0.2127,
+      "step": 9863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0745014577442953e-07,
+      "loss": 0.2088,
+      "step": 9864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0660489020116435e-07,
+      "loss": 0.2114,
+      "step": 9865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.057604940854465e-07,
+      "loss": 0.2183,
+      "step": 9866
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0491695750292325e-07,
+      "loss": 0.208,
+      "step": 9867
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0407428052916753e-07,
+      "loss": 0.2016,
+      "step": 9868
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0323246323967224e-07,
+      "loss": 0.2205,
+      "step": 9869
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0239150570985374e-07,
+      "loss": 0.2273,
+      "step": 9870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0155140801505396e-07,
+      "loss": 0.2384,
+      "step": 9871
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.007121702305361e-07,
+      "loss": 0.2064,
+      "step": 9872
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.998737924314833e-07,
+      "loss": 0.1823,
+      "step": 9873
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9903627469300656e-07,
+      "loss": 0.2139,
+      "step": 9874
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.981996170901381e-07,
+      "loss": 0.1978,
+      "step": 9875
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.973638196978324e-07,
+      "loss": 0.2191,
+      "step": 9876
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9652888259096635e-07,
+      "loss": 0.2029,
+      "step": 9877
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9569480584434217e-07,
+      "loss": 0.2037,
+      "step": 9878
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9486158953268237e-07,
+      "loss": 0.2355,
+      "step": 9879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9402923373063396e-07,
+      "loss": 0.201,
+      "step": 9880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9319773851276834e-07,
+      "loss": 0.2169,
+      "step": 9881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.923671039535748e-07,
+      "loss": 0.1955,
+      "step": 9882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9153733012746943e-07,
+      "loss": 0.2274,
+      "step": 9883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.907084171087894e-07,
+      "loss": 0.198,
+      "step": 9884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8988036497179973e-07,
+      "loss": 0.2084,
+      "step": 9885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.890531737906811e-07,
+      "loss": 0.2246,
+      "step": 9886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.882268436395398e-07,
+      "loss": 0.2245,
+      "step": 9887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.874013745924077e-07,
+      "loss": 0.1944,
+      "step": 9888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.865767667232356e-07,
+      "loss": 0.1952,
+      "step": 9889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.857530201059012e-07,
+      "loss": 0.2087,
+      "step": 9890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8493013481419984e-07,
+      "loss": 0.1835,
+      "step": 9891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.841081109218514e-07,
+      "loss": 0.1897,
+      "step": 9892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.832869485025048e-07,
+      "loss": 0.187,
+      "step": 9893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.824666476297245e-07,
+      "loss": 0.2058,
+      "step": 9894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8164720837699733e-07,
+      "loss": 0.216,
+      "step": 9895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.808286308177389e-07,
+      "loss": 0.2021,
+      "step": 9896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.800109150252829e-07,
+      "loss": 0.2344,
+      "step": 9897
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7919406107288725e-07,
+      "loss": 0.217,
+      "step": 9898
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.783780690337335e-07,
+      "loss": 0.2003,
+      "step": 9899
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.775629389809232e-07,
+      "loss": 0.2004,
+      "step": 9900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.767486709874846e-07,
+      "loss": 0.2161,
+      "step": 9901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7593526512636704e-07,
+      "loss": 0.1904,
+      "step": 9902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7512272147044004e-07,
+      "loss": 0.1962,
+      "step": 9903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7431104009250096e-07,
+      "loss": 0.2075,
+      "step": 9904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7350022106526385e-07,
+      "loss": 0.2098,
+      "step": 9905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.726902644613706e-07,
+      "loss": 0.2279,
+      "step": 9906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7188117035338424e-07,
+      "loss": 0.2034,
+      "step": 9907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7107293881378904e-07,
+      "loss": 0.1856,
+      "step": 9908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7026556991499486e-07,
+      "loss": 0.2266,
+      "step": 9909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.694590637293316e-07,
+      "loss": 0.2098,
+      "step": 9910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.686534203290526e-07,
+      "loss": 0.2154,
+      "step": 9911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6784863978633455e-07,
+      "loss": 0.2089,
+      "step": 9912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.670447221732787e-07,
+      "loss": 0.2033,
+      "step": 9913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6624166756190296e-07,
+      "loss": 0.21,
+      "step": 9914
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6543947602415307e-07,
+      "loss": 0.2223,
+      "step": 9915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.646381476318972e-07,
+      "loss": 0.1999,
+      "step": 9916
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6383768245692453e-07,
+      "loss": 0.2349,
+      "step": 9917
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.630380805709477e-07,
+      "loss": 0.2107,
+      "step": 9918
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6223934204560165e-07,
+      "loss": 0.2043,
+      "step": 9919
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.614414669524435e-07,
+      "loss": 0.1829,
+      "step": 9920
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606444553629551e-07,
+      "loss": 0.2048,
+      "step": 9921
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5984830734853927e-07,
+      "loss": 0.1968,
+      "step": 9922
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5905302298052e-07,
+      "loss": 0.1987,
+      "step": 9923
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5825860233014596e-07,
+      "loss": 0.1842,
+      "step": 9924
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.574650454685902e-07,
+      "loss": 0.2082,
+      "step": 9925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5667235246694463e-07,
+      "loss": 0.2052,
+      "step": 9926
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5588052339622594e-07,
+      "loss": 0.1853,
+      "step": 9927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.550895583273717e-07,
+      "loss": 0.1927,
+      "step": 9928
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.542994573312442e-07,
+      "loss": 0.2263,
+      "step": 9929
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.535102204786278e-07,
+      "loss": 0.2117,
+      "step": 9930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5272184784022834e-07,
+      "loss": 0.2062,
+      "step": 9931
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5193433948667476e-07,
+      "loss": 0.2485,
+      "step": 9932
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5114769548851955e-07,
+      "loss": 0.2242,
+      "step": 9933
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5036191591623637e-07,
+      "loss": 0.1979,
+      "step": 9934
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.495770008402233e-07,
+      "loss": 0.1966,
+      "step": 9935
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.487929503307974e-07,
+      "loss": 0.2268,
+      "step": 9936
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4800976445820146e-07,
+      "loss": 0.2118,
+      "step": 9937
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.472274432926004e-07,
+      "loss": 0.2333,
+      "step": 9938
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.464459869040804e-07,
+      "loss": 0.2006,
+      "step": 9939
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.45665395362651e-07,
+      "loss": 0.2207,
+      "step": 9940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4488566873824514e-07,
+      "loss": 0.1998,
+      "step": 9941
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4410680710071697e-07,
+      "loss": 0.2041,
+      "step": 9942
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4332881051984293e-07,
+      "loss": 0.2032,
+      "step": 9943
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.425516790653227e-07,
+      "loss": 0.2293,
+      "step": 9944
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.417754128067796e-07,
+      "loss": 0.2205,
+      "step": 9945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4100001181375687e-07,
+      "loss": 0.2305,
+      "step": 9946
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.402254761557211e-07,
+      "loss": 0.228,
+      "step": 9947
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3945180590206237e-07,
+      "loss": 0.2412,
+      "step": 9948
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3867900112209195e-07,
+      "loss": 0.1972,
+      "step": 9949
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3790706188504553e-07,
+      "loss": 0.2046,
+      "step": 9950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.371359882600789e-07,
+      "loss": 0.2223,
+      "step": 9951
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3636578031627234e-07,
+      "loss": 0.1899,
+      "step": 9952
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.355964381226262e-07,
+      "loss": 0.1882,
+      "step": 9953
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3482796174806763e-07,
+      "loss": 0.2285,
+      "step": 9954
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3406035126143933e-07,
+      "loss": 0.1829,
+      "step": 9955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3329360673151067e-07,
+      "loss": 0.2094,
+      "step": 9956
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3252772822697565e-07,
+      "loss": 0.2112,
+      "step": 9957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3176271581644934e-07,
+      "loss": 0.2032,
+      "step": 9958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3099856956846365e-07,
+      "loss": 0.1906,
+      "step": 9959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.302352895514793e-07,
+      "loss": 0.2052,
+      "step": 9960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.294728758338772e-07,
+      "loss": 0.2044,
+      "step": 9961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.287113284839616e-07,
+      "loss": 0.2225,
+      "step": 9962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.27950647569959e-07,
+      "loss": 0.2136,
+      "step": 9963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2719083316001375e-07,
+      "loss": 0.2289,
+      "step": 9964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.264318853222015e-07,
+      "loss": 0.2158,
+      "step": 9965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.256738041245111e-07,
+      "loss": 0.1961,
+      "step": 9966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.249165896348616e-07,
+      "loss": 0.2198,
+      "step": 9967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2416024192108765e-07,
+      "loss": 0.2243,
+      "step": 9968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.234047610509505e-07,
+      "loss": 0.2123,
+      "step": 9969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2265014709213283e-07,
+      "loss": 0.2086,
+      "step": 9970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.218964001122382e-07,
+      "loss": 0.2126,
+      "step": 9971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2114352017879493e-07,
+      "loss": 0.2176,
+      "step": 9972
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2039150735925117e-07,
+      "loss": 0.2313,
+      "step": 9973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.196403617209787e-07,
+      "loss": 0.2312,
+      "step": 9974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.188900833312736e-07,
+      "loss": 0.2178,
+      "step": 9975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.18140672257351e-07,
+      "loss": 0.2071,
+      "step": 9976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1739212856634725e-07,
+      "loss": 0.222,
+      "step": 9977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1664445232532647e-07,
+      "loss": 0.2157,
+      "step": 9978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.158976436012695e-07,
+      "loss": 0.2078,
+      "step": 9979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1515170246108285e-07,
+      "loss": 0.1823,
+      "step": 9980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1440662897159414e-07,
+      "loss": 0.2038,
+      "step": 9981
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.136624231995533e-07,
+      "loss": 0.2259,
+      "step": 9982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.129190852116326e-07,
+      "loss": 0.2095,
+      "step": 9983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1217661507442654e-07,
+      "loss": 0.1835,
