diff --git "a/sft/trainer_state.json" "b/sft/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/sft/trainer_state.json"
@@ -0,0 +1,114052 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 19004,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2.0068,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.9365,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.5098,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.3105,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.252,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.1855,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.0566,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.0552,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.0151,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.0405,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.0312,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9971,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9883,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9062,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9058,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8379,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9028,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8545,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8677,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8203,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.8408,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.7812,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.811,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.8818,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7998,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.8423,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7559,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7756,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7053,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6865,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.8066,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.8438,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7656,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7842,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7124,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7207,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6868,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7056,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6321,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7251,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7539,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7151,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6738,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7197,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7205,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6028,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6763,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6851,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6997,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6597,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6641,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6486,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7222,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6401,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6304,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6687,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7029,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6877,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6616,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5576,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7256,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5098,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5942,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7529,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.7341,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5671,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6213,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5881,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5991,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5793,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6284,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5493,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.543,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6165,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6104,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5308,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5322,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6875,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5449,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6055,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5779,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5625,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6274,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5874,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5613,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5898,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5735,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5371,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5237,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5728,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5933,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.594,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4995,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4836,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5476,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5205,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5242,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5439,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6692,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5256,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4504,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5769,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4084,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5278,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5149,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5208,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.6555,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5129,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4836,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4553,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5056,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5049,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4862,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5498,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5686,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5852,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.4858,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 0.5081,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5205,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5918,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5867,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4458,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5151,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4421,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4724,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4965,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4521,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4863,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5151,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5225,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4619,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4718,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5364,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5481,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.46,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4534,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4976,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5259,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3962,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5466,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5884,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.6555,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4961,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4258,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4968,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.52,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4993,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4817,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4275,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4663,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4385,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4268,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4077,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4932,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4994,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5059,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3867,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4244,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4819,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4844,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4056,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4927,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5579,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.439,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4602,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.4131,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4253,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.5127,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4177,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4919,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4756,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4368,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4116,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4828,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4448,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4227,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3527,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4856,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4478,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4692,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3666,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.436,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4392,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4343,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.489,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3855,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4017,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4417,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4911,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4082,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.358,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4827,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4312,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4116,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4526,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4028,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3688,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4402,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4236,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4994,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4364,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3881,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3936,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4247,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4204,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4335,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4084,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4187,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3995,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4939,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.3384,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4595,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2e-05,
+      "loss": 0.4213,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3899,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4978,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3772,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4454,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4634,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4033,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3615,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4293,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4087,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4661,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.5669,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3848,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4198,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4514,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.48,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4431,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4333,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4318,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4011,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4092,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.436,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4509,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.5039,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4441,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.424,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3564,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3788,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4976,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4387,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4395,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4873,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3914,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4242,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4668,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3904,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.33,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4032,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3856,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3525,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3954,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.415,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4385,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4729,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3977,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3347,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.3458,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.5129,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.4539,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4485,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4241,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3975,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3784,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.385,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4224,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4072,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4104,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3771,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3654,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3584,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4543,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3606,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4336,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3953,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3337,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4144,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3726,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3948,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3895,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4553,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4175,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4324,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.374,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.449,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.334,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3313,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3145,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.335,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3665,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3948,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3324,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3977,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4338,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4336,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4167,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4214,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4399,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.4148,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3436,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3979,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3871,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3969,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3024,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3008,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3716,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.3409,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3369,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4202,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3439,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3978,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4137,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3935,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3777,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3811,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3275,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3687,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.377,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3997,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3625,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4683,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3573,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3431,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.406,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3331,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.447,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4497,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4302,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3123,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3778,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3115,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2885,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4138,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3586,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3444,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3533,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.385,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3474,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3411,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3423,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3994,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3662,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3508,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3954,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3225,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.325,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3928,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2936,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3643,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3717,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3495,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.2887,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.322,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.4282,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2e-05,
+      "loss": 0.3834,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3304,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3158,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.4171,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2797,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3463,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3635,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3225,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3118,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2828,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2949,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3621,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3696,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3264,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3802,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.4155,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3649,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3496,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3025,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3558,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3717,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3685,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3955,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.361,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3215,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3546,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.4049,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3853,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3003,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2769,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3726,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.4044,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3862,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2715,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2565,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3348,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.2872,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3516,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.4548,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.316,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3301,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3754,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3466,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3462,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3588,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3508,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3293,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 0.3662,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3109,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3826,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2908,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3064,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3016,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.418,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3556,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2917,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3551,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3345,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3365,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3364,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.395,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3235,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.333,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2968,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3224,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3046,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.4061,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2603,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3069,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3884,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2881,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3199,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3147,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3311,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3099,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3727,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3433,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3398,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3459,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3887,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2615,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3206,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3127,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3138,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.45,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.397,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.364,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3706,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3761,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2991,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.327,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3416,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3262,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3978,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.3015,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2788,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.312,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.4041,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3186,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2992,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3409,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3365,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3848,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.4037,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3618,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3422,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3289,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.315,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3148,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.4402,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3484,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3674,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3568,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3281,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3312,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.4169,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3231,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.437,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3494,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3424,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3085,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2769,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3755,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3501,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.344,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3545,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3696,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.349,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.4467,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2963,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2118,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3181,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3369,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2755,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3176,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3337,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3428,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2438,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2689,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3627,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.2734,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2e-05,
+      "loss": 0.3137,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2501,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3175,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2952,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3335,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2822,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3518,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.434,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3879,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2736,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3204,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2958,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3928,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2511,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3209,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2827,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2791,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.231,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3096,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.26,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2443,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3428,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.356,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3589,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3562,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.464,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3667,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2826,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3149,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3907,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3429,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3113,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2905,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2739,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3206,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3477,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3679,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3577,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2997,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.345,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3502,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.2659,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3258,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3539,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3354,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3141,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3822,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.4287,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2e-05,
+      "loss": 0.3038,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.4126,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3101,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3795,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2991,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3833,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2739,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2631,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2786,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2981,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3113,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2792,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.26,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3418,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3588,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2891,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2563,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2358,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.29,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2798,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2711,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2129,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3694,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2413,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2935,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3433,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2371,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2843,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2766,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2714,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3356,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2646,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3016,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1798,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2495,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2863,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2616,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.377,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3187,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3142,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2933,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2902,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3716,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2532,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.3278,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.318,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2e-05,
+      "loss": 0.2969,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2734,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2721,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3282,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3789,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2383,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3378,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3016,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3063,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2867,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2552,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2181,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3031,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3268,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2595,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2534,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2712,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3571,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3047,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2876,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2969,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3214,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3092,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2776,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1839,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2607,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2361,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3273,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.275,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2186,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2189,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3126,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3086,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1965,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3405,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2778,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3088,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2579,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2217,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3096,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3368,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2363,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2736,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3928,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2499,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2701,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.2831,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.3038,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2e-05,
+      "loss": 0.355,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3105,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2537,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2786,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3076,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2749,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2986,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2927,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2277,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.276,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2784,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2764,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3182,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2754,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2699,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3506,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3229,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2341,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.205,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2538,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2679,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3054,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2231,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2603,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3013,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2501,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3827,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2571,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.229,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1942,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3002,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3363,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2604,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3048,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.318,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3286,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2805,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2537,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2971,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2723,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2247,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.313,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2636,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.295,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2579,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.2622,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.3093,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1984,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3214,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2756,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.288,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2455,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3376,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3107,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2795,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2701,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2802,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2632,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2823,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2793,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2971,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2462,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2672,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2507,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2557,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2033,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2283,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3297,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2739,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2105,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2953,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2328,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2495,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2568,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2691,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.243,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2325,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2938,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.291,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2679,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3057,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2924,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2427,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2538,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2339,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2856,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.301,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2533,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2762,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2372,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2645,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2983,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2932,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3186,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2454,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2955,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2715,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2179,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3458,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3009,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2635,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3328,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2455,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2394,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.301,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2999,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2158,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2604,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.239,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2357,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2903,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2314,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1788,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2112,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2681,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2926,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2512,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2637,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2151,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2142,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3212,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3094,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2268,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2681,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3232,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2363,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2275,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2877,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.255,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2081,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2508,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2931,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1934,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2684,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2731,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2143,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2675,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.3015,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2988,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2e-05,
+      "loss": 0.2087,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2351,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2446,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2842,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2623,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2782,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2818,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2396,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2594,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2605,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2511,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.308,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3169,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2396,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.257,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2344,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2479,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1974,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2917,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2621,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.293,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2806,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.4491,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2254,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3059,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2106,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2124,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3226,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2463,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3457,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3145,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2025,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3041,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2428,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2284,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3007,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2327,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2534,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2871,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2605,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2825,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3433,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.223,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3135,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2014,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.2804,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.3057,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2e-05,
+      "loss": 0.264,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2296,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2559,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3392,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2959,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2689,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2405,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2804,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2936,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2385,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3412,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2461,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2404,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.296,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2987,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2686,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2401,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3055,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.202,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2032,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.277,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2539,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2607,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2503,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2954,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2117,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.196,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2272,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2419,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2781,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2414,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1775,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2787,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.296,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2061,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.211,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2585,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2829,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.3109,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2196,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2869,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2029,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2379,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2e-05,
+      "loss": 0.2363,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2374,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2194,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3767,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2816,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2183,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2822,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2321,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2896,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2961,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.324,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2363,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2489,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2957,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2653,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2274,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2952,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2749,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2841,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2383,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3473,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1937,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.179,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3407,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.234,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2815,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2374,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.324,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2218,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2269,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2384,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2838,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2258,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2418,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2726,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2109,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2446,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2056,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2789,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1937,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.259,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3414,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.3698,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.254,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.2136,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1836,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2296,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2698,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3091,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.238,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1981,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2113,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2086,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2627,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2546,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2699,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1641,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2903,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2415,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2368,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2614,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1974,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2498,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2346,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2926,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2219,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2368,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2086,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2125,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3331,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2944,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1987,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3175,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3064,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2034,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2427,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2622,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2339,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2291,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2154,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2274,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2099,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2239,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.3123,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2597,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2678,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2594,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2682,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.277,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2656,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 0.2465,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2086,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1767,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1868,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2528,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2516,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.232,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2196,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.4016,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2089,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2737,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2819,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1663,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2408,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.295,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1793,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2311,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.204,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2949,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2925,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2394,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2541,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2216,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2468,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2008,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2122,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2181,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2469,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2246,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2837,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2311,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.3687,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.251,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2068,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2435,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2717,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2576,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2162,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2233,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.248,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2877,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2072,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.2144,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1719,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2455,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.27,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2109,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2488,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2131,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2318,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2772,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1509,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.175,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1948,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2939,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2511,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2013,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1541,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2239,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2424,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3108,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2549,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.213,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2535,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2881,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2495,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1796,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.29,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.201,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2098,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2581,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1931,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1835,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1765,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2452,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.3002,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2466,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2106,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2084,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2018,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2073,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1701,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2477,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.224,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2536,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2609,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.236,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2229,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2e-05,
+      "loss": 0.2448,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2556,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1819,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2297,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.262,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2692,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2294,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2795,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2336,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1864,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2677,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2664,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.241,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2185,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2353,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1913,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2774,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1817,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2044,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2437,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2402,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.17,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2004,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2366,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2584,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.275,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.3667,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1733,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2411,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2334,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1949,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2262,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2234,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.314,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2527,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2644,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2346,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2119,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.255,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.234,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.193,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2419,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2587,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2606,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.2125,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1836,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2306,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2547,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1949,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2526,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2289,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.266,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2457,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2433,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2531,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2292,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2734,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1887,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1896,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2286,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.197,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2213,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2197,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2458,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3147,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2128,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2275,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.3022,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2723,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2383,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.232,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2307,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2125,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2271,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2926,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2639,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2034,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2637,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2132,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2179,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.199,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.261,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1879,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2808,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.16,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.2759,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1912,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.257,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2772,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2466,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2839,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2461,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1756,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2114,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2375,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2124,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2287,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.219,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2548,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1735,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2609,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2106,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2924,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2606,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2438,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2646,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2346,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2661,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2583,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2589,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2323,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2342,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1333,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2012,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2638,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2349,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2285,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2808,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1854,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1859,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1827,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.222,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1939,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2038,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2256,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1545,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2093,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2357,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2006,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.267,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2189,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2e-05,
+      "loss": 0.2646,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2207,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2214,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2228,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1932,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1918,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2745,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2849,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2556,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2548,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2816,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2847,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2684,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2741,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2866,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.3362,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2189,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2347,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2007,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.181,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.235,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2155,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2022,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2399,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.17,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2036,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1759,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2304,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.241,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2369,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2805,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1947,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2876,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2232,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2457,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1996,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2815,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.23,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2325,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1887,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2514,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2093,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.189,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2e-05,
+      "loss": 0.2083,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2195,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.184,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2823,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2315,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1663,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2141,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.226,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1924,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2637,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2624,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2328,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2119,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1732,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2729,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1736,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1909,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1895,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2345,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.28,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1967,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2287,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1869,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2061,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2532,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2452,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2272,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2053,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2461,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2679,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2273,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2669,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2563,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1962,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2917,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2139,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2676,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2283,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1923,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.3163,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.2285,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.219,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.256,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2693,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2287,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1486,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2157,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2464,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1636,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2196,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2928,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2602,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2077,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2318,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.211,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1932,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.212,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2966,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1735,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1651,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2351,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2067,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1502,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.3219,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2006,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2407,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2208,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2792,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.231,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2234,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.159,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1298,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1918,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1882,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1689,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1892,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2217,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1862,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1624,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1981,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2807,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1981,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2e-05,
+      "loss": 0.2231,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2379,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2215,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2246,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.3055,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.22,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.224,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2242,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2136,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2159,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1668,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2004,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1444,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2457,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2096,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2182,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2341,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.139,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2205,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1796,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2149,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2248,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1907,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2189,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1727,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2178,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2195,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2133,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1563,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2561,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2415,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2482,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2339,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2177,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1938,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2012,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1705,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1948,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1701,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2493,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2227,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2565,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2501,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2e-05,
+      "loss": 0.2759,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1837,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2068,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2441,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2655,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2197,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2241,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2377,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2502,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1898,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1865,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2532,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2172,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2237,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2827,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2297,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2103,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2141,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2368,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1503,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2628,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1744,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2542,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2231,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.172,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2426,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1364,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1906,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.162,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2159,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1748,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1735,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2501,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2505,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2092,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1695,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2245,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1788,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1688,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1857,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.2882,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1795,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1496,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1602,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2451,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2198,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1624,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2046,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1785,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1874,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1967,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1697,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1878,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2062,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1679,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2458,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1907,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2065,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1812,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2534,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1925,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1927,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1928,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2162,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2118,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2082,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2335,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2399,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2023,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2441,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2168,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2048,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2525,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2394,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2075,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.17,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2191,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2599,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.2978,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2e-05,
+      "loss": 0.217,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2037,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1835,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2467,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2687,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2078,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1798,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2679,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1771,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1775,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2169,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.233,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.205,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2161,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2331,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1906,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1709,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2205,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2485,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2194,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1509,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1572,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1431,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1681,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2142,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2736,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1864,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2247,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2212,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1867,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.185,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1899,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2466,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1959,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2147,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2257,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2562,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2077,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1961,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2e-05,
+      "loss": 0.2473,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1809,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.214,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.183,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2278,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2384,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2248,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2095,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2423,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2512,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2145,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1535,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2257,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2607,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1783,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2185,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1973,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1848,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1716,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1819,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2338,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1903,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1816,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2021,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2279,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2133,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2099,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1921,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1708,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2001,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1714,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2149,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.229,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2112,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2006,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1702,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1355,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2518,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1956,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2085,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1741,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1991,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.2476,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1913,
+      "step": 1591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2212,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1809,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2221,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.134,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.249,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2074,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2156,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1774,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1979,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1829,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1849,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2142,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2173,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1981,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2102,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1721,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1777,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.244,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2004,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1927,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1564,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2106,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2184,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2323,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.3038,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1983,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1943,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.174,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2457,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1692,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2593,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2009,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2178,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1611,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1685,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2106,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2007,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2012,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.181,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2037,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2015,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2e-05,
+      "loss": 0.2206,
+      "step": 1639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2141,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2382,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1943,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2009,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2236,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.23,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1736,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2282,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1613,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2765,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2468,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1621,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1473,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1924,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2145,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1748,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1552,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1468,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1979,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1618,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1593,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2908,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2044,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1842,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2264,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1342,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.22,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1786,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2548,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1754,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1419,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1469,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1895,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2445,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1451,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1533,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1737,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2241,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2706,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2227,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2e-05,
+      "loss": 0.2591,
+      "step": 1686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.204,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1874,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2178,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1664,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1819,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1845,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1923,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.244,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.14,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2126,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1299,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1688,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1523,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2039,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1532,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.202,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2181,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.3066,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.234,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2451,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2148,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1665,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1903,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2008,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1605,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2371,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1444,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1758,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1971,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1944,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1847,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1992,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2053,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1771,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.256,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2413,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.2062,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1298,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1698,
+      "step": 1734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.21,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2517,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1853,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2243,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2437,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2514,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2325,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1525,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2482,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1954,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1672,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1643,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1818,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2642,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2266,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.3315,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1622,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2063,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1598,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2194,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2252,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1801,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2198,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2177,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1422,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2163,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1958,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1668,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1342,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1538,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1649,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1921,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.19,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1532,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1966,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2364,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1793,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1613,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1695,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1915,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.2058,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1529,
+      "step": 1781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2318,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1616,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2001,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2037,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2112,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2081,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1492,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2047,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.187,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2177,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.257,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1771,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1903,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.16,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2458,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2209,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2588,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2048,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1702,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1814,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2116,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1418,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1542,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1846,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2565,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1818,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2584,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1734,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2222,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1606,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.21,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.252,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.2583,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1884,
+      "step": 1829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2173,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1966,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.162,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1628,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2117,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2147,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2052,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2649,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1775,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.191,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1298,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1725,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.197,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1788,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2094,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2172,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1413,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2429,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1732,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2024,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1911,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1736,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2071,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2803,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2649,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2671,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1799,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.18,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1662,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2148,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1359,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2291,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2044,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1703,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1554,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.2321,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1854,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1972,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2e-05,
+      "loss": 0.194,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.172,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2078,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1695,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2548,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1888,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1439,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2341,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1864,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.164,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2699,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2039,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2216,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1885,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1627,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2232,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.196,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1643,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2341,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1763,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1965,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1865,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2086,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2123,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1805,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2182,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1556,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1874,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1732,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1555,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1788,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1894,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.154,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2023,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.26,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.2114,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1514,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.192,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1634,
+      "step": 1924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1456,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2073,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.21,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1922,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2589,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.196,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2356,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1441,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.178,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2028,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1576,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1863,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1692,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2223,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2388,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.231,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1949,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.162,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1196,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1777,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1937,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1408,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2269,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2034,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1769,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2488,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1708,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2375,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2297,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1924,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1632,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1883,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1927,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.2061,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 1971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2283,
+      "step": 1972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2094,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1844,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.189,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 1976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 1977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1741,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.138,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2075,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.201,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1686,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1624,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1859,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2259,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1603,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1812,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1731,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1495,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1882,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1202,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.3179,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.195,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.246,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1934,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1786,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1886,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1528,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2203,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1725,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2316,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.207,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1774,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1874,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1688,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2122,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2404,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.197,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2148,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1887,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2042,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1863,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2e-05,
+      "loss": 0.2137,
+      "step": 2019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1348,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2072,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1617,
+      "step": 2023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2629,
+      "step": 2024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1505,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1841,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.178,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2067,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.154,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2168,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1824,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1657,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2034,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1541,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2218,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1877,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.254,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2309,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.173,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1349,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2394,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1896,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2199,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2278,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1455,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1456,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1938,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1545,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1698,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1903,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2272,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1773,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1546,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2e-05,
+      "loss": 0.2178,
+      "step": 2066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1548,
+      "step": 2067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2664,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1773,
+      "step": 2069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2586,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2446,
+      "step": 2071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1473,
+      "step": 2072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1724,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1585,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2428,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.16,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.181,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1858,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1966,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1724,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2139,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1349,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2125,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2192,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.188,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.16,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2082,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1545,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1902,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1827,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2115,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1995,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2083,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1725,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1706,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1978,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1511,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1595,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.2073,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1904,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1475,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1439,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1986,
+      "step": 2114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1505,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 2116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.204,
+      "step": 2117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1994,
+      "step": 2118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1511,
+      "step": 2119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.139,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2195,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1611,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1587,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1425,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1842,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1864,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1537,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2256,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2197,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1688,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1642,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2496,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.134,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1433,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1597,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1415,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1586,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1825,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1358,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2365,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2094,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1404,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1704,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.212,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2393,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2411,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1629,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1836,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2698,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.2033,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 2161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 2162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1445,
+      "step": 2163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1948,
+      "step": 2164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1598,
+      "step": 2166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1512,
+      "step": 2167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 2168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1758,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2051,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1872,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1759,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2229,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1447,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1726,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2122,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.175,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1508,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1736,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1878,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1831,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.226,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1829,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1948,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1932,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1521,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.184,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1819,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1544,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2319,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.175,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2079,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1976,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 2208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2e-05,
+      "loss": 0.2003,
+      "step": 2209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2209,
+      "step": 2211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1957,
+      "step": 2212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1462,
+      "step": 2213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 2214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1673,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1709,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2155,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1696,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1614,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1504,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2068,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1549,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1359,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1585,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1406,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1835,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1815,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2321,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2053,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1802,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1738,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.2008,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1406,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1925,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1514,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1726,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1924,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1476,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1677,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.242,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1982,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1424,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1372,
+      "step": 2256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1725,
+      "step": 2257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2139,
+      "step": 2258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1712,
+      "step": 2259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2195,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1506,
+      "step": 2261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 2262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1453,
+      "step": 2263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2338,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.155,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1912,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1754,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1558,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2546,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1843,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1536,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1887,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1757,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1954,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1804,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1833,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2257,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1718,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1595,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1672,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1349,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1823,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1663,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2096,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2321,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.111,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.179,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1568,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1706,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2434,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1508,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.214,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1184,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1575,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.2191,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 2304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1687,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1716,
+      "step": 2306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1893,
+      "step": 2307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1616,
+      "step": 2308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2466,
+      "step": 2309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.185,