+      "step": 9984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.114350128544519e-07,
+      "loss": 0.2187,
+      "step": 9985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.106942786181488e-07,
+      "loss": 0.2201,
+      "step": 9986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0995441243187654e-07,
+      "loss": 0.2239,
+      "step": 9987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.092154143619186e-07,
+      "loss": 0.2334,
+      "step": 9988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.08477284474481e-07,
+      "loss": 0.1986,
+      "step": 9989
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0774002283569414e-07,
+      "loss": 0.2248,
+      "step": 9990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.070036295116052e-07,
+      "loss": 0.2006,
+      "step": 9991
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.062681045681881e-07,
+      "loss": 0.1998,
+      "step": 9992
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0553344807133566e-07,
+      "loss": 0.2076,
+      "step": 9993
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.047996600868675e-07,
+      "loss": 0.2161,
+      "step": 9994
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.040667406805198e-07,
+      "loss": 0.1903,
+      "step": 9995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0333468991795456e-07,
+      "loss": 0.2126,
+      "step": 9996
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.026035078647549e-07,
+      "loss": 0.186,
+      "step": 9997
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0187319458642726e-07,
+      "loss": 0.2159,
+      "step": 9998
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.011437501483994e-07,
+      "loss": 0.224,
+      "step": 9999
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0041517461602e-07,
+      "loss": 0.1948,
+      "step": 10000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.996874680545603e-07,
+      "loss": 0.193,
+      "step": 10001
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9896063052921584e-07,
+      "loss": 0.2105,
+      "step": 10002
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9823466210510357e-07,
+      "loss": 0.211,
+      "step": 10003
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.975095628472602e-07,
+      "loss": 0.2273,
+      "step": 10004
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9678533282064714e-07,
+      "loss": 0.2241,
+      "step": 10005
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.96061972090147e-07,
+      "loss": 0.2043,
+      "step": 10006
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.953394807205634e-07,
+      "loss": 0.2209,
+      "step": 10007
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9461785877662575e-07,
+      "loss": 0.2073,
+      "step": 10008
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9389710632298117e-07,
+      "loss": 0.2222,
+      "step": 10009
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.931772234242003e-07,
+      "loss": 0.2041,
+      "step": 10010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9245821014477813e-07,
+      "loss": 0.1963,
+      "step": 10011
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.917400665491288e-07,
+      "loss": 0.21,
+      "step": 10012
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.910227927015896e-07,
+      "loss": 0.2133,
+      "step": 10013
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9030638866642036e-07,
+      "loss": 0.2264,
+      "step": 10014
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.895908545078019e-07,
+      "loss": 0.2307,
+      "step": 10015
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.888761902898396e-07,
+      "loss": 0.1992,
+      "step": 10016
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8816239607655893e-07,
+      "loss": 0.2193,
+      "step": 10017
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.874494719319043e-07,
+      "loss": 0.2259,
+      "step": 10018
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8673741791974907e-07,
+      "loss": 0.2159,
+      "step": 10019
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8602623410388443e-07,
+      "loss": 0.1985,
+      "step": 10020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.853159205480216e-07,
+      "loss": 0.2124,
+      "step": 10021
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.846064773158008e-07,
+      "loss": 0.2024,
+      "step": 10022
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.838979044707768e-07,
+      "loss": 0.1882,
+      "step": 10023
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.83190202076431e-07,
+      "loss": 0.1947,
+      "step": 10024
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.824833701961649e-07,
+      "loss": 0.198,
+      "step": 10025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.817774088933023e-07,
+      "loss": 0.1955,
+      "step": 10026
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.810723182310904e-07,
+      "loss": 0.2142,
+      "step": 10027
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8036809827269416e-07,
+      "loss": 0.2238,
+      "step": 10028
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7966474908120544e-07,
+      "loss": 0.2301,
+      "step": 10029
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.789622707196371e-07,
+      "loss": 0.1976,
+      "step": 10030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7826066325092327e-07,
+      "loss": 0.2091,
+      "step": 10031
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.775599267379181e-07,
+      "loss": 0.2159,
+      "step": 10032
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.768600612434003e-07,
+      "loss": 0.2242,
+      "step": 10033
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.761610668300696e-07,
+      "loss": 0.2083,
+      "step": 10034
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7546294356054823e-07,
+      "loss": 0.1915,
+      "step": 10035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7476569149738066e-07,
+      "loss": 0.2313,
+      "step": 10036
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.740693107030301e-07,
+      "loss": 0.2191,
+      "step": 10037
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7337380123988676e-07,
+      "loss": 0.2172,
+      "step": 10038
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7267916317025967e-07,
+      "loss": 0.2236,
+      "step": 10039
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.719853965563801e-07,
+      "loss": 0.1973,
+      "step": 10040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7129250146040175e-07,
+      "loss": 0.2047,
+      "step": 10041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7060047794439937e-07,
+      "loss": 0.1973,
+      "step": 10042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6990932607037e-07,
+      "loss": 0.188,
+      "step": 10043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.692190459002342e-07,
+      "loss": 0.2132,
+      "step": 10044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.685296374958324e-07,
+      "loss": 0.1966,
+      "step": 10045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.678411009189286e-07,
+      "loss": 0.2145,
+      "step": 10046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.671534362312056e-07,
+      "loss": 0.2102,
+      "step": 10047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6646664349427197e-07,
+      "loss": 0.2016,
+      "step": 10048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.657807227696574e-07,
+      "loss": 0.2019,
+      "step": 10049
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6509567411880933e-07,
+      "loss": 0.2006,
+      "step": 10050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6441149760310313e-07,
+      "loss": 0.2117,
+      "step": 10051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6372819328383095e-07,
+      "loss": 0.1977,
+      "step": 10052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6304576122221035e-07,
+      "loss": 0.189,
+      "step": 10053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6236420147937816e-07,
+      "loss": 0.2249,
+      "step": 10054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.616835141163965e-07,
+      "loss": 0.1943,
+      "step": 10055
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.610036991942444e-07,
+      "loss": 0.1954,
+      "step": 10056
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6032475677382763e-07,
+      "loss": 0.2257,
+      "step": 10057
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.596466869159708e-07,
+      "loss": 0.2179,
+      "step": 10058
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.58969489681421e-07,
+      "loss": 0.2273,
+      "step": 10059
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.582931651308462e-07,
+      "loss": 0.2094,
+      "step": 10060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.576177133248381e-07,
+      "loss": 0.1981,
+      "step": 10061
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.569431343239104e-07,
+      "loss": 0.2377,
+      "step": 10062
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.562694281884992e-07,
+      "loss": 0.2098,
+      "step": 10063
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.555965949789563e-07,
+      "loss": 0.2078,
+      "step": 10064
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5492463475556227e-07,
+      "loss": 0.195,
+      "step": 10065
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.542535475785168e-07,
+      "loss": 0.2078,
+      "step": 10066
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.535833335079407e-07,
+      "loss": 0.2171,
+      "step": 10067
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5291399260387926e-07,
+      "loss": 0.2191,
+      "step": 10068
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.522455249262956e-07,
+      "loss": 0.2145,
+      "step": 10069
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.515779305350785e-07,
+      "loss": 0.1812,
+      "step": 10070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5091120949003567e-07,
+      "loss": 0.182,
+      "step": 10071
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.502453618508993e-07,
+      "loss": 0.2043,
+      "step": 10072
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4958038767731954e-07,
+      "loss": 0.2045,
+      "step": 10073
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.489162870288708e-07,
+      "loss": 0.22,
+      "step": 10074
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.48253059965049e-07,
+      "loss": 0.2522,
+      "step": 10075
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.475907065452721e-07,
+      "loss": 0.1981,
+      "step": 10076
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4692922682887923e-07,
+      "loss": 0.2047,
+      "step": 10077
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.462686208751308e-07,
+      "loss": 0.2306,
+      "step": 10078
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4560888874320955e-07,
+      "loss": 0.203,
+      "step": 10079
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.449500304922203e-07,
+      "loss": 0.1797,
+      "step": 10080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.442920461811904e-07,
+      "loss": 0.1942,
+      "step": 10081
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4363493586906483e-07,
+      "loss": 0.2156,
+      "step": 10082
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4297869961471544e-07,
+      "loss": 0.2076,
+      "step": 10083
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.423233374769318e-07,
+      "loss": 0.2149,
+      "step": 10084
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4166884951442706e-07,
+      "loss": 0.2059,
+      "step": 10085
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.410152357858375e-07,
+      "loss": 0.1818,
+      "step": 10086
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4036249634971753e-07,
+      "loss": 0.1982,
+      "step": 10087
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3971063126454584e-07,
+      "loss": 0.1915,
+      "step": 10088
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3905964058872135e-07,
+      "loss": 0.2108,
+      "step": 10089
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.384095243805673e-07,
+      "loss": 0.199,
+      "step": 10090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3776028269832384e-07,
+      "loss": 0.2104,
+      "step": 10091
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3711191560015667e-07,
+      "loss": 0.2025,
+      "step": 10092
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3646442314415153e-07,
+      "loss": 0.2344,
+      "step": 10093
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3581780538831755e-07,
+      "loss": 0.1902,
+      "step": 10094
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.351720623905851e-07,
+      "loss": 0.1964,
+      "step": 10095
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.345271942088023e-07,
+      "loss": 0.1911,
+      "step": 10096
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3388320090074412e-07,
+      "loss": 0.2165,
+      "step": 10097
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3324008252410325e-07,
+      "loss": 0.2405,
+      "step": 10098
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3259783913649804e-07,
+      "loss": 0.2048,
+      "step": 10099
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3195647079546468e-07,
+      "loss": 0.2013,
+      "step": 10100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3131597755845946e-07,
+      "loss": 0.2139,
+      "step": 10101
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.306763594828687e-07,
+      "loss": 0.212,
+      "step": 10102
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.30037616625991e-07,
+      "loss": 0.203,
+      "step": 10103
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.293997490450528e-07,
+      "loss": 0.2089,
+      "step": 10104
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2876275679719838e-07,