+      "step": 2311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1526,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1703,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2109,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1967,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2203,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1466,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1681,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1586,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2551,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1971,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2317,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2303,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1384,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1654,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1484,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1825,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2083,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1648,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.205,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1713,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.201,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.2512,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1984,
+      "step": 2351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 2352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1747,
+      "step": 2353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1666,
+      "step": 2354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.3026,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1849,
+      "step": 2356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1192,
+      "step": 2357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 2358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1825,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.23,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2247,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.213,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1754,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1506,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.131,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2159,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2393,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2376,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1672,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2127,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2061,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1413,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.2307,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1622,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1475,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1447,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.182,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1625,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1631,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.138,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1973,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1757,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1853,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1454,
+      "step": 2399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1852,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2333,
+      "step": 2401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 2402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 2403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 2404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1654,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.187,
+      "step": 2406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2002,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1558,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1859,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1757,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2691,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1953,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1612,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1366,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1431,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1629,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2003,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2056,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2193,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1275,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1778,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2584,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.201,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1595,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1633,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1663,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1873,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2046,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1279,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2164,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1626,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.2058,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1857,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1664,
+      "step": 2446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1749,
+      "step": 2447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1574,
+      "step": 2448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.2307,
+      "step": 2449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1677,
+      "step": 2451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.2291,
+      "step": 2452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1953,
+      "step": 2453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1647,
+      "step": 2454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.2305,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1824,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1848,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.159,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1685,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1644,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1673,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1642,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1544,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1684,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1511,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1943,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1606,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1913,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1889,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1554,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1979,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1366,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1531,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1775,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1833,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1853,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1492,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1944,
+      "step": 2493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 2494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1531,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.142,
+      "step": 2496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1692,
+      "step": 2497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1738,
+      "step": 2498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1395,
+      "step": 2499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1557,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1658,
+      "step": 2501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1672,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1606,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1642,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1892,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.243,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.162,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1375,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.155,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2021,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1181,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2168,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1704,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1594,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1681,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1708,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.195,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1755,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1501,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2158,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1716,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2651,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2339,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.145,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2014,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.2347,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1767,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.134,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 2541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.159,
+      "step": 2542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1686,
+      "step": 2543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1558,
+      "step": 2544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.217,
+      "step": 2546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 2547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.222,
+      "step": 2548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1939,
+      "step": 2549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2123,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1138,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1848,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2537,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1365,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1386,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1466,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.159,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1569,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1439,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1688,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1644,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2859,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.168,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1904,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2524,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1497,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1737,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.173,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1877,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1287,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1881,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.2195,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1421,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 2588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 2589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 2590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2211,
+      "step": 2591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1861,
+      "step": 2592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 2593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1951,
+      "step": 2594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1628,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2324,
+      "step": 2596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.165,
+      "step": 2597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1294,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1557,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2428,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1444,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.157,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1901,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1511,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2642,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2319,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1945,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1757,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1745,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1931,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.222,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.172,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.148,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1422,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.2488,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1476,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1685,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1938,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1824,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1618,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.138,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1678,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1692,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 2636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1815,
+      "step": 2637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 2638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1542,
+      "step": 2639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2284,
+      "step": 2641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 2642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1419,
+      "step": 2643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2334,
+      "step": 2644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1885,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1651,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2003,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2314,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1599,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1724,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1004,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.177,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1461,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1594,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1314,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1685,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1893,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1551,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1898,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1696,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1407,
+      "step": 2683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1645,
+      "step": 2684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 2685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 2686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 2687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.2487,
+      "step": 2688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1736,
+      "step": 2689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1276,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1481,
+      "step": 2691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 2692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1579,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1257,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.226,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1746,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1305,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1852,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1799,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1455,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.2481,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1551,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1768,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1713,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1383,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.189,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.2063,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1679,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1749,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1605,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.2234,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1754,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.183,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1114,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.218,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1947,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 2731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.2375,
+      "step": 2732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 2733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1265,
+      "step": 2734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.2056,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1783,
+      "step": 2736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1441,
+      "step": 2737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1864,
+      "step": 2738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1369,
+      "step": 2739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 2740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.141,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1733,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1911,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1433,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1515,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1877,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1481,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1438,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1853,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.2363,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1903,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1932,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1925,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1959,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1287,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.157,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1633,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0971,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1852,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1625,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1503,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1293,
+      "step": 2778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1652,
+      "step": 2779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.174,
+      "step": 2780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 2781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1354,
+      "step": 2782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 2783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 2784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1352,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1451,
+      "step": 2786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1869,
+      "step": 2787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1599,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1504,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1877,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1683,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.197,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1718,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1504,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1473,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1875,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1868,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.2035,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1038,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1326,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1302,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1508,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1304,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1849,
+      "step": 2826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1943,
+      "step": 2827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2142,
+      "step": 2828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 2829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1422,
+      "step": 2831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 2832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.307,
+      "step": 2833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2059,
+      "step": 2834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2092,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1886,
+      "step": 2836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2034,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1514,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1415,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1506,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1744,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1566,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1659,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1895,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1342,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2108,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1712,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1582,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1121,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1445,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1222,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1482,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.2045,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1447,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1815,
+      "step": 2874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 2875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1686,
+      "step": 2876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 2877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1936,
+      "step": 2878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 2879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.191,
+      "step": 2880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 2881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 2882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 2883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 2884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1525,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.141,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1779,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.2027,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1734,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1404,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1449,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1646,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1593,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1514,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1548,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1225,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1538,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1859,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1552,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1991,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1512,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.161,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1602,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.2175,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1795,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1838,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1638,
+      "step": 2921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 2922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1454,
+      "step": 2923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1808,
+      "step": 2924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1631,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 2926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1646,
+      "step": 2927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 2928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1542,
+      "step": 2929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 2930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.224,
+      "step": 2931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1844,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.18,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2083,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1481,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1824,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2607,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1646,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2117,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1944,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.188,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1757,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.2166,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1843,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.199,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.181,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.167,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1789,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 2968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 2969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1633,
+      "step": 2970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1138,
+      "step": 2971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1408,
+      "step": 2972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2031,
+      "step": 2973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.182,
+      "step": 2974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 2975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 2976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.154,
+      "step": 2977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 2978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1865,
+      "step": 2979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.205,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2538,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2096,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1266,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.2198,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1612,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.169,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1634,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1854,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1691,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1715,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1913,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1783,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1765,
+      "step": 3004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1467,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1603,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1451,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 3016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2148,
+      "step": 3017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1573,
+      "step": 3018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1466,
+      "step": 3019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1733,
+      "step": 3020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 3021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 3022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 3023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 3024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 3025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 3026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 3027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1302,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2142,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1447,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1631,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1486,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2365,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1848,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2103,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2074,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1333,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1782,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.174,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1545,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 3052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1468,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1533,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1628,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1515,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.2254,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1476,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1365,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 3063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1993,
+      "step": 3064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1798,
+      "step": 3065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 3066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1496,
+      "step": 3067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 3068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1462,
+      "step": 3069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1804,
+      "step": 3070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 3071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1742,
+      "step": 3072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1724,
+      "step": 3073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 3074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1551,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.2181,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1202,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.185,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1515,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1753,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1261,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1823,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1444,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.145,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.2185,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 3099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1247,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1616,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1882,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1567,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1805,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1823,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1264,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1544,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1992,
+      "step": 3111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2128,
+      "step": 3112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1615,
+      "step": 3113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1305,
+      "step": 3114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1964,
+      "step": 3115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 3116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1206,
+      "step": 3117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 3118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2646,
+      "step": 3119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 3120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 3121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1541,
+      "step": 3122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1582,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1518,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1408,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2212,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1612,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1461,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2419,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1848,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1869,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2174,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1824,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2178,
+      "step": 3146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1552,
+      "step": 3147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 3148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1573,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1611,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.2399,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1482,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1459,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1355,
+      "step": 3158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 3159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 3160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 3161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 3162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 3163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2298,
+      "step": 3164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 3165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1295,
+      "step": 3166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1548,
+      "step": 3167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.19,
+      "step": 3168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1721,
+      "step": 3169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 3170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2139,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1446,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1854,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1081,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1837,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1414,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2092,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1893,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1225,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1594,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1314,
+      "step": 3194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1977,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1509,
+      "step": 3196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1659,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1478,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1833,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.2227,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1583,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1463,
+      "step": 3206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1501,
+      "step": 3207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1606,
+      "step": 3208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1433,
+      "step": 3209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 3210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1768,
+      "step": 3211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 3212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1502,
+      "step": 3213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 3214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1878,
+      "step": 3215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1531,
+      "step": 3216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1761,
+      "step": 3217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1894,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1462,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.2336,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1786,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1966,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1714,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1984,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1634,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 3241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1285,
+      "step": 3242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1733,
+      "step": 3243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1265,
+      "step": 3244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.2393,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1847,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1575,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1196,
+      "step": 3253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 3254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 3255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 3256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1217,
+      "step": 3257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 3258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.155,
+      "step": 3259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 3260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 3261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 3262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1518,
+      "step": 3263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2233,
+      "step": 3264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.11,
+      "step": 3265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1407,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2121,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2372,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2121,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1464,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2241,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1888,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1958,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2166,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1572,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1529,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1418,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1226,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1506,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2015,
+      "step": 3288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 3289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1253,
+      "step": 3290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1821,
+      "step": 3291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.2427,
+      "step": 3292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.128,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1722,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1404,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1425,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1552,
+      "step": 3301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 3302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1198,
+      "step": 3303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1446,
+      "step": 3304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3234,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1461,
+      "step": 3306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 3307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1375,
+      "step": 3308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1988,
+      "step": 3309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1656,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.147,
+      "step": 3311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1533,
+      "step": 3312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 3313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1463,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1549,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1291,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.2005,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1395,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1532,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1441,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1616,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1648,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1896,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1728,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1615,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1184,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1578,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1175,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.2154,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1555,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1798,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1638,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1542,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 3348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1799,
+      "step": 3349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 3350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 3351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1786,
+      "step": 3352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 3353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1478,
+      "step": 3354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1275,
+      "step": 3355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 3356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1768,
+      "step": 3357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1668,
+      "step": 3358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 3359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 3360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1159,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.157,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1753,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1364,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.2543,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.148,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1453,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1454,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.2228,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1565,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1941,
+      "step": 3383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 3384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1852,
+      "step": 3385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1985,
+      "step": 3386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 3387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1672,
+      "step": 3388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1081,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1821,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.166,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1604,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 3396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 3397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1618,
+      "step": 3398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 3399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.199,
+      "step": 3400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 3401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 3402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 3403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 3404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 3405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 3406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 3407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 3408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1608,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1275,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1198,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1121,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1664,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1684,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.2074,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1646,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1335,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1727,
+      "step": 3430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 3431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 3432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 3433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 3434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 3435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 3436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1718,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1443,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 3443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 3444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 3445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 3446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1418,
+      "step": 3447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 3448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1636,
+      "step": 3449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.21,
+      "step": 3450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 3451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 3452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.2032,
+      "step": 3453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 3454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 3455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 3456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1478,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1794,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1528,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.2145,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1583,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1258,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1484,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1293,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1495,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1993,
+      "step": 3478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1906,
+      "step": 3479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1459,
+      "step": 3480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 3481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.169,
+      "step": 3482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1297,
+      "step": 3483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1264,
+      "step": 3484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1487,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1598,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1668,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1372,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1921,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 3491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 3492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1664,
+      "step": 3493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1557,
+      "step": 3494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1258,
+      "step": 3495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 3496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1484,
+      "step": 3497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1299,
+      "step": 3498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 3499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 3500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1307,
+      "step": 3501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1295,
+      "step": 3502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 3503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 3504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.195,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1594,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1844,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1631,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1698,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1303,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1453,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1844,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1299,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.2011,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1776,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1872,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1673,
+      "step": 3525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 3526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 3527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 3528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 3529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 3530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 3531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1652,
+      "step": 3532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1924,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1721,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1505,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 3538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 3539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 3540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1072,
+      "step": 3541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1375,
+      "step": 3542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1485,
+      "step": 3543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 3544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.178,
+      "step": 3545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 3546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 3547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 3548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1991,
+      "step": 3549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 3550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 3551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.2263,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.2078,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1586,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1616,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1659,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1471,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1863,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1275,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1531,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 3572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 3573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 3574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 3575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 3576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 3577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 3578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 3579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1328,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1647,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1546,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1881,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1355,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1291,
+      "step": 3586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1535,
+      "step": 3587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1815,
+      "step": 3588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1898,
+      "step": 3589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 3590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 3591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 3592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1431,
+      "step": 3593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 3594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 3595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1782,
+      "step": 3596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 3597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 3598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 3599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2629,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1814,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1729,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1838,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1625,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1287,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2411,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2058,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2107,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1239,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1415,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 3619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.2351,
+      "step": 3620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 3621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1295,
+      "step": 3622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1535,
+      "step": 3623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1906,
+      "step": 3624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 3625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1556,
+      "step": 3626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1538,
+      "step": 3627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1197,
+      "step": 3628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 3629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.19,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 3633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1625,
+      "step": 3634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.141,
+      "step": 3635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1629,
+      "step": 3636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.186,
+      "step": 3637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1325,
+      "step": 3638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1725,
+      "step": 3639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2059,
+      "step": 3640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1726,
+      "step": 3641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 3642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1634,
+      "step": 3643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 3644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 3645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1342,
+      "step": 3646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1909,
+      "step": 3647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1869,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.179,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1283,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.18,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1568,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1748,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.147,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1958,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1839,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1897,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1349,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2292,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 3667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1501,
+      "step": 3668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 3669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1182,
+      "step": 3670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1493,
+      "step": 3671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1838,
+      "step": 3672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0977,
+      "step": 3673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 3674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 3675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2317,
+      "step": 3676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.2086,
+      "step": 3677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.148,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1607,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1471,
+      "step": 3681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 3682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 3683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 3684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 3685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2355,
+      "step": 3686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1411,
+      "step": 3687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 3688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 3689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 3690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0985,
+      "step": 3691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 3692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1747,
+      "step": 3693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1291,
+      "step": 3694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1383,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1633,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1515,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1283,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2209,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 3714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 3715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 3716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.2179,
+      "step": 3717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 3718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1735,
+      "step": 3719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1651,
+      "step": 3720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1496,
+      "step": 3721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 3722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.187,
+      "step": 3723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 3724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1649,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 3728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 3729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2422,
+      "step": 3730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 3731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1781,
+      "step": 3732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 3733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2261,
+      "step": 3734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 3735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.158,
+      "step": 3736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 3737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 3738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1879,
+      "step": 3739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 3740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 3741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2137,
+      "step": 3742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1939,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1453,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.133,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2174,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.2012,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1804,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1451,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1751,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1766,
+      "step": 3761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 3762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1594,
+      "step": 3763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 3764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 3765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 3766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1797,
+      "step": 3767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 3768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.201,
+      "step": 3769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1905,
+      "step": 3770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 3771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1414,
+      "step": 3772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1996,
+      "step": 3773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1863,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 3776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 3777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 3778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1655,
+      "step": 3779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1293,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 3781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 3782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 3783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1526,
+      "step": 3784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1364,
+      "step": 3785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 3786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1831,
+      "step": 3787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 3788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1769,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.2277,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1481,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1251,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1406,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1303,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1455,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1812,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1571,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.18,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1359,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1332,
+      "step": 3809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1338,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 3811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 3812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 3813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1456,
+      "step": 3814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 3816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 3817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 3818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 3819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1833,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.141,
+      "step": 3821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1896,
+      "step": 3823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1318,
+      "step": 3824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 3825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 3826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 3827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 3828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1683,
+      "step": 3829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 3830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 3831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.122,
+      "step": 3832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 3833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 3834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 3835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 3836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 3837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1659,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2052,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.166,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2222,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.164,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1573,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1901,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1359,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1599,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.167,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1693,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 3856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 3857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1701,
+      "step": 3858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1426,
+      "step": 3859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 3860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 3861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 3862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1503,
+      "step": 3863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 3864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1266,
+      "step": 3865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.2209,
+      "step": 3866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 3867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 3868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 3869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1408,
+      "step": 3871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1637,
+      "step": 3872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.244,
+      "step": 3873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 3874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1752,
+      "step": 3875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1442,
+      "step": 3876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 3877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 3878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1935,
+      "step": 3879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1912,
+      "step": 3880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.2003,
+      "step": 3881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 3882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 3883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 3884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 3885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1756,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1868,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1798,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1689,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1348,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.189,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1271,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1777,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1414,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1406,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 3903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 3904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 3905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 3906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 3907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 3908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.16,
+      "step": 3909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 3910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 3911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 3912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.191,
+      "step": 3913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 3914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 3915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 3916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 3917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1862,
+      "step": 3918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 3919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1658,
+      "step": 3920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1634,
+      "step": 3921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 3922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 3923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 3924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 3925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1622,
+      "step": 3926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 3927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1803,
+      "step": 3928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 3929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 3930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.2291,
+      "step": 3931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 3932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 3933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1352,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1221,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1727,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.122,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1579,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.111,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1781,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1636,
+      "step": 3951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 3952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 3953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1233,
+      "step": 3954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1919,
+      "step": 3955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 3956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1796,
+      "step": 3957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 3958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1563,
+      "step": 3959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1982,
+      "step": 3960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1587,
+      "step": 3961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 3962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1004,
+      "step": 3963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 3964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1257,
+      "step": 3966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1138,
+      "step": 3967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1332,
+      "step": 3968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 3969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 3970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 3971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1694,
+      "step": 3972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1526,
+      "step": 3973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1748,
+      "step": 3974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1469,
+      "step": 3975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 3976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1748,
+      "step": 3977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1175,
+      "step": 3978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1279,
+      "step": 3979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 3980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.152,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1213,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1617,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1759,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1253,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1544,
+      "step": 3998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1217,
+      "step": 3999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1443,
+      "step": 4000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1709,
+      "step": 4001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1437,
+      "step": 4002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 4003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 4004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 4005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 4006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 4007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.139,
+      "step": 4008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1707,
+      "step": 4009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1255,
+      "step": 4010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.217,
+      "step": 4011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1871,
+      "step": 4012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1718,
+      "step": 4013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 4014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1829,
+      "step": 4015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 4016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 4017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 4018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 4019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1685,
+      "step": 4020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.141,
+      "step": 4021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1532,
+      "step": 4022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1629,
+      "step": 4023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1689,
+      "step": 4024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 4025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 4026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 4027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1929,
+      "step": 4028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1767,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1603,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1341,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1753,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1187,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1421,
+      "step": 4045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1435,
+      "step": 4046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 4047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 4048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1233,
+      "step": 4049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.139,
+      "step": 4050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1422,
+      "step": 4051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 4052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1613,
+      "step": 4053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1771,
+      "step": 4054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1484,
+      "step": 4055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.2077,
+      "step": 4056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.159,
+      "step": 4057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1721,
+      "step": 4058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 4059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1359,
+      "step": 4060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1892,
+      "step": 4061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1666,
+      "step": 4062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1365,
+      "step": 4063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.203,
+      "step": 4064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 4065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 4066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 4067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1549,
+      "step": 4068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 4069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1603,
+      "step": 4070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 4071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1255,
+      "step": 4072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1509,
+      "step": 4073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 4074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 4075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1732,
+      "step": 4076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1744,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1492,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1638,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 4093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 4094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 4095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1464,
+      "step": 4096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 4097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 4098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1251,
+      "step": 4099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1509,
+      "step": 4100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.2084,
+      "step": 4101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.2007,
+      "step": 4102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 4103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 4104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 4105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1668,
+      "step": 4106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.148,
+      "step": 4107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 4108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1493,
+      "step": 4109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 4110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0958,
+      "step": 4111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 4112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1004,
+      "step": 4113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1549,
+      "step": 4114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1959,
+      "step": 4115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 4116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 4117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 4118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 4119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1683,
+      "step": 4120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1774,
+      "step": 4121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1742,
+      "step": 4122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1861,
+      "step": 4123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.175,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1689,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1148,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1133,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1361,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1235,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1939,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 4140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.25,
+      "step": 4141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1562,
+      "step": 4142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 4143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 4144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 4145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1267,
+      "step": 4146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 4147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 4148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.243,
+      "step": 4149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1788,
+      "step": 4150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1061,
+      "step": 4151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 4152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1506,
+      "step": 4153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 4154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 4155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.128,
+      "step": 4156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1622,
+      "step": 4157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 4158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 4159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 4160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1525,
+      "step": 4161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 4162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1488,
+      "step": 4163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1731,
+      "step": 4164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1148,
+      "step": 4165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1724,
+      "step": 4166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2075,
+      "step": 4167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.176,
+      "step": 4168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 4169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 4170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1567,
+      "step": 4171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1522,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.14,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.14,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2323,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.2107,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1239,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1628,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 4187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 4188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1648,
+      "step": 4189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 4190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 4191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 4192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 4193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 4194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1912,
+      "step": 4195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1982,
+      "step": 4196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 4197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 4198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 4199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 4200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 4201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1411,
+      "step": 4202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1706,
+      "step": 4203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 4204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 4205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1631,
+      "step": 4206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1967,
+      "step": 4207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 4208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 4209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 4210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 4211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1384,
+      "step": 4212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1916,
+      "step": 4213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 4214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 4215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 4216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 4217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1564,
+      "step": 4218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1832,
+      "step": 4219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1623,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.171,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1642,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1582,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1803,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 4235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 4236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1652,
+      "step": 4237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1424,
+      "step": 4238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 4239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 4240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1521,
+      "step": 4241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 4242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.2026,
+      "step": 4243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 4244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 4245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1691,
+      "step": 4246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 4247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 4248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 4249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1499,
+      "step": 4250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 4251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 4252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 4253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1528,
+      "step": 4254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1611,
+      "step": 4255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.131,
+      "step": 4256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 4257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 4258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 4259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1229,
+      "step": 4261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1755,
+      "step": 4262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 4263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 4264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 4265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1512,
+      "step": 4266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 4267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1855,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1904,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1516,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1926,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1415,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1354,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1463,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 4282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.2148,
+      "step": 4283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 4284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 4285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 4286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1266,
+      "step": 4287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 4288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 4289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 4290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 4291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 4292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 4293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1815,
+      "step": 4294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1546,
+      "step": 4295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1791,
+      "step": 4296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 4297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 4298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 4299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1081,
+      "step": 4300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 4301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 4302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 4303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 4304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 4305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 4306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 4307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 4308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1731,
+      "step": 4309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 4310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 4311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.142,
+      "step": 4312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 4313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1945,
+      "step": 4314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1417,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.2293,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1198,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1871,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1101,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1739,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0934,
+      "step": 4329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1681,
+      "step": 4330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 4331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 4332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 4333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 4334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 4335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 4336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1582,
+      "step": 4337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 4338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1649,
+      "step": 4339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 4340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 4341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1593,
+      "step": 4342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 4343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 4344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 4345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 4346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 4347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 4348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 4349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 4350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 4351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1574,
+      "step": 4352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.2191,
+      "step": 4353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1402,
+      "step": 4354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 4355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 4356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 4357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1761,
+      "step": 4358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1548,
+      "step": 4359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1526,
+      "step": 4360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1191,
+      "step": 4361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0985,
+      "step": 4362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1641,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1729,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1703,
+      "step": 4377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 4378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 4379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 4380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 4381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 4382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1888,
+      "step": 4383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1884,
+      "step": 4384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1609,
+      "step": 4385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 4386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1671,
+      "step": 4387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 4388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 4389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 4390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 4391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1217,
+      "step": 4392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 4393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2e-05,
+      "loss": 0.2079,
+      "step": 4394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1402,
+      "step": 4395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 4396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.2054,
+      "step": 4397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 4398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 4399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1895,
+      "step": 4400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 4401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 4402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 4403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1945,
+      "step": 4404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 4405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1514,
+      "step": 4406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 4407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 4408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 4409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 4410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1402,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1626,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1902,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1641,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1556,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1747,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 4424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 4425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 4426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1239,
+      "step": 4427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1628,
+      "step": 4428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1787,
+      "step": 4429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1951,
+      "step": 4430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 4431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 4432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 4433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1565,
+      "step": 4434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1502,
+      "step": 4435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.122,
+      "step": 4436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1258,
+      "step": 4437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 4438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 4439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 4440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.2063,
+      "step": 4441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 4442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1898,
+      "step": 4443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 4444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 4445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 4446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1297,
+      "step": 4447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 4448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1553,
+      "step": 4449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 4450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1667,
+      "step": 4451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 4452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 4453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 4454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 4455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 4456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 4457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1411,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2016,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2246,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.155,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1567,
+      "step": 4471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 4472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 4473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 4474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 4475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 4476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1307,
+      "step": 4477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1416,
+      "step": 4478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 4479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 4480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 4481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1482,
+      "step": 4482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1745,
+      "step": 4483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.2049,
+      "step": 4484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1737,
+      "step": 4485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1286,
+      "step": 4486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 4487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1817,
+      "step": 4488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 4489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 4490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 4491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 4492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 4493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 4494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 4495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 4496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 4497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 4498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1618,
+      "step": 4499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 4500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 4501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.2314,
+      "step": 4502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1598,
+      "step": 4503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 4504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 4505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1253,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1564,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.147,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1583,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1597,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1437,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 4519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 4520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 4521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 4522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 4523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 4524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 4525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 4526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1474,
+      "step": 4527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 4528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1573,
+      "step": 4529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 4530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 4531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 4532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 4533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1598,
+      "step": 4534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 4535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 4536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 4537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1437,
+      "step": 4538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 4539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 4540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 4541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 4542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 4543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 4544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1554,
+      "step": 4545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 4546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.19,
+      "step": 4547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1297,
+      "step": 4548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 4549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1638,
+      "step": 4550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 4551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 4552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1091,
+      "step": 4553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1556,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.143,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1639,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1691,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.157,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1358,
+      "step": 4566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 4567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1447,
+      "step": 4568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1666,
+      "step": 4569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1591,
+      "step": 4570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 4571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 4572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 4573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1351,
+      "step": 4574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 4575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1995,
+      "step": 4576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 4577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 4578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 4579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 4580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1613,
+      "step": 4581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 4582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 4583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1175,
+      "step": 4584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 4585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.2045,
+      "step": 4586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1114,
+      "step": 4587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 4588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 4589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 4590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1291,
+      "step": 4591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1696,
+      "step": 4592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 4593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1328,
+      "step": 4594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 4595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1792,
+      "step": 4596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 4597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1614,
+      "step": 4598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 4599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 4600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.166,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1402,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1475,
+      "step": 4613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1567,
+      "step": 4614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 4615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1866,
+      "step": 4616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 4617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1222,
+      "step": 4618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 4619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 4620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1452,
+      "step": 4621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 4622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 4623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.151,
+      "step": 4624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 4625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1772,
+      "step": 4626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1677,
+      "step": 4627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 4628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 4629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 4630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 4631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 4632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 4633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1406,
+      "step": 4634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1674,
+      "step": 4635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 4636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1332,
+      "step": 4637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 4638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1641,
+      "step": 4639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 4640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.2129,
+      "step": 4641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 4642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 4643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 4644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 4645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 4646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 4647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1762,
+      "step": 4648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1251,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1332,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1454,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1626,
+      "step": 4661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 4662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 4663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1526,
+      "step": 4664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 4665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 4666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 4667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 4668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1501,
+      "step": 4669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 4670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1369,
+      "step": 4671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 4672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 4673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1375,
+      "step": 4674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 4675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1495,
+      "step": 4676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1385,
+      "step": 4677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 4678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 4679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 4680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2017,
+      "step": 4681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 4682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 4683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 4684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 4685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1978,
+      "step": 4686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 4687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 4688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 4689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1264,
+      "step": 4690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 4691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 4692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 4693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 4694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 4695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 4696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0955,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1518,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1299,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1723,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1636,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.163,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.169,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1301,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 4708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 4709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 4710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 4711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2181,
+      "step": 4712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 4713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 4714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 4715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1633,
+      "step": 4716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1533,
+      "step": 4717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 4718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 4719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1261,
+      "step": 4720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.188,
+      "step": 4721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.2247,
+      "step": 4722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 4723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 4724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 4725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1613,
+      "step": 4726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1512,
+      "step": 4727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 4728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 4729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 4730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 4731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 4732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 4733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 4734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 4735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.168,
+      "step": 4736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.179,
+      "step": 4737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.2138,
+      "step": 4738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 4739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 4740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 4741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 4742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 4743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 4744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 4745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 4746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 4747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 4748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 4749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1299,
+      "step": 4750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 4751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1675,
+      "step": 4752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1257,
+      "step": 4753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 4754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 4755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 4756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1376,
+      "step": 4757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 4758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 4759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 4760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 4761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1138,
+      "step": 4762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 4763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 4764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 4765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 4766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 4767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 4768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 4769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 4770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 4771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 4772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 4773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 4774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 4775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 4776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 4777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 4778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 4779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 4780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 4781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 4782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 4783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 4784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1676,
+      "step": 4785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 4786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 4787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 4788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 4789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1912,
+      "step": 4790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 4791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1196,
+      "step": 4792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 4793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 4794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 4795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 4796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 4797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 4798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 4799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 4800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 4801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 4802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 4803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 4804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 4805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.135,
+      "step": 4806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1283,
+      "step": 4807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 4808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 4809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 4810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 4811
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 4812
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1311,
+      "step": 4813
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 4814
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 4815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 4816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 4817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 4818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 4819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 4820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 4821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 4822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 4823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 4824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 4825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 4826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 4827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 4828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1537,
+      "step": 4829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 4830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 4831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 4832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 4833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 4834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 4835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 4836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0925,
+      "step": 4837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 4838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 4839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 4840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 4841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 4842
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 4843
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 4844
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 4845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 4846
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 4847
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 4848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 4849
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 4850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 4851
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 4852
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 4853
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 4854
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 4855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 4856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 4857
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 4858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 4859
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 4860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 4861
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 4862
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 4863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 4864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 4865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 4866
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 4867
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 4868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 4869
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.178,
+      "step": 4870
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 4871
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 4872
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 4873
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 4874
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 4875
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 4876
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 4877
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 4878
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 4879
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 4880
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 4881
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 4882
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 4883
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 4884
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 4885
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 4886
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 4887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 4888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1253,
+      "step": 4889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 4890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 4891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 4892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 4893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 4894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 4895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 4896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 4897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 4898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 4899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 4900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 4901