+      "loss": 0.2104,
+      "step": 10105
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2812663993949435e-07,
+      "loss": 0.2029,
+      "step": 10106
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2749139852892954e-07,
+      "loss": 0.1847,
+      "step": 10107
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.268570326224151e-07,
+      "loss": 0.2039,
+      "step": 10108
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2622354227678335e-07,
+      "loss": 0.2046,
+      "step": 10109
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2559092754878552e-07,
+      "loss": 0.202,
+      "step": 10110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2495918849509856e-07,
+      "loss": 0.2113,
+      "step": 10111
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2432832517231939e-07,
+      "loss": 0.2017,
+      "step": 10112
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2369833763696502e-07,
+      "loss": 0.2016,
+      "step": 10113
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2306922594547365e-07,
+      "loss": 0.2181,
+      "step": 10114
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2244099015420906e-07,
+      "loss": 0.2101,
+      "step": 10115
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.218136303194529e-07,
+      "loss": 0.2031,
+      "step": 10116
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2118714649740912e-07,
+      "loss": 0.2065,
+      "step": 10117
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2056153874420283e-07,
+      "loss": 0.2041,
+      "step": 10118
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.199368071158825e-07,
+      "loss": 0.2414,
+      "step": 10119
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.193129516684167e-07,
+      "loss": 0.1913,
+      "step": 10120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.186899724576952e-07,
+      "loss": 0.2119,
+      "step": 10121
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1806786953953107e-07,
+      "loss": 0.2143,
+      "step": 10122
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1744664296965534e-07,
+      "loss": 0.1869,
+      "step": 10123
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.168262928037246e-07,
+      "loss": 0.2196,
+      "step": 10124
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1620681909731324e-07,
+      "loss": 0.22,
+      "step": 10125
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.155882219059202e-07,
+      "loss": 0.2143,
+      "step": 10126
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1497050128496567e-07,
+      "loss": 0.2092,
+      "step": 10127
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1435365728978862e-07,
+      "loss": 0.2173,
+      "step": 10128
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1373768997565047e-07,
+      "loss": 0.1989,
+      "step": 10129
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1312259939773705e-07,
+      "loss": 0.2297,
+      "step": 10130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1250838561115318e-07,
+      "loss": 0.2182,
+      "step": 10131
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.118950486709237e-07,
+      "loss": 0.2135,
+      "step": 10132
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.112825886319958e-07,
+      "loss": 0.2008,
+      "step": 10133
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1067100554924225e-07,
+      "loss": 0.1815,
+      "step": 10134
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1006029947745033e-07,
+      "loss": 0.2272,
+      "step": 10135
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0945047047133627e-07,
+      "loss": 0.1986,
+      "step": 10136
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.088415185855297e-07,
+      "loss": 0.2118,
+      "step": 10137
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0823344387458698e-07,
+      "loss": 0.2051,
+      "step": 10138
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0762624639298568e-07,
+      "loss": 0.2178,
+      "step": 10139
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0701992619512334e-07,
+      "loss": 0.1938,
+      "step": 10140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0641448333531765e-07,
+      "loss": 0.2301,
+      "step": 10141
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.058099178678108e-07,
+      "loss": 0.2051,
+      "step": 10142
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0520622984676386e-07,
+      "loss": 0.2231,
+      "step": 10143
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0460341932626138e-07,
+      "loss": 0.2226,
+      "step": 10144
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0400148636030904e-07,
+      "loss": 0.1951,
+      "step": 10145
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0340043100283146e-07,
+      "loss": 0.2037,
+      "step": 10146
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0280025330767673e-07,
+      "loss": 0.212,
+      "step": 10147
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.022009533286129e-07,
+      "loss": 0.2084,
+      "step": 10148
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0160253111933149e-07,
+      "loss": 0.2268,
+      "step": 10149
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0100498673344516e-07,
+      "loss": 0.2095,
+      "step": 10150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0040832022448443e-07,
+      "loss": 0.1966,
+      "step": 10151
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.998125316459054e-07,
+      "loss": 0.211,
+      "step": 10152
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9921762105108432e-07,
+      "loss": 0.2213,
+      "step": 10153
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9862358849331853e-07,
+      "loss": 0.2091,
+      "step": 10154
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9803043402582322e-07,
+      "loss": 0.2107,
+      "step": 10155
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9743815770174145e-07,
+      "loss": 0.1958,
+      "step": 10156
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9684675957413414e-07,
+      "loss": 0.216,
+      "step": 10157
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9625623969598106e-07,
+      "loss": 0.2212,
+      "step": 10158
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9566659812018995e-07,
+      "loss": 0.1761,
+      "step": 10159
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.950778348995841e-07,
+      "loss": 0.2071,
+      "step": 10160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9448995008690796e-07,
+      "loss": 0.2149,
+      "step": 10161
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9390294373483276e-07,
+      "loss": 0.2088,
+      "step": 10162
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9331681589594532e-07,
+      "loss": 0.2052,
+      "step": 10163
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9273156662275582e-07,
+      "loss": 0.2065,
+      "step": 10164
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.2189,
+      "step": 10165
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9156370398311863e-07,
+      "loss": 0.2119,
+      "step": 10166
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9098109072129945e-07,
+      "loss": 0.2106,
+      "step": 10167
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9039935623443307e-07,
+      "loss": 0.1933,
+      "step": 10168
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8981850057463446e-07,
+      "loss": 0.2122,
+      "step": 10169
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8923852379394424e-07,
+      "loss": 0.2019,
+      "step": 10170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8865942594431974e-07,
+      "loss": 0.2162,
+      "step": 10171
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8808120707764276e-07,
+      "loss": 0.1899,
+      "step": 10172
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.87503867245713e-07,
+      "loss": 0.1828,
+      "step": 10173
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8692740650025576e-07,
+      "loss": 0.2161,
+      "step": 10174
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8635182489291415e-07,
+      "loss": 0.2171,
+      "step": 10175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.857771224752536e-07,
+      "loss": 0.1779,
+      "step": 10176
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8520329929876291e-07,
+      "loss": 0.1975,
+      "step": 10177
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8463035541484653e-07,
+      "loss": 0.2279,
+      "step": 10178
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8405829087483563e-07,
+      "loss": 0.1986,
+      "step": 10179
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8348710572997918e-07,
+      "loss": 0.2193,
+      "step": 10180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8291680003145074e-07,
+      "loss": 0.2103,
+      "step": 10181
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8234737383034162e-07,
+      "loss": 0.2077,
+      "step": 10182
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.817788271776677e-07,
+      "loss": 0.1809,
+      "step": 10183
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8121116012436158e-07,
+      "loss": 0.2221,
+      "step": 10184
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8064437272128144e-07,
+      "loss": 0.2399,
+      "step": 10185
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8007846501920557e-07,
+      "loss": 0.2043,
+      "step": 10186
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7951343706883005e-07,
+      "loss": 0.2036,
+      "step": 10187
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.789492889207778e-07,
+      "loss": 0.2238,
+      "step": 10188
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.783860206255883e-07,
+      "loss": 0.2356,
+      "step": 10189
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7782363223372234e-07,
+      "loss": 0.2043,
+      "step": 10190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.772621237955685e-07,
+      "loss": 0.1918,
+      "step": 10191
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7670149536142656e-07,
+      "loss": 0.2045,
+      "step": 10192
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.761417469815241e-07,
+      "loss": 0.2151,
+      "step": 10193
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7558287870600876e-07,
+      "loss": 0.204,
+      "step": 10194
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7502489058494832e-07,
+      "loss": 0.2122,
+      "step": 10195
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7446778266833165e-07,
+      "loss": 0.2265,
+      "step": 10196
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.739115550060688e-07,
+      "loss": 0.2033,
+      "step": 10197
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.733562076479911e-07,
+      "loss": 0.2081,
+      "step": 10198
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7280174064385314e-07,
+      "loss": 0.2114,
+      "step": 10199
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.722481540433285e-07,
+      "loss": 0.2148,
+      "step": 10200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.716954478960109e-07,
+      "loss": 0.1934,
+      "step": 10201
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7114362225141623e-07,
+      "loss": 0.2188,
+      "step": 10202
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7059267715898275e-07,
+      "loss": 0.2102,
+      "step": 10203
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7004261266806987e-07,
+      "loss": 0.1936,
+      "step": 10204
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6949342882795262e-07,
+      "loss": 0.1925,
+      "step": 10205
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6894512568783717e-07,
+      "loss": 0.2099,
+      "step": 10206
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.683977032968409e-07,
+      "loss": 0.1941,
+      "step": 10207
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6785116170400907e-07,
+      "loss": 0.1944,
+      "step": 10208
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6730550095830466e-07,
+      "loss": 0.2051,
+      "step": 10209
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6676072110861198e-07,
+      "loss": 0.2076,
+      "step": 10210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6621682220373747e-07,
+      "loss": 0.2173,
+      "step": 10211
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6567380429240775e-07,
+      "loss": 0.1838,
+      "step": 10212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.651316674232717e-07,
+      "loss": 0.2067,
+      "step": 10213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.645904116448993e-07,
+      "loss": 0.1922,
+      "step": 10214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6405003700577848e-07,
+      "loss": 0.2025,
+      "step": 10215
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.635105435543227e-07,
+      "loss": 0.2107,
+      "step": 10216
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6297193133886337e-07,
+      "loss": 0.2119,
+      "step": 10217
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6243420040765511e-07,
+      "loss": 0.2362,
+      "step": 10218
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.618973508088706e-07,
+      "loss": 0.1874,
+      "step": 10219
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6136138259060573e-07,
+      "loss": 0.1876,
+      "step": 10220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6082629580087772e-07,
+      "loss": 0.2362,
+      "step": 10221
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6029209048762485e-07,
+      "loss": 0.2032,
+      "step": 10222
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5975876669870328e-07,
+      "loss": 0.2091,
+      "step": 10223
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.592263244818959e-07,
+      "loss": 0.1983,
+      "step": 10224
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5869476388490124e-07,