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1257,
+      "step": 4902
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 4903
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 4904
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 4905
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 4906
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1294,
+      "step": 4907
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 4908
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 4909
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 4910
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 4911
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 4912
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 4913
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 4914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 4915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.1425,
+      "step": 4916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 4917
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1417,
+      "step": 4918
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 4919
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 4920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 4921
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 4922
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 4923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 4924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 4925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 4926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 4927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 4928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1247,
+      "step": 4929
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 4930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 4931
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 4932
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 4933
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 4934
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 4935
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1162,
+      "step": 4936
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 4937
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 4938
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1424,
+      "step": 4939
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.135,
+      "step": 4940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 4941
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 4942
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 4943
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1635,
+      "step": 4944
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 4945
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 4946
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 4947
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 4948
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1294,
+      "step": 4949
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 4950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 4951
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.127,
+      "step": 4952
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 4953
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 4954
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 4955
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1355,
+      "step": 4956
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 4957
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 4958
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 4959
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 4960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 4961
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 4962
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 4963
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 4964
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 4965
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 4966
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 4967
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 4968
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 4969
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1451,
+      "step": 4970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 4971
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 4972
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 4973
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1527,
+      "step": 4974
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.151,
+      "step": 4975
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 4976
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 4977
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 4978
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 4979
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 4980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 4981
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1441,
+      "step": 4982
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 4983
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 4984
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 4985
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 4986
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 4987
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 4988
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 4989
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 4990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 4991
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 4992
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.149,
+      "step": 4993
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 4994
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 4995
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 4996
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 4997
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 4998
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 4999
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 5000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 5001
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 5002
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 5003
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 5004
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.126,
+      "step": 5005
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1214,
+      "step": 5006
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 5007
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 5008
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 5009
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 5010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 5011
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 5012
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 5013
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 5014
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 5015
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 5016
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 5017
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 5018
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1225,
+      "step": 5019
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 5020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 5021
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 5022
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 5023
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1021,
+      "step": 5024
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 5025
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 5026
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 5027
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1497,
+      "step": 5028
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 5029
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 5030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 5031
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 5032
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 5033
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 5034
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 5035
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 5036
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 5037
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 5038
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 5039
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 5040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 5041
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 5042
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 5043
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 5044
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1084,
+      "step": 5045
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 5046
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 5047
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 5048
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 5049
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 5050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 5051
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 5052
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 5053
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 5054
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 5055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 5056
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.124,
+      "step": 5057
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.135,
+      "step": 5058
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1545,
+      "step": 5059
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 5060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 5061
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 5062
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1135,
+      "step": 5063
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 5064
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 5065
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 5066
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 5067
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 5068
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 5069
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 5070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 5071
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 5072
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 5073
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 5074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 5075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 5076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 5077
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 5078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 5079
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 5080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 5081
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 5082
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 5083
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 5084
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 5085
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1159,
+      "step": 5086
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 5087
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 5088
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 5089
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 5090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 5091
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 5092
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1271,
+      "step": 5093
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1004,
+      "step": 5094
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1498,
+      "step": 5095
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 5096
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 5097
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 5098
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 5099
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 5100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1883,
+      "step": 5101
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 5102
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 5103
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 5104
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 5105
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 5106
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1527,
+      "step": 5107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 5108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 5109
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 5110
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 5111
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 5112
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 5113
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 5114
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 5115
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 5116
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 5117
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 5118
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1592,
+      "step": 5119
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 5120
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 5121
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 5122
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 5123
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 5124
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 5125
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 5126
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 5127
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1372,
+      "step": 5128
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 5129
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 5130
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 5131
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 5132
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 5133
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 5134
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 5135
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 5136
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 5137
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 5138
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 5139
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 5140
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 5141
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 5142
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 5143
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 5144
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 5145
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 5146
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 5147
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 5148
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 5149
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 5150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 5151
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 5152
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 5153
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 5154
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 5155
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 5156
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 5157
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 5158
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 5159
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 5160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 5161
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 5162
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 5163
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 5164
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 5165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1561,
+      "step": 5166
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 5167
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1536,
+      "step": 5168
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 5169
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 5170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 5171
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 5172
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 5173
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 5174
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 5175
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 5176
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 5177
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 5178
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 5179
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 5180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 5181
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 5182
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 5183
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 5184
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 5185
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 5186
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 5187
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 5188
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 5189
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 5190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 5191
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1563,
+      "step": 5192
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 5193
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 5194
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 5195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 5196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 5197
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 5198
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 5199
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 5200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 5201
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1284,
+      "step": 5202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 5203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1442,
+      "step": 5204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 5205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 5206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 5207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 5208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 5209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 5210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1468,
+      "step": 5211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 5212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 5213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 5214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 5215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 5216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 5217
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 5218
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 5219
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 5220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 5221
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 5222
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 5223
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 5224
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 5225
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 5226
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 5227
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 5228
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.145,
+      "step": 5229
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 5230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 5231
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 5232
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 5233
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 5234
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 5235
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 5236
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 5237
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 5238
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 5239
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 5240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 5241
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 5242
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 5243
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 5244
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 5245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0925,
+      "step": 5246
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 5247
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.194,
+      "step": 5248
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 5249
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 5250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.111,
+      "step": 5251
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 5252
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 5253
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 5254
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 5255
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 5256
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 5257
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 5258
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 5259
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 5260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 5261
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 5262
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 5263
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 5264
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 5265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 5266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 5267
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1159,
+      "step": 5268
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 5269
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 5270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 5271
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 5272
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 5273
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 5274
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 5275
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 5276
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 5277
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1365,
+      "step": 5278
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1352,
+      "step": 5279
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.11,
+      "step": 5280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 5281
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 5282
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 5283
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 5284
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 5285
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 5286
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 5287
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 5288
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 5289
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 5290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 5291
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 5292
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 5293
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 5294
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 5295
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 5296
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1314,
+      "step": 5297
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 5298
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 5299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 5300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1162,
+      "step": 5301
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 5302
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 5303
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 5304
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 5305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 5306
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1774,
+      "step": 5307
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 5308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 5309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 5310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 5311
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 5312
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 5313
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1407,
+      "step": 5314
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 5315
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 5316
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 5317
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 5318
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1385,
+      "step": 5319
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 5320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 5321
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 5322
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 5323
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 5324
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 5325
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 5326
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 5327
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 5328
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 5329
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 5330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 5331
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 5332
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1293,
+      "step": 5333
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 5334
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5335
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 5336
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 5337
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 5338
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 5339
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 5341
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5342
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 5343
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 5344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1072,
+      "step": 5345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.127,
+      "step": 5346
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 5347
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 5348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1584,
+      "step": 5349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 5350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 5351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1779,
+      "step": 5352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 5353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 5354
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 5355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 5356
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 5357
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1247,
+      "step": 5358
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 5359
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 5360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 5361
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 5362
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 5363
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 5364
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 5365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 5366
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0843,
+      "step": 5367
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 5368
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5369
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 5370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 5371
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 5372
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 5373
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 5374
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 5375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 5376
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 5377
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 5378
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 5379
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 5380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 5381
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 5382
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 5383
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1351,
+      "step": 5384
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 5385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 5386
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 5387
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1081,
+      "step": 5388
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 5389
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 5390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 5391
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 5392
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 5393
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0998,
+      "step": 5394
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 5395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 5396
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1016,
+      "step": 5397
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 5398
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 5399
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 5400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 5401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 5402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 5403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 5404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 5405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 5406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 5407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 5408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 5409
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 5410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 5411
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1184,
+      "step": 5412
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 5413
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 5414
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 5415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 5416
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 5417
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.128,
+      "step": 5418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 5419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 5420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 5421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 5422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 5423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 5424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 5425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 5427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 5429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1206,
+      "step": 5430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 5431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 5432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 5433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1187,
+      "step": 5434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 5435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 5436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 5437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 5438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 5439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1497,
+      "step": 5440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 5441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 5442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 5443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1285,
+      "step": 5444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 5445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 5446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 5447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 5449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 5450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 5451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 5452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 5455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 5456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 5457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 5458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 5459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 5460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1325,
+      "step": 5461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1433,
+      "step": 5462
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 5463
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 5464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 5465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 5466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 5467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 5468
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 5469
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 5470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 5471
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 5472
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 5473
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 5474
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 5475
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 5476
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 5477
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 5478
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1407,
+      "step": 5479
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 5480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 5481
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 5482
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 5483
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 5484
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 5485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 5486
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1709,
+      "step": 5487
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 5488
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 5489
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 5490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 5491
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 5492
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 5493
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 5494
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 5495
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 5496
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 5497
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 5498
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 5499
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 5500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5501
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 5502
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 5503
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 5504
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 5505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 5506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 5507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 5508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 5509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1186,
+      "step": 5510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 5511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 5512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 5513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 5514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 5515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 5516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 5517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 5518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 5519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 5520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 5521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 5522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 5523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 5524
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 5525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 5526
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 5527
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 5528
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 5529
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 5530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 5531
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1652,
+      "step": 5532
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1384,
+      "step": 5533
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2e-05,
+      "loss": 0.1472,
+      "step": 5534
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 5535
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 5536
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 5537
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 5538
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 5539
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 5540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 5541
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1539,
+      "step": 5542
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1135,
+      "step": 5543
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 5544
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 5545
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 5546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 5547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 5548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 5549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1202,
+      "step": 5550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 5551
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 5552
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 5553
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 5554
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 5555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 5556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 5557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 5558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 5559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 5560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 5561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 5562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 5563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 5564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 5565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 5566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 5567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 5568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 5569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 5570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 5571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 5572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 5573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 5574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 5575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 5576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 5577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 5578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 5579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 5580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.156,
+      "step": 5581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 5582
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 5583
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 5584
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 5586
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 5587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 5588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 5589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 5590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 5591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 5592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 5593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 5594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 5595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 5596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 5597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 5598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 5599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 5600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 5601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 5602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 5603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 5604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 5605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 5606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 5607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 5608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 5609
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1374,
+      "step": 5610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 5611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 5612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 5613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 5614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 5615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 5616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 5617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 5618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1356,
+      "step": 5619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 5620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 5621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 5622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 5623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 5624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 5625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1528,
+      "step": 5626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 5627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 5628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 5629
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 5630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 5631
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 5632
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 5633
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 5634
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 5635
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 5636
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 5637
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 5638
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 5639
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 5640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 5641
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 5642
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 5643
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 5644
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 5645
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 5646
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 5647
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 5648
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 5649
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.126,
+      "step": 5650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 5651
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 5652
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 5653
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 5654
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 5655
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 5656
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 5657
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 5658
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 5659
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 5660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 5661
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 5662
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 5663
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 5664
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 5665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 5666
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 5667
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1551,
+      "step": 5668
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 5669
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 5670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1361,
+      "step": 5671
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 5672
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 5673
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 5674
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 5675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 5676
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 5677
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 5678
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 5679
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 5680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 5681
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 5682
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 5683
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 5684
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 5685
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 5686
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 5687
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 5688
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 5689
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 5690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 5691
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 5692
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 5693
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 5694
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 5695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 5696
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 5697
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1784,
+      "step": 5698
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 5699
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 5700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 5701
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1159,
+      "step": 5702
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1558,
+      "step": 5703
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 5704
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 5705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 5706
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 5707
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 5708
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 5709
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 5710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 5711
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.14,
+      "step": 5712
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 5713
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 5714
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 5715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 5716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 5717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 5718
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 5719
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 5720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 5721
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 5722
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 5723
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 5724
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 5725
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0943,
+      "step": 5726
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 5727
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0924,
+      "step": 5728
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.148,
+      "step": 5729
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 5730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 5731
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 5732
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1619,
+      "step": 5733
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 5734
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 5735
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 5736
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 5737
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 5738
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 5739
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 5740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 5741
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 5742
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 5743
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 5744
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.165,
+      "step": 5745
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 5746
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 5747
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 5748
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 5749
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1101,
+      "step": 5750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1175,
+      "step": 5751
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 5752
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 5753
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0982,
+      "step": 5754
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 5755
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5756
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 5757
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 5758
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 5759
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 5760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 5761
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 5762
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 5763
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1314,
+      "step": 5764
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 5765
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 5766
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 5767
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 5768
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 5769
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 5770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 5771
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 5772
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 5773
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 5774
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 5775
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 5776
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 5777
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 5778
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 5779
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 5780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 5781
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 5782
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 5783
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 5784
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 5785
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 5786
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 5787
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 5788
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 5789
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 5790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 5791
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 5792
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 5793
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 5794
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1289,
+      "step": 5795
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 5796
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 5797
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 5798
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 5799
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 5800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 5801
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 5802
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 5803
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 5804
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 5805
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1433,
+      "step": 5806
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 5807
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 5808
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 5809
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 5810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 5811
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 5812
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 5813
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 5814
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 5815
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 5816
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1525,
+      "step": 5817
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 5818
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 5819
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 5820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 5821
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 5822
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1287,
+      "step": 5823
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 5824
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 5825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 5826
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 5827
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 5828
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 5829
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 5830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 5831
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 5832
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 5833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 5834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 5835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 5836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 5837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 5838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 5839
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 5840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 5841
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 5842
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 5843
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1693,
+      "step": 5844
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 5845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 5846
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 5847
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 5848
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 5849
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 5850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 5851
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 5852
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 5853
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 5854
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 5855
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 5856
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 5857
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 5858
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 5859
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 5860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 5861
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 5862
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 5863
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 5864
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 5865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 5866
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 5867
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 5868
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 5869
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 5870
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 5871
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 5872
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 5873
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 5874
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 5875
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 5876
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 5877
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 5878
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.135,
+      "step": 5879
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 5880
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 5881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 5882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 5883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 5884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 5885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 5886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1251,
+      "step": 5887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 5888
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 5889
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 5890
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 5891
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 5892
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 5893
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 5894
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1579,
+      "step": 5895
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 5896
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 5897
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 5898
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 5899
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 5900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 5901
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 5902
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 5903
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 5904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 5905
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 5906
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 5907