+      "loss": 0.1835,
+      "step": 10225
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.581640849553412e-07,
+      "loss": 0.2173,
+      "step": 10226
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5763428774075996e-07,
+      "loss": 0.2274,
+      "step": 10227
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5710537228861843e-07,
+      "loss": 0.213,
+      "step": 10228
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5657733864630432e-07,
+      "loss": 0.2082,
+      "step": 10229
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5605018686111972e-07,
+      "loss": 0.2038,
+      "step": 10230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.555239169802958e-07,
+      "loss": 0.2179,
+      "step": 10231
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5499852905097813e-07,
+      "loss": 0.1939,
+      "step": 10232
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5447402312023573e-07,
+      "loss": 0.2002,
+      "step": 10233
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.539503992350566e-07,
+      "loss": 0.2111,
+      "step": 10234
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.534276574423532e-07,
+      "loss": 0.1994,
+      "step": 10235
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5290579778895808e-07,
+      "loss": 0.2051,
+      "step": 10236
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5238482032162162e-07,
+      "loss": 0.2188,
+      "step": 10237
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5186472508701643e-07,
+      "loss": 0.2021,
+      "step": 10238
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5134551213174087e-07,
+      "loss": 0.1765,
+      "step": 10239
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.508271815023077e-07,
+      "loss": 0.1874,
+      "step": 10240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.503097332451553e-07,
+      "loss": 0.2035,
+      "step": 10241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4979316740663886e-07,
+      "loss": 0.2275,
+      "step": 10242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.49277484033038e-07,
+      "loss": 0.2033,
+      "step": 10243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4876268317055242e-07,
+      "loss": 0.1999,
+      "step": 10244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.482487648653008e-07,
+      "loss": 0.2021,
+      "step": 10245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4773572916332523e-07,
+      "loss": 0.2107,
+      "step": 10246
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.472235761105878e-07,
+      "loss": 0.209,
+      "step": 10247
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.467123057529707e-07,
+      "loss": 0.1971,
+      "step": 10248
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.462019181362795e-07,
+      "loss": 0.2147,
+      "step": 10249
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4569241330623872e-07,
+      "loss": 0.1807,
+      "step": 10250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4518379130849302e-07,
+      "loss": 0.2149,
+      "step": 10251
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4467605218860924e-07,
+      "loss": 0.2193,
+      "step": 10252
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4416919599207547e-07,
+      "loss": 0.2003,
+      "step": 10253
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.436632227642998e-07,
+      "loss": 0.2018,
+      "step": 10254
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.431581325506115e-07,
+      "loss": 0.2499,
+      "step": 10255
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4265392539626112e-07,
+      "loss": 0.181,
+      "step": 10256
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4215060134642022e-07,
+      "loss": 0.2018,
+      "step": 10257
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4164816044618058e-07,
+      "loss": 0.2118,
+      "step": 10258
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4114660274055503e-07,
+      "loss": 0.214,
+      "step": 10259
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4064592827447764e-07,
+      "loss": 0.2052,
+      "step": 10260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4014613709280144e-07,
+      "loss": 0.2323,
+      "step": 10261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3964722924030395e-07,
+      "loss": 0.2044,
+      "step": 10262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3914920476168048e-07,
+      "loss": 0.2281,
+      "step": 10263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3865206370154872e-07,
+      "loss": 0.2316,
+      "step": 10264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3815580610444636e-07,
+      "loss": 0.2086,
+      "step": 10265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3766043201483336e-07,
+      "loss": 0.1909,
+      "step": 10266
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3716594147708874e-07,
+      "loss": 0.2073,
+      "step": 10267
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3667233453551365e-07,
+      "loss": 0.2141,
+      "step": 10268
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3617961123432831e-07,
+      "loss": 0.2021,
+      "step": 10269
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3568777161767523e-07,
+      "loss": 0.1977,
+      "step": 10270
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3519681572961907e-07,
+      "loss": 0.2023,
+      "step": 10271
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.347067436141425e-07,
+      "loss": 0.2102,
+      "step": 10272
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3421755531515258e-07,
+      "loss": 0.1928,
+      "step": 10273
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3372925087647094e-07,
+      "loss": 0.191,
+      "step": 10274
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.332418303418481e-07,
+      "loss": 0.2351,
+      "step": 10275
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.327552937549481e-07,
+      "loss": 0.2113,
+      "step": 10276
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3226964115936046e-07,
+      "loss": 0.2295,
+      "step": 10277
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3178487259859485e-07,
+      "loss": 0.2066,
+      "step": 10278
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3130098811607983e-07,
+      "loss": 0.2088,
+      "step": 10279
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.308179877551652e-07,
+      "loss": 0.2216,
+      "step": 10280
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3033587155912408e-07,
+      "loss": 0.2203,
+      "step": 10281
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2985463957114752e-07,
+      "loss": 0.2307,
+      "step": 10282
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2937429183434768e-07,
+      "loss": 0.2137,
+      "step": 10283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2889482839176015e-07,
+      "loss": 0.1863,
+      "step": 10284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2841624928633722e-07,
+      "loss": 0.2053,
+      "step": 10285
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.279385545609546e-07,
+      "loss": 0.2088,
+      "step": 10286
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.274617442584092e-07,
+      "loss": 0.1793,
+      "step": 10287
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2698581842141567e-07,
+      "loss": 0.1934,
+      "step": 10288
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.265107770926144e-07,
+      "loss": 0.2132,
+      "step": 10289
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2603662031456133e-07,
+      "loss": 0.2025,
+      "step": 10290
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2556334812973693e-07,
+      "loss": 0.2128,
+      "step": 10291
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.250909605805395e-07,
+      "loss": 0.2408,
+      "step": 10292
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2461945770929073e-07,
+      "loss": 0.1894,
+      "step": 10293
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2414883955823133e-07,
+      "loss": 0.2173,
+      "step": 10294
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2367910616952305e-07,
+      "loss": 0.2087,
+      "step": 10295
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2321025758524784e-07,
+      "loss": 0.2123,
+      "step": 10296
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.22742293847411e-07,
+      "loss": 0.1977,
+      "step": 10297
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.222752149979356e-07,
+      "loss": 0.2266,
+      "step": 10298
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.21809021078666e-07,
+      "loss": 0.2071,
+      "step": 10299
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2134371213136985e-07,
+      "loss": 0.1986,
+      "step": 10300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2087928819773165e-07,
+      "loss": 0.1946,
+      "step": 10301
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2041574931935696e-07,
+      "loss": 0.2118,
+      "step": 10302
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.19953095537777e-07,
+      "loss": 0.2109,
+      "step": 10303
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1949132689443975e-07,
+      "loss": 0.1955,
+      "step": 10304
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1903044343071213e-07,
+      "loss": 0.195,
+      "step": 10305
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1857044518788441e-07,
+      "loss": 0.1817,
+      "step": 10306
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1811133220716808e-07,
+      "loss": 0.1894,
+      "step": 10307
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.176531045296947e-07,
+      "loss": 0.1858,
+      "step": 10308
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1719576219651585e-07,
+      "loss": 0.2116,
+      "step": 10309
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1673930524860211e-07,
+      "loss": 0.2095,
+      "step": 10310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1628373372684965e-07,
+      "loss": 0.2165,
+      "step": 10311
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1582904767207025e-07,
+      "loss": 0.1845,
+      "step": 10312
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1537524712500137e-07,
+      "loss": 0.1975,
+      "step": 10313
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.14922332126296e-07,
+      "loss": 0.1942,
+      "step": 10314
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1447030271652948e-07,
+      "loss": 0.1923,
+      "step": 10315
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1401915893620053e-07,
+      "loss": 0.1944,
+      "step": 10316
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1356890082572459e-07,
+      "loss": 0.2226,
+      "step": 10317
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.131195284254405e-07,
+      "loss": 0.2224,
+      "step": 10318
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1267104177560717e-07,
+      "loss": 0.2155,
+      "step": 10319
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1222344091640358e-07,
+      "loss": 0.2052,
+      "step": 10320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1177672588792876e-07,
+      "loss": 0.2071,
+      "step": 10321
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1133089673020515e-07,
+      "loss": 0.2025,
+      "step": 10322
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1088595348317189e-07,
+      "loss": 0.2184,
+      "step": 10323
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1044189618669154e-07,
+      "loss": 0.1968,
+      "step": 10324
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0999872488054674e-07,
+      "loss": 0.1987,
+      "step": 10325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0955643960444018e-07,
+      "loss": 0.234,
+      "step": 10326
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.091150403979957e-07,
+      "loss": 0.2085,
+      "step": 10327
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0867452730075722e-07,
+      "loss": 0.2021,
+      "step": 10328
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0823490035218986e-07,
+      "loss": 0.2345,
+      "step": 10329
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0779615959167989e-07,
+      "loss": 0.2218,
+      "step": 10330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0735830505853251e-07,
+      "loss": 0.1974,
+      "step": 10331
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0692133679197525e-07,
+      "loss": 0.2148,
+      "step": 10332
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0648525483115346e-07,
+      "loss": 0.1893,
+      "step": 10333
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.06050059215137e-07,
+      "loss": 0.2129,
+      "step": 10334
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0561574998291358e-07,
+      "loss": 0.2307,
+      "step": 10335
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0518232717339317e-07,
+      "loss": 0.2092,
+      "step": 10336
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.047497908254047e-07,
+      "loss": 0.2076,
+      "step": 10337
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0431814097769833e-07,
+      "loss": 0.2141,
+      "step": 10338
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0388737766894419e-07,
+      "loss": 0.2103,
+      "step": 10339
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0345750093773588e-07,
+      "loss": 0.2097,
+      "step": 10340
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.030285108225837e-07,
+      "loss": 0.2252,
+      "step": 10341
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0260040736192022e-07,
+      "loss": 0.2028,
+      "step": 10342
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0217319059409925e-07,
+      "loss": 0.1892,
+      "step": 10343
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0174686055739458e-07,
+      "loss": 0.1839,
+      "step": 10344
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0132141729000123e-07,
+      "loss": 0.1977,
+      "step": 10345
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0089686083003203e-07,
+      "loss": 0.2182,
+      "step": 10346
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0047319121552213e-07,
+      "loss": 0.2265,
+      "step": 10347
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0005040848443004e-07,
+      "loss": 0.1977,
+      "step": 10348
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.962851267462991e-08,
+      "loss": 0.1994,
+      "step": 10349
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.92075038239193e-08,
+      "loss": 0.2086,
+      "step": 10350
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.878738197001692e-08,
+      "loss": 0.2108,
+      "step": 10351
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.836814715055931e-08,
+      "loss": 0.2076,
+      "step": 10352
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.794979940310533e-08,
+      "loss": 0.2194,
+      "step": 10353
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.753233876513501e-08,
+      "loss": 0.2074,
+      "step": 10354
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.711576527404731e-08,
+      "loss": 0.216,
+      "step": 10355
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.670007896716238e-08,
+      "loss": 0.2066,
+      "step": 10356
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.628527988172154e-08,
+      "loss": 0.2063,
+      "step": 10357
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.587136805488506e-08,
+      "loss": 0.2054,
+      "step": 10358
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.545834352373551e-08,
+      "loss": 0.1848,
+      "step": 10359
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.50462063252744e-08,
+      "loss": 0.1785,
+      "step": 10360
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.463495649642551e-08,
+      "loss": 0.2082,
+      "step": 10361
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.42245940740305e-08,
+      "loss": 0.2083,
+      "step": 10362
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.381511909485441e-08,
+      "loss": 0.2079,
+      "step": 10363
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.340653159558233e-08,
+      "loss": 0.2399,
+      "step": 10364
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.299883161281609e-08,
+      "loss": 0.1904,
+      "step": 10365
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.259201918308314e-08,
+      "loss": 0.2071,
+      "step": 10366
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.218609434282876e-08,
+      "loss": 0.2336,
+      "step": 10367
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.178105712841945e-08,
+      "loss": 0.1895,
+      "step": 10368
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.137690757614059e-08,
+      "loss": 0.2215,
+      "step": 10369
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.09736457221999e-08,
+      "loss": 0.2197,
+      "step": 10370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.057127160272517e-08,
+      "loss": 0.2017,
+      "step": 10371
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.01697852537653e-08,
+      "loss": 0.2097,
+      "step": 10372
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.97691867112882e-08,
+      "loss": 0.2197,
+      "step": 10373
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.936947601118296e-08,
+      "loss": 0.2024,
+      "step": 10374
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.897065318925757e-08,
+      "loss": 0.2137,
+      "step": 10375
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.857271828124348e-08,
+      "loss": 0.218,
+      "step": 10376
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.817567132279104e-08,
+      "loss": 0.2324,
+      "step": 10377
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.777951234947179e-08,
+      "loss": 0.2247,
+      "step": 10378
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.738424139677515e-08,
+      "loss": 0.2114,
+      "step": 10379
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.698985850011388e-08,
+      "loss": 0.224,
+      "step": 10380
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.659636369481972e-08,
+      "loss": 0.2477,
+      "step": 10381
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.620375701614559e-08,
+      "loss": 0.2184,
+      "step": 10382
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.581203849926445e-08,
+      "loss": 0.2033,
+      "step": 10383
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.542120817927047e-08,
+      "loss": 0.2047,
+      "step": 10384
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.503126609117673e-08,
+      "loss": 0.2195,
+      "step": 10385
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.464221226991753e-08,
+      "loss": 0.2042,
+      "step": 10386
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.425404675034721e-08,
+      "loss": 0.2169,
+      "step": 10387
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.386676956724238e-08,
+      "loss": 0.1989,
+      "step": 10388
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.348038075529641e-08,
+      "loss": 0.2014,
+      "step": 10389
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.309488034912827e-08,
+      "loss": 0.2089,
+      "step": 10390
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.271026838327145e-08,
+      "loss": 0.2106,
+      "step": 10391
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.232654489218395e-08,
+      "loss": 0.1978,
+      "step": 10392
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.19437099102427e-08,
+      "loss": 0.2125,
+      "step": 10393
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.156176347174583e-08,
+      "loss": 0.2128,
+      "step": 10394
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.118070561091151e-08,
+      "loss": 0.237,
+      "step": 10395
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.080053636187801e-08,
+      "loss": 0.2242,
+      "step": 10396
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.042125575870362e-08,
+      "loss": 0.2213,
+      "step": 10397
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.004286383536675e-08,
+      "loss": 0.2043,
+      "step": 10398
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.966536062576912e-08,
+      "loss": 0.2072,
+      "step": 10399
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.928874616372817e-08,
+      "loss": 0.22,
+      "step": 10400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.891302048298576e-08,
+      "loss": 0.2253,
+      "step": 10401
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.853818361720277e-08,
+      "loss": 0.1987,
+      "step": 10402
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.8164235599959e-08,
+      "loss": 0.2045,
+      "step": 10403
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.779117646475765e-08,
+      "loss": 0.1967,
+      "step": 10404
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.741900624501975e-08,
+      "loss": 0.2,
+      "step": 10405
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.70477249740853e-08,
+      "loss": 0.2076,
+      "step": 10406
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.667733268521993e-08,
+      "loss": 0.2143,
+      "step": 10407
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.630782941160486e-08,
+      "loss": 0.2031,
+      "step": 10408
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.593921518634472e-08,
+      "loss": 0.2286,
+      "step": 10409
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.5571490042462e-08,
+      "loss": 0.2111,
+      "step": 10410
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.520465401290033e-08,
+      "loss": 0.2137,
+      "step": 10411
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.483870713052454e-08,
+      "loss": 0.2096,
+      "step": 10412
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.447364942811841e-08,
+      "loss": 0.2314,
+      "step": 10413
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.41094809383891e-08,
+      "loss": 0.2458,
+      "step": 10414
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.374620169395941e-08,
+      "loss": 0.2467,
+      "step": 10415
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.338381172737663e-08,
+      "loss": 0.1934,
+      "step": 10416
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.302231107110591e-08,
+      "loss": 0.2169,
+      "step": 10417
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.266169975753357e-08,
+      "loss": 0.231,
+      "step": 10418
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.230197781896597e-08,
+      "loss": 0.2023,
+      "step": 10419
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.19431452876318e-08,
+      "loss": 0.2025,
+      "step": 10420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.158520219567532e-08,
+      "loss": 0.1739,
+      "step": 10421
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.122814857516646e-08,
+      "loss": 0.2058,
+      "step": 10422
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.087198445809296e-08,
+      "loss": 0.179,
+      "step": 10423
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.05167098763615e-08,
+      "loss": 0.202,
+      "step": 10424
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.016232486180218e-08,
+      "loss": 0.2064,
+      "step": 10425
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.980882944616296e-08,
+      "loss": 0.1948,
+      "step": 10426
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.945622366111404e-08,
+      "loss": 0.2068,
+      "step": 10427
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.910450753824238e-08,
+      "loss": 0.2172,
+      "step": 10428
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.875368110906055e-08,
+      "loss": 0.2018,
+      "step": 10429
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.840374440499675e-08,
+      "loss": 0.1941,
+      "step": 10430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.805469745740145e-08,
+      "loss": 0.2223,
+      "step": 10431
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.770654029754631e-08,
+      "loss": 0.1966,
+      "step": 10432
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.735927295662081e-08,
+      "loss": 0.2025,
+      "step": 10433
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.701289546573675e-08,
+      "loss": 0.2196,
+      "step": 10434
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.666740785592596e-08,
+      "loss": 0.2019,
+      "step": 10435
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.632281015813924e-08,
+      "loss": 0.2016,
+      "step": 10436
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.597910240324967e-08,
+      "loss": 0.1927,
+      "step": 10437
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.563628462204819e-08,
+      "loss": 0.2108,
+      "step": 10438
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.529435684524798e-08,
+      "loss": 0.2076,
+      "step": 10439
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.495331910348235e-08,
+      "loss": 0.1973,
+      "step": 10440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.461317142730461e-08,
+      "loss": 0.2103,
+      "step": 10441
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.427391384718707e-08,
+      "loss": 0.2079,
+      "step": 10442
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.393554639352206e-08,
+      "loss": 0.2028,
+      "step": 10443
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.359806909662647e-08,
+      "loss": 0.2236,
+      "step": 10444
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.326148198673165e-08,
+      "loss": 0.1921,
+      "step": 10445
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.29257850939946e-08,
+      "loss": 0.2113,
+      "step": 10446
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.259097844848683e-08,
+      "loss": 0.2279,
+      "step": 10447
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.225706208020543e-08,
+      "loss": 0.2311,
+      "step": 10448
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.192403601906427e-08,
+      "loss": 0.2304,
+      "step": 10449
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.159190029489948e-08,
+      "loss": 0.2253,
+      "step": 10450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.126065493746614e-08,
+      "loss": 0.1964,
+      "step": 10451
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.09302999764394e-08,
+      "loss": 0.2124,
+      "step": 10452
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.06008354414167e-08,
+      "loss": 0.2149,
+      "step": 10453
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.027226136191333e-08,
+      "loss": 0.2164,
+      "step": 10454
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.994457776736573e-08,
+      "loss": 0.2357,
+      "step": 10455
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.961778468713042e-08,
+      "loss": 0.2186,
+      "step": 10456
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.929188215048509e-08,
+      "loss": 0.208,
+      "step": 10457
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.896687018662639e-08,
+      "loss": 0.2117,
+      "step": 10458
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8642748824672135e-08,
+      "loss": 0.2103,
+      "step": 10459
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.831951809365799e-08,
+      "loss": 0.224,
+      "step": 10460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.799717802254412e-08,
+      "loss": 0.205,
+      "step": 10461
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7675728640207426e-08,
+      "loss": 0.1797,
+      "step": 10462
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.735516997544488e-08,
+      "loss": 0.2111,
+      "step": 10463
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.703550205697683e-08,
+      "loss": 0.2097,
+      "step": 10464
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.671672491344149e-08,