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 5908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 5909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 5910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 5911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 5912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 5913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 5914
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 5915
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 5916
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.122,
+      "step": 5917
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 5918
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 5919
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 5920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 5921
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 5922
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 5923
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1311,
+      "step": 5924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 5925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 5926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 5927
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 5928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 5929
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 5930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 5931
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 5932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 5933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 5934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 5935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 5936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 5937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 5938
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1393,
+      "step": 5939
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 5940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 5941
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 5942
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 5943
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 5944
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 5945
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 5946
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 5947
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 5948
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 5949
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 5950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 5951
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 5952
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 5953
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 5954
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 5955
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 5956
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 5957
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 5958
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 5959
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 5960
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 5961
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 5962
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 5963
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 5964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 5965
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 5966
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 5967
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 5968
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 5969
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 5970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 5971
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 5972
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 5973
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 5974
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 5975
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1148,
+      "step": 5976
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 5977
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 5978
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 5979
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 5980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 5981
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 5982
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 5983
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 5984
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 5985
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1285,
+      "step": 5986
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 5987
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 5988
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 5989
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1385,
+      "step": 5990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 5991
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 5992
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1321,
+      "step": 5993
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 5994
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1302,
+      "step": 5995
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 5996
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 5997
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1279,
+      "step": 5998
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.124,
+      "step": 5999
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 6000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 6001
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 6002
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 6003
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 6004
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0958,
+      "step": 6005
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 6006
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 6007
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 6008
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1698,
+      "step": 6009
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1283,
+      "step": 6010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 6011
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 6012
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 6013
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 6014
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 6015
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 6016
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 6017
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 6018
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 6019
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 6020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 6021
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1238,
+      "step": 6022
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 6023
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 6024
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1324,
+      "step": 6025
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 6026
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 6027
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 6028
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 6029
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 6030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 6031
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1138,
+      "step": 6032
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 6033
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 6034
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 6035
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 6036
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 6037
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 6038
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 6039
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 6040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 6041
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 6042
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1665,
+      "step": 6043
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1855,
+      "step": 6044
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 6045
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 6046
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 6047
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 6048
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 6049
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 6050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 6051
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 6052
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 6053
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 6054
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 6055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 6056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2e-05,
+      "loss": 0.1389,
+      "step": 6057
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 6058
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 6059
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 6060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 6061
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 6062
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 6063
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1571,
+      "step": 6064
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 6065
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 6066
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1379,
+      "step": 6067
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 6068
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 6069
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 6070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1534,
+      "step": 6071
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 6072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 6073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 6074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 6075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 6076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 6077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 6078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 6079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 6080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 6081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 6082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 6083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 6084
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 6085
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 6086
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 6087
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 6088
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 6089
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 6090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 6091
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 6092
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 6093
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 6094
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 6095
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1148,
+      "step": 6096
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1091,
+      "step": 6097
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 6098
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 6099
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 6100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 6101
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 6102
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 6103
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 6104
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 6105
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 6106
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 6107
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 6108
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1214,
+      "step": 6109
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 6110
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 6111
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1295,
+      "step": 6112
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.113,
+      "step": 6113
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 6114
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6115
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 6116
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 6117
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 6118
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 6119
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 6120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 6121
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 6122
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 6123
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 6124
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 6125
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 6126
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 6127
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 6128
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 6129
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 6130
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 6131
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 6132
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.15,
+      "step": 6133
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 6134
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 6135
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 6136
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 6137
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 6138
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 6139
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 6140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 6141
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 6142
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 6143
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 6144
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 6145
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 6146
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 6147
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 6148
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 6149
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 6150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 6151
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 6152
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 6153
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 6154
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 6155
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 6156
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 6157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 6158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1298,
+      "step": 6159
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 6160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 6161
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 6162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1226,
+      "step": 6163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 6164
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 6165
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 6166
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 6167
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1285,
+      "step": 6168
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 6169
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 6170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 6171
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 6172
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 6173
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 6174
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 6175
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 6176
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 6177
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 6178
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 6179
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 6180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 6181
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 6182
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 6183
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 6184
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 6185
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 6186
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 6187
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 6188
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 6189
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 6190
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 6191
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 6192
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 6193
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 6194
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 6195
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 6196
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 6197
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 6198
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 6199
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1211,
+      "step": 6200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 6201
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 6202
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1121,
+      "step": 6203
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 6204
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 6205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 6206
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 6207
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 6208
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 6209
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 6210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 6211
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 6212
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 6213
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 6214
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 6215
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 6216
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 6217
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 6218
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 6219
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 6220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 6221
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 6222
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 6223
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.154,
+      "step": 6224
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 6225
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 6226
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 6227
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 6228
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 6229
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 6230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 6231
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1368,
+      "step": 6232
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 6233
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 6234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 6235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 6236
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 6237
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 6238
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 6239
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 6240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1207,
+      "step": 6241
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 6242
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1061,
+      "step": 6243
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 6244
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 6245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 6246
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 6247
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 6248
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 6249
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 6250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 6251
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 6252
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 6253
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 6254
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 6255
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 6256
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 6257
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 6258
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 6259
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 6260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 6261
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 6262
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 6263
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 6264
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 6265
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 6266
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 6267
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 6268
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 6269
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 6270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 6271
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 6272
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 6273
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 6274
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 6275
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 6276
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 6277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 6278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 6279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 6280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1729,
+      "step": 6281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 6282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 6283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 6284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 6285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 6286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 6287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 6288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 6289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 6290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 6291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 6292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 6293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 6294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 6295
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 6296
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 6297
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 6298
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 6299
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 6300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 6301
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 6302
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 6303
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 6304
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 6306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1761,
+      "step": 6307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 6308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 6309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 6310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 6311
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1265,
+      "step": 6312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 6313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 6314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 6315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 6316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 6317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1615,
+      "step": 6318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 6319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0982,
+      "step": 6320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 6321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1084,
+      "step": 6322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 6323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 6324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 6325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1261,
+      "step": 6326
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 6327
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 6328
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 6329
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 6330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1002,
+      "step": 6331
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 6332
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 6333
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1455,
+      "step": 6334
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 6335
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 6336
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 6337
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 6338
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 6339
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 6340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 6341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1318,
+      "step": 6342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 6343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1352,
+      "step": 6344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 6345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 6346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 6347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 6348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 6349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 6350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 6351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 6352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 6353
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 6354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 6355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 6356
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 6357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.111,
+      "step": 6358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 6359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1021,
+      "step": 6360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 6361
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 6362
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 6363
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 6364
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 6365
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 6366
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 6367
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 6368
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 6369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 6370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 6371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1355,
+      "step": 6372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 6373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 6374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 6375
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1214,
+      "step": 6376
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6377
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 6378
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1436,
+      "step": 6379
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 6380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 6381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 6382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 6383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 6384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 6385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 6386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 6387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 6388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 6389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 6390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 6391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 6392
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 6393
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 6394
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 6395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 6396
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 6397
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 6398
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 6399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 6400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 6401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 6402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.111,
+      "step": 6403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 6404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 6405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1438,
+      "step": 6406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 6407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 6408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 6409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1436,
+      "step": 6410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 6411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 6412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 6413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 6414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1302,
+      "step": 6415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 6416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 6417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 6418
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1579,
+      "step": 6419
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 6420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 6421
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 6422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 6423
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 6424
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 6425
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 6426
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 6427
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 6428
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 6429
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 6430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 6431
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6432
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 6433
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 6434
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 6435
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 6436
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 6437
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 6438
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 6439
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1374,
+      "step": 6440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 6441
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1257,
+      "step": 6442
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 6443
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 6444
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1321,
+      "step": 6445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 6446
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 6447
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 6448
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 6449
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 6450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 6451
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1742,
+      "step": 6452
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 6453
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 6454
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 6455
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 6456
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 6457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 6458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 6459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1415,
+      "step": 6460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 6461
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 6462
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 6463
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 6464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.17,
+      "step": 6465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 6466
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 6467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 6468
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 6469
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 6470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1502,
+      "step": 6471
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 6472
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 6473
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 6474
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 6475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 6476
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 6477
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 6478
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 6479
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 6480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 6481
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 6482
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 6483
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 6484
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 6485
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 6486
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 6487
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 6488
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.137,
+      "step": 6489
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 6490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 6491
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 6492
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 6493
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 6494
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 6495
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 6496
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 6497
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 6498
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 6499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 6500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 6501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 6502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 6503
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 6504
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 6505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 6506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 6507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 6508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 6509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1344,
+      "step": 6510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1325,
+      "step": 6511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 6512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 6513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 6514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 6515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 6516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 6517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 6518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 6519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 6520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 6521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 6522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 6523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 6524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 6525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 6526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 6527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 6528
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 6529
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 6530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 6531
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 6532
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 6533
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 6534
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 6535
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 6536
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 6537
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 6538
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 6539
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 6540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 6541
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 6542
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 6543
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 6544
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 6545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 6546
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1348,
+      "step": 6547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 6548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 6549
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 6550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 6551
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 6552
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 6553
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1328,
+      "step": 6554
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 6555
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 6556
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 6557
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 6558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 6559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 6560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 6561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 6562
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.165,
+      "step": 6563
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 6564
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 6565
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 6566
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 6567
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 6568
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1354,
+      "step": 6569
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 6570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1135,
+      "step": 6571
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 6572
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 6573
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 6574
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 6575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.1301,
+      "step": 6576
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 6577
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 6578
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 6579
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 6580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 6581
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1575,
+      "step": 6582
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 6583
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 6584
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 6585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.151,
+      "step": 6586
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 6587
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 6588
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 6589
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 6590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1187,
+      "step": 6591
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 6592
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1192,
+      "step": 6593
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 6594
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 6595
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 6596
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 6597
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 6598
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 6599
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 6600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 6601
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 6602
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 6603
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 6604
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1271,
+      "step": 6605
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1465,
+      "step": 6606
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6607
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 6608
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 6609
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 6610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 6611
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 6612
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 6613
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1458,
+      "step": 6614
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 6615
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 6616
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1211,
+      "step": 6617
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1819,
+      "step": 6618
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 6619
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 6620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 6621
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 6622
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1589,
+      "step": 6623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 6624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 6625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 6626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2e-05,
+      "loss": 0.1279,
+      "step": 6627
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 6628
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1311,
+      "step": 6629
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 6630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 6631
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1159,
+      "step": 6632
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 6633
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 6634
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 6635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 6636
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1384,
+      "step": 6637
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 6638
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 6639
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 6640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1189,
+      "step": 6641
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1254,
+      "step": 6642
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 6643
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 6644
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 6645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 6646
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1523,
+      "step": 6647
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 6648
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 6649
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 6650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 6651
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 6652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 6653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 6654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 6655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 6656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 6657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 6658
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 6659
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 6660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 6661
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1061,
+      "step": 6662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 6663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 6664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 6665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 6666
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 6667
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 6668
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6669
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 6670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 6671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 6672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 6673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 6674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 6675
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 6676
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 6677
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1248,
+      "step": 6678
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 6679
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 6680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 6681
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 6682
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 6683
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 6684
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.188,
+      "step": 6685
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 6686
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 6687
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 6688
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 6689
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 6690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 6691
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 6692
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1084,
+      "step": 6693
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 6694
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 6695
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 6696
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 6697
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 6698
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6699
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1427,
+      "step": 6701
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 6702
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 6703
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 6704
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 6705
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 6706
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 6707
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 6708
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 6709
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 6710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 6711
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 6712
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 6713
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 6714
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 6715
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1084,
+      "step": 6716
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 6717
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 6718
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 6719
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 6720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 6721
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 6722
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 6723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1525,
+      "step": 6724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 6725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 6726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 6727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0985,
+      "step": 6728
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1265,
+      "step": 6729
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 6730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 6731
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 6732
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1207,
+      "step": 6733
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 6734
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 6735
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 6736
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 6737
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 6738
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 6739
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 6740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6741
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 6742
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 6743
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 6744
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 6745
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 6746
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 6747
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 6748
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 6749
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 6750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 6751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 6752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 6753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 6754
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 6756
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 6757
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 6758
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 6759
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 6760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 6761
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 6762
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 6763
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 6764
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 6765
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 6766
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 6767
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 6768
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 6769
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 6770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 6771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 6772
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 6773
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 6774
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1596,
+      "step": 6775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 6776
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 6777
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1868,
+      "step": 6778
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1058,
+      "step": 6779
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 6780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 6781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1095,
+      "step": 6782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 6783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1191,
+      "step": 6784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 6785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 6786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 6787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 6788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 6789
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1426,
+      "step": 6790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1572,
+      "step": 6791
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 6792
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 6793
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 6794
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 6795
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1133,
+      "step": 6796
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 6797
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 6798
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 6799
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 6800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 6801
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 6802
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 6803
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 6804
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 6805
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 6806
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 6807
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 6808
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 6809
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 6810
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 6811
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 6812
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 6813
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 6814
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 6815
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 6816
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1354,
+      "step": 6817
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 6818
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 6819
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 6820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1316,
+      "step": 6821
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 6822
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 6823
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 6824
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 6825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 6826
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 6827
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 6828
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 6829
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 6830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 6831
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 6832
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 6833
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 6834
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 6835
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 6836
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 6837
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 6838
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1101,
+      "step": 6839
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 6840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 6841
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 6842
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 6843
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 6844
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 6845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 6846
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 6847
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 6848
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 6849
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 6850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 6851
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 6852
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 6853
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 6854
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 6855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 6856
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 6857
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 6858
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 6859
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 6860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 6861
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 6862
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 6863
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 6864
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1229,
+      "step": 6865
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 6866
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 6867
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 6868
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1314,
+      "step": 6869
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 6870
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 6871
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 6872
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 6873
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 6874
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 6875
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 6876
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 6877
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1072,
+      "step": 6878
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1302,
+      "step": 6879
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 6880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1222,
+      "step": 6881
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 6882
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 6883
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 6884
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 6885
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1341,
+      "step": 6886
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 6887
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 6888
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 6889
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 6890
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 6891
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 6892
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 6893
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 6894
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 6895
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 6896
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 6897
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 6898
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 6899
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 6900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 6901
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 6902
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 6903
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 6904
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 6905
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 6906
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 6907
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 6908
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 6909
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1519,
+      "step": 6910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 6911
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2e-05,
+      "loss": 0.1743,
+      "step": 6912