+      "loss": 0.2124,
+      "step": 10465
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.639883857339601e-08,
+      "loss": 0.2008,
+      "step": 10466
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.608184306532094e-08,
+      "loss": 0.2204,
+      "step": 10467
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.576573841761579e-08,
+      "loss": 0.2141,
+      "step": 10468
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.545052465859791e-08,
+      "loss": 0.2201,
+      "step": 10469
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.513620181650803e-08,
+      "loss": 0.2145,
+      "step": 10470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.482276991950697e-08,
+      "loss": 0.2005,
+      "step": 10471
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.451022899567115e-08,
+      "loss": 0.2076,
+      "step": 10472
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.419857907300485e-08,
+      "loss": 0.2179,
+      "step": 10473
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.388782017942573e-08,
+      "loss": 0.1952,
+      "step": 10474
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3577952342774854e-08,
+      "loss": 0.2331,
+      "step": 10475
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.326897559081334e-08,
+      "loss": 0.2125,
+      "step": 10476
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.296088995122017e-08,
+      "loss": 0.2009,
+      "step": 10477
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.265369545159882e-08,
+      "loss": 0.2164,
+      "step": 10478
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2347392119467265e-08,
+      "loss": 0.2357,
+      "step": 10479
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.204197998227023e-08,
+      "loss": 0.1995,
+      "step": 10480
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.173745906736694e-08,
+      "loss": 0.2008,
+      "step": 10481
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.143382940203889e-08,
+      "loss": 0.2221,
+      "step": 10482
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.113109101348879e-08,
+      "loss": 0.195,
+      "step": 10483
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.082924392883826e-08,
+      "loss": 0.202,
+      "step": 10484
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0528288175129005e-08,
+      "loss": 0.2223,
+      "step": 10485
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0228223779322794e-08,
+      "loss": 0.2196,
+      "step": 10486
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9929050768302564e-08,
+      "loss": 0.2236,
+      "step": 10487
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.963076916887022e-08,
+      "loss": 0.2036,
+      "step": 10488
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.933337900774993e-08,
+      "loss": 0.2116,
+      "step": 10489
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9036880311582604e-08,
+      "loss": 0.2061,
+      "step": 10490
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.874127310693144e-08,
+      "loss": 0.2135,
+      "step": 10491
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.844655742027971e-08,
+      "loss": 0.1892,
+      "step": 10492
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.815273327803183e-08,
+      "loss": 0.2066,
+      "step": 10493
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.785980070650786e-08,
+      "loss": 0.2395,
+      "step": 10494
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.75677597319546e-08,
+      "loss": 0.2167,
+      "step": 10495
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7276610380533325e-08,
+      "loss": 0.2178,
+      "step": 10496
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.698635267832874e-08,
+      "loss": 0.1801,
+      "step": 10497
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.669698665134448e-08,
+      "loss": 0.196,
+      "step": 10498
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.640851232550425e-08,
+      "loss": 0.2241,
+      "step": 10499
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.612092972665183e-08,
+      "loss": 0.2088,
+      "step": 10500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.583423888055105e-08,
+      "loss": 0.2056,
+      "step": 10501
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.554843981288692e-08,
+      "loss": 0.2214,
+      "step": 10502
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5263532549262304e-08,
+      "loss": 0.2219,
+      "step": 10503
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.497951711520343e-08,
+      "loss": 0.2092,
+      "step": 10504
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.46963935361544e-08,
+      "loss": 0.221,
+      "step": 10505
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.441416183747826e-08,
+      "loss": 0.213,
+      "step": 10506
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.413282204446145e-08,
+      "loss": 0.1865,
+      "step": 10507
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.385237418230825e-08,
+      "loss": 0.2244,
+      "step": 10508
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.357281827614413e-08,
+      "loss": 0.2174,
+      "step": 10509
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3294154351012366e-08,
+      "loss": 0.2022,
+      "step": 10510
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.301638243187967e-08,
+      "loss": 0.1949,
+      "step": 10511
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.273950254362946e-08,
+      "loss": 0.2163,
+      "step": 10512
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2463514711068575e-08,
+      "loss": 0.2152,
+      "step": 10513
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2188418958922785e-08,
+      "loss": 0.2059,
+      "step": 10514
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1914215311836814e-08,
+      "loss": 0.2141,
+      "step": 10515
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1640903794375464e-08,
+      "loss": 0.1957,
+      "step": 10516
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1368484431023593e-08,
+      "loss": 0.1901,
+      "step": 10517
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.109695724618945e-08,
+      "loss": 0.2027,
+      "step": 10518
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0826322264196916e-08,
+      "loss": 0.193,
+      "step": 10519
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.055657950929215e-08,
+      "loss": 0.2238,
+      "step": 10520
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.028772900564137e-08,
+      "loss": 0.1953,
+      "step": 10521
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.001977077732977e-08,
+      "loss": 0.1944,
+      "step": 10522
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9752704848363686e-08,
+      "loss": 0.2093,
+      "step": 10523
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9486531242668435e-08,
+      "loss": 0.2116,
+      "step": 10524
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.92212499840916e-08,
+      "loss": 0.2072,
+      "step": 10525
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8956861096398624e-08,
+      "loss": 0.2384,
+      "step": 10526
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8693364603274995e-08,
+      "loss": 0.2049,
+      "step": 10527
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8430760528327396e-08,
+      "loss": 0.1854,
+      "step": 10528
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.816904889508255e-08,
+      "loss": 0.2233,
+      "step": 10529
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7908229726987264e-08,
+      "loss": 0.1971,
+      "step": 10530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7648303047406185e-08,
+      "loss": 0.2114,
+      "step": 10531
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.738926887962735e-08,
+      "loss": 0.2202,
+      "step": 10532
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7131127246856636e-08,
+      "loss": 0.1943,
+      "step": 10533
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.687387817221999e-08,
+      "loss": 0.2143,
+      "step": 10534
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.661752167876453e-08,
+      "loss": 0.2008,
+      "step": 10535
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.636205778945745e-08,
+      "loss": 0.2369,
+      "step": 10536
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.610748652718488e-08,
+      "loss": 0.2322,
+      "step": 10537
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.58538079147519e-08,
+      "loss": 0.2153,
+      "step": 10538
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.560102197488813e-08,
+      "loss": 0.1957,
+      "step": 10539
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.534912873023766e-08,
+      "loss": 0.2154,
+      "step": 10540
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5098128203369106e-08,
+      "loss": 0.1848,
+      "step": 10541
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.484802041676894e-08,
+      "loss": 0.2105,
+      "step": 10542
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.459880539284255e-08,
+      "loss": 0.2255,
+      "step": 10543
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4350483153917644e-08,
+      "loss": 0.2132,
+      "step": 10544
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.410305372224088e-08,
+      "loss": 0.2162,
+      "step": 10545
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.385651711998006e-08,
+      "loss": 0.2096,
+      "step": 10546
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3610873369219754e-08,
+      "loss": 0.2013,
+      "step": 10547
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.336612249196902e-08,
+      "loss": 0.1763,
+      "step": 10548
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3122264510154764e-08,
+      "loss": 0.1852,
+      "step": 10549
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2879299445621736e-08,
+      "loss": 0.204,
+      "step": 10550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2637227320138074e-08,
+      "loss": 0.2164,
+      "step": 10551
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.239604815539088e-08,
+      "loss": 0.1918,
+      "step": 10552
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.215576197298731e-08,
+      "loss": 0.2338,
+      "step": 10553
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.191636879445348e-08,
+      "loss": 0.2266,
+      "step": 10554
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1677868641236676e-08,
+      "loss": 0.1872,
+      "step": 10555
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.144026153470314e-08,
+      "loss": 0.2096,
+      "step": 10556
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1203547496140295e-08,
+      "loss": 0.2074,
+      "step": 10557
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.096772654675451e-08,
+      "loss": 0.1838,
+      "step": 10558
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.073279870767332e-08,
+      "loss": 0.2134,
+      "step": 10559
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.049876399994323e-08,
+      "loss": 0.2182,
+      "step": 10560
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.02656224445308e-08,
+      "loss": 0.2131,
+      "step": 10561
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.003337406232265e-08,
+      "loss": 0.2042,
+      "step": 10562
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9802018874126593e-08,
+      "loss": 0.217,
+      "step": 10563
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9571556900668264e-08,
+      "loss": 0.2071,
+      "step": 10564
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9341988162595593e-08,
+      "loss": 0.1735,
+      "step": 10565
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9113312680474348e-08,
+      "loss": 0.2099,
+      "step": 10566
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.888553047479037e-08,
+      "loss": 0.215,
+      "step": 10567
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8658641565952882e-08,
+      "loss": 0.2078,
+      "step": 10568
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8432645974285632e-08,
+      "loss": 0.2398,
+      "step": 10569
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.820754372003798e-08,
+      "loss": 0.2105,
+      "step": 10570
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7983334823373786e-08,
+      "loss": 0.1955,
+      "step": 10571
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7760019304381436e-08,
+      "loss": 0.2094,
+      "step": 10572
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7537597183068254e-08,
+      "loss": 0.2165,
+      "step": 10573
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7316068479357193e-08,
+      "loss": 0.2118,
+      "step": 10574
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7095433213097933e-08,
+      "loss": 0.2226,
+      "step": 10575
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6875691404055772e-08,
+      "loss": 0.1971,
+      "step": 10576
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6656843071916073e-08,
+      "loss": 0.23,
+      "step": 10577
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6438888236287596e-08,
+      "loss": 0.2113,
+      "step": 10578
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6221826916693614e-08,
+      "loss": 0.2317,
+      "step": 10579
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6005659132581907e-08,
+      "loss": 0.2128,
+      "step": 10580
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5790384903319198e-08,
+      "loss": 0.1756,
+      "step": 10581
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5576004248190068e-08,
+      "loss": 0.213,
+      "step": 10582
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.536251718640248e-08,
+      "loss": 0.2328,
+      "step": 10583
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.514992373708003e-08,
+      "loss": 0.2287,
+      "step": 10584
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.49382239192697e-08,
+      "loss": 0.2195,
+      "step": 10585
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4727417751937433e-08,
+      "loss": 0.2497,
+      "step": 10586
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.451750525397034e-08,
+      "loss": 0.2345,
+      "step": 10587
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4308486444172273e-08,
+      "loss": 0.2107,
+      "step": 10588
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4100361341269363e-08,
+      "loss": 0.2217,
+      "step": 10589
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3893129963907803e-08,
+      "loss": 0.1831,
+      "step": 10590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.368679233065163e-08,
+      "loss": 0.2197,
+      "step": 10591
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3481348459988285e-08,
+      "loss": 0.216,
+      "step": 10592
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.327679837032304e-08,
+      "loss": 0.2076,
+      "step": 10593
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3073142079979015e-08,
+      "loss": 0.2148,
+      "step": 10594
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.287037960720384e-08,
+      "loss": 0.2126,
+      "step": 10595
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2668510970161874e-08,
+      "loss": 0.2089,
+      "step": 10596
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2467536186937532e-08,
+      "loss": 0.21,
+      "step": 10597
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.226745527553642e-08,
+      "loss": 0.2175,
+      "step": 10598
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2068268253884195e-08,
+      "loss": 0.2195,
+      "step": 10599
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1869975139824363e-08,
+      "loss": 0.1883,
+      "step": 10600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.167257595112271e-08,
+      "loss": 0.2035,
+      "step": 10601
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1476070705463982e-08,
+      "loss": 0.2092,
+      "step": 10602
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1280459420451872e-08,
+      "loss": 0.2233,
+      "step": 10603
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.108574211361125e-08,
+      "loss": 0.2172,
+      "step": 10604
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0891918802387058e-08,
+      "loss": 0.1948,
+      "step": 10605
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0698989504142064e-08,
+      "loss": 0.2298,
+      "step": 10606
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0506954236163557e-08,
+      "loss": 0.2191,
+      "step": 10607
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.031581301565222e-08,
+      "loss": 0.1965,
+      "step": 10608
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0125565859734353e-08,
+      "loss": 0.1983,
+      "step": 10609
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9936212785452992e-08,
+      "loss": 0.1935,
+      "step": 10610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.974775380977345e-08,
+      "loss": 0.2009,
+      "step": 10611
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9560188949576676e-08,
+      "loss": 0.225,
+      "step": 10612
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9373518221669218e-08,
+      "loss": 0.2258,
+      "step": 10613
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.918774164277326e-08,
+      "loss": 0.1853,
+      "step": 10614
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9002859229531046e-08,
+      "loss": 0.2076,
+      "step": 10615
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.881887099850821e-08,
+      "loss": 0.216,
+      "step": 10616
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8635776966186015e-08,
+      "loss": 0.2067,
+      "step": 10617
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8453577148969116e-08,
+      "loss": 0.2174,
+      "step": 10618
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.827227156318001e-08,
+      "loss": 0.2169,
+      "step": 10619
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8091860225061266e-08,
+      "loss": 0.2141,
+      "step": 10620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.791234315077661e-08,
+      "loss": 0.2054,
+      "step": 10621
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7733720356407637e-08,
+      "loss": 0.2088,
+      "step": 10622
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.755599185795598e-08,
+      "loss": 0.2186,
+      "step": 10623
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.737915767134668e-08,
+      "loss": 0.2122,
+      "step": 10624
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7203217812420403e-08,
+      "loss": 0.205,
+      "step": 10625
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7028172296938984e-08,
+      "loss": 0.2281,
+      "step": 10626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.685402114058543e-08,
+      "loss": 0.2159,
+      "step": 10627
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.668076435896171e-08,
+      "loss": 0.2155,
+      "step": 10628
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6508401967588738e-08,
+      "loss": 0.2133,
+      "step": 10629
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.633693398190861e-08,
+      "loss": 0.1983,
+      "step": 10630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.616636041728348e-08,
+      "loss": 0.1968,
+      "step": 10631
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5996681288993343e-08,
+      "loss": 0.219,
+      "step": 10632
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.58278966122416e-08,
+      "loss": 0.1935,
+      "step": 10633
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5660006402147266e-08,
+      "loss": 0.179,
+      "step": 10634
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5493010673752752e-08,
+      "loss": 0.2276,
+      "step": 10635
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5326909442018313e-08,
+      "loss": 0.1834,
+      "step": 10636
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.516170272182538e-08,
+      "loss": 0.1984,
+      "step": 10637
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.499739052797433e-08,
+      "loss": 0.2121,
+      "step": 10638
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.483397287518451e-08,
+      "loss": 0.2297,
+      "step": 10639
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4671449778097535e-08,
+      "loss": 0.2079,
+      "step": 10640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4509821251272871e-08,
+      "loss": 0.2108,
+      "step": 10641
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4349087309191156e-08,
+      "loss": 0.2601,
+      "step": 10642
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4189247966253094e-08,
+      "loss": 0.2168,
+      "step": 10643
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4030303236776122e-08,
+      "loss": 0.2115,
+      "step": 10644
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3872253135001069e-08,
+      "loss": 0.1923,
+      "step": 10645
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.371509767508883e-08,
+      "loss": 0.218,
+      "step": 10646
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3558836871115922e-08,
+      "loss": 0.1885,
+      "step": 10647
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3403470737083368e-08,
+      "loss": 0.2117,
+      "step": 10648
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3248999286910035e-08,
+      "loss": 0.2067,
+      "step": 10649
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3095422534434854e-08,
+      "loss": 0.2051,
+      "step": 10650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2942740493415706e-08,
+      "loss": 0.1913,
+      "step": 10651
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.279095317753165e-08,
+      "loss": 0.2026,
+      "step": 10652
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2640060600380699e-08,
+      "loss": 0.2333,
+      "step": 10653
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2490062775482037e-08,
+      "loss": 0.19,
+      "step": 10654
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2340959716273804e-08,
+      "loss": 0.2151,
+      "step": 10655
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2192751436111983e-08,
+      "loss": 0.1968,
+      "step": 10656
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2045437948275952e-08,
+      "loss": 0.2193,
+      "step": 10657
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.189901926596293e-08,
+      "loss": 0.1884,
+      "step": 10658
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1753495402290205e-08,
+      "loss": 0.2107,
+      "step": 10659
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1608866370296234e-08,
+      "loss": 0.1967,
+      "step": 10660
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1465132182936211e-08,
+      "loss": 0.1801,
+      "step": 10661
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1322292853088722e-08,
+      "loss": 0.2153,
+      "step": 10662
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.118034839354909e-08,
+      "loss": 0.2317,
+      "step": 10663
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1039298817033806e-08,
+      "loss": 0.2199,
+      "step": 10664
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.089914413618054e-08,
+      "loss": 0.1949,
+      "step": 10665
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0759884363545913e-08,
+      "loss": 0.2218,
+      "step": 10666
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0621519511604394e-08,
+      "loss": 0.2025,
+      "step": 10667
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0484049592752732e-08,
+      "loss": 0.2347,
+      "step": 10668
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.034747461930552e-08,
+      "loss": 0.2257,
+      "step": 10669
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0211794603500747e-08,
+      "loss": 0.1988,
+      "step": 10670
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0077009557490913e-08,
+      "loss": 0.1996,
+      "step": 10671
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.943119493354137e-09,
+      "loss": 0.1882,
+      "step": 10672
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.810124423083046e-09,
+      "loss": 0.1958,
+      "step": 10673
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.678024358592553e-09,
+      "loss": 0.198,
+      "step": 10674
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.546819311718747e-09,
+      "loss": 0.2244,
+      "step": 10675
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.416509294215559e-09,
+      "loss": 0.1961,
+      "step": 10676
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.287094317756985e-09,
+      "loss": 0.2145,
+      "step": 10677
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.158574393938192e-09,
+      "loss": 0.2125,
+      "step": 10678
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.030949534271082e-09,
+      "loss": 0.2085,
+      "step": 10679
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.904219750189846e-09,
+      "loss": 0.1891,
+      "step": 10680
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.778385053048732e-09,
+      "loss": 0.1954,
+      "step": 10681
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.653445454122056e-09,
+      "loss": 0.2002,
+      "step": 10682
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.529400964600864e-09,
+      "loss": 0.1912,
+      "step": 10683
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.406251595599601e-09,
+      "loss": 0.2142,
+      "step": 10684
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.283997358150553e-09,
+      "loss": 0.207,
+      "step": 10685
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.162638263204959e-09,
+      "loss": 0.2025,
+      "step": 10686
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.042174321637453e-09,
+      "loss": 0.1849,
+      "step": 10687
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.922605544239403e-09,
+      "loss": 0.1987,
+      "step": 10688
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.803931941722242e-09,
+      "loss": 0.213,
+      "step": 10689
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.686153524718576e-09,
+      "loss": 0.1912,
+      "step": 10690
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.569270303778852e-09,
+      "loss": 0.2015,
+      "step": 10691
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.453282289375808e-09,
+      "loss": 0.2096,
+      "step": 10692
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.338189491900016e-09,
+      "loss": 0.2143,
+      "step": 10693
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.223991921662121e-09,
+      "loss": 0.2199,
+      "step": 10694
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.1106895888928274e-09,
+      "loss": 0.2182,
+      "step": 10695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.998282503744014e-09,
+      "loss": 0.2235,
+      "step": 10696
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.886770676284293e-09,
+      "loss": 0.1951,
+      "step": 10697
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.776154116504563e-09,
+      "loss": 0.2003,
+      "step": 10698
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6664328343157835e-09,
+      "loss": 0.2306,
+      "step": 10699
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.557606839546759e-09,
+      "loss": 0.2104,
+      "step": 10700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4496761419463595e-09,
+      "loss": 0.2376,
+      "step": 10701
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.342640751185736e-09,
+      "loss": 0.2151,
+      "step": 10702
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.236500676851664e-09,
+      "loss": 0.2153,
+      "step": 10703
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.131255928455426e-09,
+      "loss": 0.2512,
+      "step": 10704