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 6913
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 6914
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.147,
+      "step": 6915
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 6916
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 6917
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 6918
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 6919
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 6920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1461,
+      "step": 6921
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 6922
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 6923
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 6924
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 6925
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 6926
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 6927
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1494,
+      "step": 6928
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1677,
+      "step": 6929
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 6930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 6931
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 6932
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 6933
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 6934
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 6935
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 6936
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 6937
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 6938
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 6939
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 6940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 6941
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 6942
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 6943
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 6944
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1095,
+      "step": 6945
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 6946
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 6947
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 6948
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 6949
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 6950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 6951
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 6952
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 6953
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 6954
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 6955
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 6956
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 6957
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 6958
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 6959
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 6960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1096,
+      "step": 6961
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 6962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 6963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 6964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 6965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.139,
+      "step": 6966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 6967
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 6968
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 6969
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 6970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 6971
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 6972
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 6973
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 6974
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 6975
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 6976
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 6977
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 6978
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 6979
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 6980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 6981
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 6982
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 6983
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 6984
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1144,
+      "step": 6985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1121,
+      "step": 6986
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 6987
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 6988
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 6989
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1266,
+      "step": 6990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 6991
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 6992
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1697,
+      "step": 6993
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 6994
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 6995
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 6996
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 6997
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 6998
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 6999
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1361,
+      "step": 7000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 7001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 7002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1606,
+      "step": 7003
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1421,
+      "step": 7004
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 7005
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 7006
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 7007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 7008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1477,
+      "step": 7009
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 7010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1351,
+      "step": 7011
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1385,
+      "step": 7012
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 7013
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1217,
+      "step": 7014
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 7015
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0934,
+      "step": 7016
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 7017
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 7018
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 7019
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 7020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 7021
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 7022
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 7023
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 7024
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 7025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 7026
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 7027
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 7028
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 7029
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 7030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 7031
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 7032
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 7033
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 7034
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 7035
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 7036
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1305,
+      "step": 7037
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 7038
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 7040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 7041
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 7042
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 7043
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 7044
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1254,
+      "step": 7045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 7046
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 7047
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 7048
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 7049
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 7050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 7051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 7052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 7053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 7054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 7055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 7056
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 7057
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 7058
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1221,
+      "step": 7059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1432,
+      "step": 7060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 7061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 7062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 7063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 7064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 7065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 7066
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 7067
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 7068
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 7069
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 7070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 7071
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 7072
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 7073
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7074
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 7075
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 7076
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 7077
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 7078
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 7079
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 7080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1595,
+      "step": 7081
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 7082
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 7083
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 7084
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 7085
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 7086
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 7087
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1295,
+      "step": 7088
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 7089
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 7090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 7091
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 7092
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 7093
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 7094
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 7095
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 7096
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 7097
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 7098
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 7099
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 7100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 7101
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 7102
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 7103
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 7104
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 7105
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 7106
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7107
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1111,
+      "step": 7108
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 7109
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 7110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 7111
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 7112
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 7113
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 7114
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0924,
+      "step": 7115
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 7116
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 7117
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 7118
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 7119
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 7120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 7121
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7122
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 7123
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 7124
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 7125
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 7126
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 7127
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 7128
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 7129
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1734,
+      "step": 7130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 7131
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1072,
+      "step": 7132
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 7133
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 7134
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1217,
+      "step": 7135
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 7136
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 7137
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 7138
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7139
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 7140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 7141
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 7142
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 7143
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 7144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 7145
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 7146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1586,
+      "step": 7147
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 7148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 7149
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 7150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 7151
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 7152
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 7153
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 7154
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 7155
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 7156
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 7157
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 7158
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 7159
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 7160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 7161
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 7162
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 7163
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 7164
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 7165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 7166
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 7167
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 7168
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1133,
+      "step": 7169
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 7170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 7171
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1512,
+      "step": 7172
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 7173
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 7174
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 7176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 7177
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 7178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 7179
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 7180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 7181
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0934,
+      "step": 7182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 7183
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1198,
+      "step": 7184
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 7185
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1453,
+      "step": 7186
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1135,
+      "step": 7187
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 7188
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 7189
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 7190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 7191
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 7192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 7193
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 7194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 7195
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 7196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 7197
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 7198
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 7199
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 7200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 7201
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 7202
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 7203
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1369,
+      "step": 7204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 7205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 7206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 7207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 7208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 7209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 7210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 7211
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 7212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 7213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 7214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 7215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 7216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1379,
+      "step": 7217
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 7218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 7219
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 7220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 7221
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 7222
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 7223
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 7224
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 7225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 7226
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1495,
+      "step": 7227
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 7228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 7229
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 7230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 7231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 7232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 7233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 7234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 7235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 7236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 7237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 7238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 7239
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 7240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 7241
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1602,
+      "step": 7242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 7243
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 7244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 7246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0942,
+      "step": 7247
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 7248
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 7249
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 7250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 7251
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 7252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 7253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 7254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1559,
+      "step": 7255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 7256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 7257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 7258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 7259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 7260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 7261
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 7262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 7263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 7264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.117,
+      "step": 7265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 7266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 7267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 7268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 7269
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 7270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 7271
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 7272
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 7273
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 7274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 7275
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 7276
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 7277
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 7278
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 7279
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 7280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 7281
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 7282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 7283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 7284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 7285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 7286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 7287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 7288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1221,
+      "step": 7289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 7290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 7291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 7292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 7293
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 7294
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 7295
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 7296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 7297
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1016,
+      "step": 7298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 7299
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 7300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 7301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 7302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 7303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 7304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 7305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 7306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 7307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 7308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1381,
+      "step": 7309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1258,
+      "step": 7310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 7311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 7312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 7313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 7314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 7315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 7316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 7317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 7318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 7319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 7320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 7321
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 7322
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 7323
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 7324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 7325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.146,
+      "step": 7326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 7327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 7328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 7329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 7330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 7331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1366,
+      "step": 7332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1148,
+      "step": 7333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1367,
+      "step": 7334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 7335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 7336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1304,
+      "step": 7337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1893,
+      "step": 7338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 7339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 7340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 7341
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 7342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 7343
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 7344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 7345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 7346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1002,
+      "step": 7347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 7348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 7349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 7350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1869,
+      "step": 7351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 7352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 7353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 7354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 7355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 7356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 7357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 7358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 7359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 7360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 7361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 7362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 7363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 7364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 7365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 7366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 7367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 7368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 7369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 7370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 7371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0985,
+      "step": 7372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0998,
+      "step": 7373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 7374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1394,
+      "step": 7375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 7376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1202,
+      "step": 7377
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 7378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 7379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 7380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 7381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 7382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 7383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1461,
+      "step": 7384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 7385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1002,
+      "step": 7386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 7387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 7388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1171,
+      "step": 7389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 7390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 7391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 7392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 7393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.129,
+      "step": 7394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 7395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 7396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 7397
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 7398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 7399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 7400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 7401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 7402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 7403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 7404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 7405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0998,
+      "step": 7406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 7407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 7408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 7409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 7410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 7411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 7412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 7413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 7414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1796,
+      "step": 7415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1311,
+      "step": 7416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 7417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.2296,
+      "step": 7418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 7419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 7420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 7421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 7423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1297,
+      "step": 7424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 7425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 7426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 7427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 7428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 7429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 7430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.127,
+      "step": 7431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 7432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 7433
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 7434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 7435
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 7436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 7437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 7438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 7439
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 7440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 7441
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 7442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 7443
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 7444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 7445
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 7446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 7447
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1238,
+      "step": 7448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 7449
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1108,
+      "step": 7450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 7451
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 7452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 7453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 7454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 7455
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 7456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 7457
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 7458
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1214,
+      "step": 7459
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1426,
+      "step": 7460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 7461
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 7462
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 7463
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.151,
+      "step": 7464
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 7465
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1421,
+      "step": 7466
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 7467
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 7468
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 7469
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1435,
+      "step": 7470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 7471
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 7472
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 7473
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 7474
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 7475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 7476
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 7477
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 7478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 7479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 7480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 7481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1388,
+      "step": 7482
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 7483
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 7484
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.168,
+      "step": 7485
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 7486
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 7487
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 7488
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 7489
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1114,
+      "step": 7490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 7491
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 7492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1682,
+      "step": 7493
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 7494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 7495
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 7496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 7497
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 7498
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7499
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 7500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 7501
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 7502
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1277,
+      "step": 7503
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 7504
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 7505
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 7506
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 7507
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 7508
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 7509
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 7510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 7511
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 7512
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 7513
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 7514
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 7515
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 7516
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 7517
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 7518
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 7519
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1211,
+      "step": 7520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 7521
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7522
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 7523
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1289,
+      "step": 7524
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1038,
+      "step": 7525
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 7526
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 7527
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 7528
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 7529
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 7530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1424,
+      "step": 7531
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 7532
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 7533
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 7534
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1315,
+      "step": 7535
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 7536
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 7537
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 7538
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 7539
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 7540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 7541
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 7542
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1059,
+      "step": 7543
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 7544
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7545
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 7546
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 7547
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1224,
+      "step": 7548
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 7549
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 7550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 7551
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 7552
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 7553
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1483,
+      "step": 7554
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 7555
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 7556
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 7557
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 7558
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 7559
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 7560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 7561
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 7562
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7563
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1254,
+      "step": 7564
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 7565
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 7566
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 7567
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 7568
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 7569
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 7570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 7571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 7572
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 7573
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 7574
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 7575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 7576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 7577
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 7578
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 7579
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 7580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 7581
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 7582
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1428,
+      "step": 7583
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 7584
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 7585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 7586
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 7587
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 7588
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1342,
+      "step": 7589
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 7590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.145,
+      "step": 7591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 7592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 7593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 7594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.138,
+      "step": 7595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1182,
+      "step": 7597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1199,
+      "step": 7599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 7600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.169,
+      "step": 7601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 7602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 7603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 7604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 7605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 7606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 7607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 7608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 7609
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1279,
+      "step": 7610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 7611
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 7612
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 7613
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 7614
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 7615
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 7616
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 7617
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 7618
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 7619
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1202,
+      "step": 7620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.127,
+      "step": 7621
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 7622
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 7623
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 7624
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2e-05,
+      "loss": 0.1653,
+      "step": 7625
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1103,
+      "step": 7626
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 7627
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 7628
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 7629
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 7630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 7631
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 7632
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 7633
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 7634
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 7635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1599,
+      "step": 7636
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 7637
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 7638
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 7639
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 7640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 7641
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 7642
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 7643
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 7644
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 7645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 7646
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 7647
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 7648
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 7649
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 7650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 7651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 7653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 7654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 7655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1899,
+      "step": 7656
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1218,
+      "step": 7657
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 7658
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 7659
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 7660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.172,
+      "step": 7661
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 7662
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 7663
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 7664
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 7665
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 7666
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 7667
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 7668
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 7669
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 7670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 7671
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1345,
+      "step": 7672
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 7673
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 7674
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 7675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 7676
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 7677
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 7678
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 7679
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 7680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 7681
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 7682
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 7683
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 7684
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 7685
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 7686
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 7687
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 7688
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 7689
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 7690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 7691
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 7692
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 7693
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 7694
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 7695
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 7696
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 7697
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 7698
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 7699
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 7700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 7701
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 7702
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 7703
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 7704
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 7705
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 7706
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 7707
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 7708
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 7709
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 7710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1185,
+      "step": 7711
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 7712
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 7713
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 7714
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 7715
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 7716
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 7717
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 7718
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 7719
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 7720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1147,
+      "step": 7721
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1289,
+      "step": 7722
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 7723
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 7724
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 7725
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 7726
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 7727
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 7728
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 7729
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 7730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 7731
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 7732
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1275,
+      "step": 7733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 7734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 7735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.172,
+      "step": 7736
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 7737
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 7738
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 7739
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1262,
+      "step": 7740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 7741
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 7742
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 7743
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 7744
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 7745
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 7746
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 7747
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 7748
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 7749
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1305,
+      "step": 7750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 7751
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 7752
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 7753
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1434,
+      "step": 7754
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 7755
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 7756
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 7757
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 7758
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 7759
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1341,
+      "step": 7760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 7761
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 7762
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1813,
+      "step": 7763
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1661,
+      "step": 7764
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 7765
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 7766
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 7767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 7768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 7769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1416,
+      "step": 7770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 7771
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 7772
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 7773
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1184,
+      "step": 7774
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1332,
+      "step": 7775
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 7776
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 7777
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1517,
+      "step": 7778
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 7779
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1586,
+      "step": 7780
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 7781
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 7782
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 7783
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 7784
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 7785
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 7786
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 7787
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 7788
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 7789
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 7790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 7791
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 7792
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 7793
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 7794
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 7795
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 7796
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 7797
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1548,
+      "step": 7798
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1541,
+      "step": 7799
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 7800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 7801
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 7802
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 7803
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 7804
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 7805
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 7806
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 7807
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 7808
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 7809
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 7810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 7811
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 7812
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1524,
+      "step": 7813
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 7814
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 7815
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 7816
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1581,
+      "step": 7817
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 7818
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 7819
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 7820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 7821
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 7822
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 7823
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 7824
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 7825
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 7826
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 7827
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 7828
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 7829
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 7830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1098,
+      "step": 7831
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.11,
+      "step": 7832
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 7833
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1218,
+      "step": 7834
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 7835
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 7836
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 7837
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 7838
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 7839
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 7840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 7841
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 7842
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.131,
+      "step": 7843
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 7844
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 7845
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1092,
+      "step": 7846
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 7847
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 7848
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 7849
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 7850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 7851
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 7852
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 7853
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 7854
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 7855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 7856
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 7857
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 7858
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1366,
+      "step": 7859
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 7860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1435,
+      "step": 7861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 7862
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 7863
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 7864
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 7865
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 7866
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 7867
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 7868
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.131,
+      "step": 7869
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 7870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 7871
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 7872
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 7873
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 7874
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 7875
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 7876
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 7877
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 7878
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 7879
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 7880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 7881
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 7882
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 7883
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.113,
+      "step": 7884
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 7885
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1337,
+      "step": 7886
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 7887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 7888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1133,
+      "step": 7889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 7890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 7891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 7892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 7893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 7894
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1378,
+      "step": 7895
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 7896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 7897
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 7898
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1734,
+      "step": 7899
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 7900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 7901
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 7902
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 7903
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 7904
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 7905
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 7906
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 7907
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 7908
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 7909
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 7910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 7911
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 7912
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 7913
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 7914
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 7915