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.026906515425035e-09,
+      "loss": 0.1958,
+      "step": 10705
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.923452447107458e-09,
+      "loss": 0.1994,
+      "step": 10706
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.8208937327730585e-09,
+      "loss": 0.2229,
+      "step": 10707
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.7192303816089316e-09,
+      "loss": 0.2119,
+      "step": 10708
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.618462402723346e-09,
+      "loss": 0.2197,
+      "step": 10709
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.518589805143526e-09,
+      "loss": 0.2104,
+      "step": 10710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.419612597816759e-09,
+      "loss": 0.218,
+      "step": 10711
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.321530789610396e-09,
+      "loss": 0.2146,
+      "step": 10712
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.224344389311853e-09,
+      "loss": 0.2029,
+      "step": 10713
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.128053405627498e-09,
+      "loss": 0.2061,
+      "step": 10714
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.032657847183764e-09,
+      "loss": 0.207,
+      "step": 10715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.938157722528258e-09,
+      "loss": 0.2086,
+      "step": 10716
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.844553040125322e-09,
+      "loss": 0.216,
+      "step": 10717
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.751843808361578e-09,
+      "loss": 0.2306,
+      "step": 10718
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6600300355437165e-09,
+      "loss": 0.2089,
+      "step": 10719
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.56911172989516e-09,
+      "loss": 0.1913,
+      "step": 10720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.479088899562723e-09,
+      "loss": 0.2293,
+      "step": 10721
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.389961552611066e-09,
+      "loss": 0.2152,
+      "step": 10722
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.301729697023804e-09,
+      "loss": 0.2274,
+      "step": 10723
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.214393340707945e-09,
+      "loss": 0.218,
+      "step": 10724
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.12795249148612e-09,
+      "loss": 0.2005,
+      "step": 10725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.042407157102135e-09,
+      "loss": 0.1928,
+      "step": 10726
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.957757345220969e-09,
+      "loss": 0.2086,
+      "step": 10727
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.874003063425446e-09,
+      "loss": 0.2133,
+      "step": 10728
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.791144319218454e-09,
+      "loss": 0.1902,
+      "step": 10729
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7091811200251627e-09,
+      "loss": 0.2175,
+      "step": 10730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6281134731874778e-09,
+      "loss": 0.2105,
+      "step": 10731
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.547941385967368e-09,
+      "loss": 0.2372,
+      "step": 10732
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.468664865547977e-09,
+      "loss": 0.197,
+      "step": 10733
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3902839190314005e-09,
+      "loss": 0.2143,
+      "step": 10734
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.31279855344091e-09,
+      "loss": 0.1923,
+      "step": 10735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2362087757165096e-09,
+      "loss": 0.214,
+      "step": 10736
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1605145927193768e-09,
+      "loss": 0.2128,
+      "step": 10737
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0857160112329753e-09,
+      "loss": 0.2187,
+      "step": 10738
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0118130379575005e-09,
+      "loss": 0.1812,
+      "step": 10739
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.938805679514323e-09,
+      "loss": 0.2017,
+      "step": 10740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.866693942441545e-09,
+      "loss": 0.2145,
+      "step": 10741
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7954778332028863e-09,
+      "loss": 0.2117,
+      "step": 10742
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.725157358176578e-09,
+      "loss": 0.1978,
+      "step": 10743
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.655732523663135e-09,
+      "loss": 0.2067,
+      "step": 10744
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.587203335882027e-09,
+      "loss": 0.2012,
+      "step": 10745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5195698009716772e-09,
+      "loss": 0.2116,
+      "step": 10746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4528319249927935e-09,
+      "loss": 0.2143,
+      "step": 10747
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3869897139250365e-09,
+      "loss": 0.2226,
+      "step": 10748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3220431736648008e-09,
+      "loss": 0.2086,
+      "step": 10749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2579923100307655e-09,
+      "loss": 0.1981,
+      "step": 10750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1948371287638936e-09,
+      "loss": 0.2116,
+      "step": 10751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1325776355185513e-09,
+      "loss": 0.2136,
+      "step": 10752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0712138358747192e-09,
+      "loss": 0.2387,
+      "step": 10753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0107457353302217e-09,
+      "loss": 0.2232,
+      "step": 10754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.951173339300727e-09,
+      "loss": 0.2235,
+      "step": 10755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.892496653124187e-09,
+      "loss": 0.2117,
+      "step": 10756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8347156820563983e-09,
+      "loss": 0.191,
+      "step": 10757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7778304312754402e-09,
+      "loss": 0.179,
+      "step": 10758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7218409058761265e-09,
+      "loss": 0.1888,
+      "step": 10759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6667471108755552e-09,
+      "loss": 0.2262,
+      "step": 10760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6125490512086673e-09,
+      "loss": 0.186,
+      "step": 10761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5592467317315784e-09,
+      "loss": 0.2185,
+      "step": 10762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.506840157219358e-09,
+      "loss": 0.2135,
+      "step": 10763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4553293323660288e-09,
+      "loss": 0.2385,
+      "step": 10764
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4047142617890086e-09,
+      "loss": 0.2199,
+      "step": 10765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3549949500202275e-09,
+      "loss": 0.1989,
+      "step": 10766
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3061714015139005e-09,
+      "loss": 0.1937,
+      "step": 10767
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2582436206465264e-09,
+      "loss": 0.2283,
+      "step": 10768
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.211211611710228e-09,
+      "loss": 0.2555,
+      "step": 10769
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.165075378917191e-09,
+      "loss": 0.2218,
+      "step": 10770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1198349264041064e-09,
+      "loss": 0.205,
+      "step": 10771
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.075490258221068e-09,
+      "loss": 0.1943,
+      "step": 10772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.032041378341564e-09,
+      "loss": 0.182,
+      "step": 10773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.894882906591464e-10,
+      "loss": 0.2207,
+      "step": 10774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.478309989852108e-10,
+      "loss": 0.1812,
+      "step": 10775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.07069507051217e-10,
+      "loss": 0.2281,
+      "step": 10776
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.672038185097986e-10,
+      "loss": 0.2078,
+      "step": 10777
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.282339369325432e-10,
+      "loss": 0.1902,
+      "step": 10778
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.901598658111021e-10,
+      "loss": 0.2126,
+      "step": 10779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.529816085549701e-10,
+      "loss": 0.2192,
+      "step": 10780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.166991684959268e-10,
+      "loss": 0.1996,
+      "step": 10781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.813125488835947e-10,
+      "loss": 0.1888,
+      "step": 10782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.468217528898813e-10,
+      "loss": 0.2043,
+      "step": 10783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.132267836023164e-10,
+      "loss": 0.1914,
+      "step": 10784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.805276440329355e-10,
+      "loss": 0.1905,
+      "step": 10785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.487243371105067e-10,
+      "loss": 0.21,
+      "step": 10786
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.178168656838622e-10,
+      "loss": 0.2107,
+      "step": 10787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.878052325218985e-10,
+      "loss": 0.2144,
+      "step": 10788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.5868944031468575e-10,
+      "loss": 0.1958,
+      "step": 10789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.304694916679175e-10,
+      "loss": 0.2083,
+      "step": 10790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0314538911290224e-10,
+      "loss": 0.203,
+      "step": 10791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7671713509435104e-10,
+      "loss": 0.2192,
+      "step": 10792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5118473198259006e-10,
+      "loss": 0.235,
+      "step": 10793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2654818206467877e-10,
+      "loss": 0.2252,
+      "step": 10794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.028074875466303e-10,
+      "loss": 0.203,
+      "step": 10795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.799626505556319e-10,
+      "loss": 0.2057,
+      "step": 10796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5801367313782464e-10,
+      "loss": 0.2001,
+      "step": 10797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3696055726163403e-10,
+      "loss": 0.2137,
+      "step": 10798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1680330481110845e-10,
+      "loss": 0.214,
+      "step": 10799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9754191759258078e-10,
+      "loss": 0.2264,
+      "step": 10800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7917639733244785e-10,
+      "loss": 0.2049,
+      "step": 10801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6170674567606015e-10,
+      "loss": 0.2116,
+      "step": 10802
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4513296418772194e-10,
+      "loss": 0.2256,
+      "step": 10803
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.294550543518014e-10,
+      "loss": 0.2467,
+      "step": 10804
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1467301757384087e-10,
+      "loss": 0.1826,
+      "step": 10805
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.007868551772262e-10,
+      "loss": 0.2157,
+      "step": 10806
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.779656840762762e-11,
+      "loss": 0.2072,
+      "step": 10807
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.570215842744866e-11,
+      "loss": 0.2132,
+      "step": 10808
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.450362632026697e-11,
+      "loss": 0.207,
+      "step": 10809
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.420097308972416e-11,
+      "loss": 0.2559,
+      "step": 10810
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4794199659525806e-11,
+      "loss": 0.2107,
+      "step": 10811
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.628330687122095e-11,
+      "loss": 0.2027,
+      "step": 10812
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.866829548642258e-11,
+      "loss": 0.2211,
+      "step": 10813
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1949166190138316e-11,
+      "loss": 0.2104,
+      "step": 10814
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6125919581888583e-11,
+      "loss": 0.2021,
+      "step": 10815
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.119855618569865e-11,
+      "loss": 0.2105,
+      "step": 10816
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.167076440106613e-12,
+      "loss": 0.1925,
+      "step": 10817
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0314807081553995e-12,
+      "loss": 0.2261,
+      "step": 10818
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.791769270731436e-12,
+      "loss": 0.1983,
+      "step": 10819
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.479423276748662e-13,
+      "loss": 0.2114,
+      "step": 10820
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1905,
+      "step": 10821
+    },
+    {
+      "epoch": 3.0,
+      "step": 10821,
+      "total_flos": 2.3702016182399795e+17,
+      "train_loss": 0.3818010155925049,
+      "train_runtime": 12934.9481,
+      "train_samples_per_second": 107.071,
+      "train_steps_per_second": 0.837
+    }
+  ],
+  "max_steps": 10821,
+  "num_train_epochs": 3,
+  "total_flos": 2.3702016182399795e+17,
+  "trial_name": null,
+  "trial_params": null
+}