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 7916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 7917
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 7918
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 7919
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 7920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 7921
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 7922
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 7923
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 7924
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 7925
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 7926
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 7927
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 7928
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 7929
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 7930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 7931
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1582,
+      "step": 7932
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 7933
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 7934
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 7935
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 7936
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 7937
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 7938
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 7939
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 7940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 7941
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 7942
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 7943
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 7944
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1311,
+      "step": 7945
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 7946
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 7947
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 7948
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 7949
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 7950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1357,
+      "step": 7951
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 7952
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 7953
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 7954
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 7955
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 7956
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 7957
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 7958
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 7959
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 7960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 7961
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 7962
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1222,
+      "step": 7963
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 7964
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 7965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 7966
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 7967
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 7968
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 7969
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1272,
+      "step": 7970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 7971
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1375,
+      "step": 7972
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 7973
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 7974
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 7975
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 7976
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 7977
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1424,
+      "step": 7978
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 7979
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.109,
+      "step": 7980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 7981
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 7982
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 7983
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 7984
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 7985
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 7986
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 7987
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 7988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 7989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 7990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 7991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 7992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 7993
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 7994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 7995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 7996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 7997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 7998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 7999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 8000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1254,
+      "step": 8001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1038,
+      "step": 8002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 8003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 8004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 8005
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 8006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1242,
+      "step": 8007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1264,
+      "step": 8008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 8009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 8010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0925,
+      "step": 8011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1187,
+      "step": 8012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 8013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1304,
+      "step": 8014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 8015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 8016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1335,
+      "step": 8017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 8018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1699,
+      "step": 8019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 8020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 8021
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1437,
+      "step": 8022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 8023
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1002,
+      "step": 8024
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.147,
+      "step": 8025
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 8026
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1354,
+      "step": 8027
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 8028
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 8029
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 8030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 8031
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 8032
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 8033
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 8034
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 8035
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 8036
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1038,
+      "step": 8037
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 8038
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 8039
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 8040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 8041
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.128,
+      "step": 8042
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1227,
+      "step": 8043
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 8044
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 8045
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 8046
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 8047
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 8048
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 8049
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 8050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0958,
+      "step": 8051
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 8052
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 8053
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 8054
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 8055
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1182,
+      "step": 8056
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1544,
+      "step": 8057
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 8058
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 8059
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 8060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 8061
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 8062
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 8063
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 8064
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 8065
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 8066
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 8067
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 8068
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1105,
+      "step": 8069
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 8070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 8071
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 8072
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 8073
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1167,
+      "step": 8074
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 8075
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1431,
+      "step": 8076
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 8077
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 8078
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 8079
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 8080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 8081
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 8082
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1665,
+      "step": 8083
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1583,
+      "step": 8084
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1061,
+      "step": 8085
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 8086
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 8087
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 8088
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 8089
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0977,
+      "step": 8090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1755,
+      "step": 8091
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 8092
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 8093
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 8094
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 8095
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1307,
+      "step": 8096
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 8097
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1533,
+      "step": 8098
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1081,
+      "step": 8099
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 8100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 8101
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 8102
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 8103
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1363,
+      "step": 8104
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 8105
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 8106
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 8107
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 8108
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 8109
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 8110
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 8111
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 8112
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 8113
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 8114
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 8115
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 8116
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 8117
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 8118
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 8119
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 8120
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 8121
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 8122
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 8123
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 8124
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1326,
+      "step": 8125
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 8126
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 8127
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1385,
+      "step": 8128
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 8129
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 8130
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 8131
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 8132
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 8133
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 8134
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 8135
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 8136
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 8137
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 8138
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 8139
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 8140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 8141
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 8142
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 8143
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 8144
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 8145
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 8146
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1261,
+      "step": 8147
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 8148
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1399,
+      "step": 8149
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 8150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 8151
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0996,
+      "step": 8152
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 8153
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 8154
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 8155
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 8156
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 8157
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 8158
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 8159
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.153,
+      "step": 8160
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 8161
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 8162
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 8163
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 8164
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 8165
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 8166
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 8167
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 8168
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1363,
+      "step": 8169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1308,
+      "step": 8170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 8171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1449,
+      "step": 8172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 8173
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 8174
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 8175
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 8176
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 8177
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 8178
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 8179
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0955,
+      "step": 8180
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 8181
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 8182
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 8183
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 8184
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 8185
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 8186
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 8187
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 8188
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 8189
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 8190
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1226,
+      "step": 8191
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 8192
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 8193
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 8194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1199,
+      "step": 8195
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 8196
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 8197
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 8198
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 8199
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 8200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 8201
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 8202
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1496,
+      "step": 8203
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 8204
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 8205
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 8206
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 8207
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 8208
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 8209
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1321,
+      "step": 8210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1222,
+      "step": 8211
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 8212
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 8213
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 8214
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1021,
+      "step": 8215
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 8216
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 8217
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 8218
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 8219
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1338,
+      "step": 8220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 8221
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1245,
+      "step": 8222
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8223
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 8224
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 8225
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 8226
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 8227
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1348,
+      "step": 8228
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 8229
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 8230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 8231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 8232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 8233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 8234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 8235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 8236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 8237
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 8238
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 8239
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 8240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 8241
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 8242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 8243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 8244
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 8245
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 8246
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 8247
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 8248
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 8249
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 8250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 8251
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 8252
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 8253
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 8254
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 8255
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1298,
+      "step": 8256
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 8257
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1501,
+      "step": 8258
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 8259
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 8260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1133,
+      "step": 8261
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 8262
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 8263
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 8264
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 8265
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 8266
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1141,
+      "step": 8267
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 8268
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 8269
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 8270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.157,
+      "step": 8271
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 8272
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 8273
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 8274
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 8275
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 8276
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 8277
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 8278
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 8279
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 8280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 8281
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 8282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 8283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 8284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 8285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 8286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1558,
+      "step": 8287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1497,
+      "step": 8288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 8289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 8290
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1177,
+      "step": 8291
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 8292
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 8293
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 8294
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 8295
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 8296
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 8297
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 8298
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 8299
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1374,
+      "step": 8300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 8301
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 8302
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 8303
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 8304
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 8305
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 8306
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 8307
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1195,
+      "step": 8308
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 8309
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 8310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 8311
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 8312
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 8313
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0971,
+      "step": 8314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 8315
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 8316
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 8317
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 8318
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1464,
+      "step": 8319
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1328,
+      "step": 8320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 8321
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 8322
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 8323
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 8324
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.14,
+      "step": 8325
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 8326
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 8327
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 8328
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1457,
+      "step": 8329
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1401,
+      "step": 8330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 8331
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 8332
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 8333
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1125,
+      "step": 8334
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1365,
+      "step": 8335
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 8336
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 8337
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 8338
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 8339
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 8340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1513,
+      "step": 8341
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1382,
+      "step": 8342
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 8343
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1669,
+      "step": 8344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 8345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 8346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1505,
+      "step": 8347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 8348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 8349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 8350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 8351
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 8352
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 8353
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 8354
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 8355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 8356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 8357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 8358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1623,
+      "step": 8359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1795,
+      "step": 8360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 8361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 8362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 8363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 8364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 8365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 8366
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1469,
+      "step": 8367
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 8368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 8369
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 8370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 8372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1454,
+      "step": 8373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 8374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 8375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 8376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 8377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 8378
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 8379
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 8380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0985,
+      "step": 8381
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1174,
+      "step": 8382
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 8383
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 8384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1199,
+      "step": 8385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 8386
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 8387
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 8388
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 8389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 8390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 8391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 8392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 8393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 8394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 8395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 8396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 8397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 8398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 8399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 8400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 8401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 8402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1219,
+      "step": 8403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1069,
+      "step": 8404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 8405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 8406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 8407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 8408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 8409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 8410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 8411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1309,
+      "step": 8412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 8413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 8414
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1376,
+      "step": 8415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 8416
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 8417
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 8418
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 8419
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 8420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1812,
+      "step": 8421
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 8422
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 8423
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 8424
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 8425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1608,
+      "step": 8426
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 8427
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1199,
+      "step": 8428
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 8429
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 8430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 8431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 8432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 8433
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 8434
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 8435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 8436
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 8437
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 8438
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 8439
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 8440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 8441
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1343,
+      "step": 8442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 8443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 8444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 8445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 8446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 8447
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1235,
+      "step": 8448
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 8449
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 8450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 8451
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1396,
+      "step": 8452
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 8453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 8454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 8455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 8456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 8457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1856,
+      "step": 8458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1642,
+      "step": 8459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1225,
+      "step": 8460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 8461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 8462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 8463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 8464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 8465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 8466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 8467
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 8468
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 8469
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1182,
+      "step": 8470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8471
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 8472
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 8473
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 8474
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1414,
+      "step": 8475
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 8476
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 8477
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 8478
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 8479
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1346,
+      "step": 8480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 8481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 8482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 8483
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 8484
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 8485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 8486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 8487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 8488
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 8489
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 8490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 8491
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 8492
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1244,
+      "step": 8493
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 8494
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 8495
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1577,
+      "step": 8496
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1164,
+      "step": 8497
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1157,
+      "step": 8498
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1463,
+      "step": 8499
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 8500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1061,
+      "step": 8501
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 8502
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 8503
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 8504
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1746,
+      "step": 8505
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 8506
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1172,
+      "step": 8507
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1546,
+      "step": 8508
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 8509
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 8510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 8511
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1307,
+      "step": 8512
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.124,
+      "step": 8513
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 8514
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 8515
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 8516
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 8517
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 8518
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1077,
+      "step": 8519
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1104,
+      "step": 8520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 8521
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 8522
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 8523
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 8524
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 8525
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 8526
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1476,
+      "step": 8527
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 8528
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 8529
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 8530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 8531
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 8532
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 8533
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 8534
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 8535
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 8536
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 8537
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1155,
+      "step": 8538
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 8539
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 8540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 8541
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 8542
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 8543
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 8544
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 8545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 8546
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 8547
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 8548
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 8549
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 8550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 8551
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1213,
+      "step": 8552
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1333,
+      "step": 8553
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 8554
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 8555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 8556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 8557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 8558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 8559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 8560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 8561
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 8562
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 8563
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 8564
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 8565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 8566
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 8567
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1325,
+      "step": 8568
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 8569
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1179,
+      "step": 8570
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 8571
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 8572
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 8573
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 8574
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1251,
+      "step": 8575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 8576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 8577
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 8578
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 8579
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 8580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1327,
+      "step": 8581
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 8582
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 8583
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 8584
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 8585
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1225,
+      "step": 8586
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 8587
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 8588
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1377,
+      "step": 8589
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 8590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1252,
+      "step": 8591
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 8592
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 8593
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 8594
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 8595
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1134,
+      "step": 8596
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1231,
+      "step": 8597
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 8598
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 8599
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 8600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 8601
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 8602
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 8603
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 8604
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 8605
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 8606
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 8607
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 8608
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 8609
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 8610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 8611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 8612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1114,
+      "step": 8613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 8614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 8615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 8616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 8617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1095,
+      "step": 8618
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 8619
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 8620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 8621
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1387,
+      "step": 8622
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1588,
+      "step": 8623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 8624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 8625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 8626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 8627
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 8628
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 8629
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 8630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 8631
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0955,
+      "step": 8632
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 8633
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 8634
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 8635
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 8636
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 8637
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 8638
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 8639
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 8640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 8641
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8642
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 8643
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 8644
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 8645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 8646
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 8647
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 8648
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8649
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 8650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 8651
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 8652
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 8653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 8654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1235,
+      "step": 8655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 8656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 8657
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 8658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 8659
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1268,
+      "step": 8660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 8661
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 8662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 8663
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 8664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 8665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 8666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 8667
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 8668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 8669
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 8670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 8671
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 8672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 8673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 8674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 8675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 8676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 8677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 8678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 8679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 8680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 8681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 8682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 8683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 8684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1358,
+      "step": 8685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1112,
+      "step": 8686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 8687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 8688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 8689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 8690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 8691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 8692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 8693
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 8694
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 8695
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 8696
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 8697
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 8698
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 8699
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 8700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.113,
+      "step": 8701
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1191,
+      "step": 8702
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 8703
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 8704
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 8705
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 8706
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1403,
+      "step": 8707
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 8708
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 8709
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1023,
+      "step": 8710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0966,
+      "step": 8711
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 8712
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 8713
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 8714
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 8715
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 8716
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 8717
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 8718
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1366,
+      "step": 8719
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 8720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 8721
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 8722
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 8723
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1199,
+      "step": 8724
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 8725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.125,
+      "step": 8726
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 8727
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 8728
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 8729
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 8730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 8731
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 8732
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 8733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 8734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 8735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1549,
+      "step": 8736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 8737
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0958,
+      "step": 8738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1336,
+      "step": 8739
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 8740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 8741
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 8742
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 8743
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 8744
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1168,
+      "step": 8745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 8746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 8747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 8748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 8749
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 8750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 8751
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 8752
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1258,
+      "step": 8753
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 8754
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 8755
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 8756
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 8757
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 8758
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0973,
+      "step": 8759
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 8760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 8761
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 8762
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1417,
+      "step": 8763
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 8764
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 8765
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 8766
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 8767
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 8768
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 8769
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 8770
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 8771
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 8772
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 8773
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 8774
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 8775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 8776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 8777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 8778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 8779
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 8780
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 8781
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 8782
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 8783
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 8784
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 8785
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 8786
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 8787
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 8788
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 8789
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 8790
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 8791
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 8792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 8793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 8794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 8795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 8796
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 8797
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1059,
+      "step": 8798
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 8799
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 8800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 8801
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 8802
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 8803
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0998,
+      "step": 8804
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 8805
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1054,
+      "step": 8806
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 8807
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 8808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 8809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 8810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1149,
+      "step": 8811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1169,
+      "step": 8812
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 8813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1711,
+      "step": 8814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 8815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 8816
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 8817
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 8818
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 8819
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1319,
+      "step": 8820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 8821
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 8822
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 8823
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 8824
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1731,
+      "step": 8825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 8826
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 8827
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 8828
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 8829
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 8830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.144,
+      "step": 8831
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 8832
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1439,
+      "step": 8833
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 8834
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 8835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 8836
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 8837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 8838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 8839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1063,
+      "step": 8840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 8841
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 8842
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 8843
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 8844
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 8845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 8846
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 8847
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 8848
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1205,
+      "step": 8849
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 8850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 8851
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 8852
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 8853
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 8854
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 8855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 8856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1281,
+      "step": 8857
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 8858
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 8859
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 8860
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 8861
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 8862
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 8863
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 8864
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 8865
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1122,
+      "step": 8866
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 8867
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1321,
+      "step": 8868
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 8869
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 8870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 8871
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 8872
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 8873
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1192,
+      "step": 8874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 8875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1043,
+      "step": 8876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 8877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 8878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 8879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 8880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 8881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1173,
+      "step": 8882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0934,
+      "step": 8883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 8884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 8885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 8886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 8887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 8888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 8889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 8890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 8891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 8892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 8893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 8894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1228,
+      "step": 8895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 8896
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 8897
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 8898
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1089,
+      "step": 8899
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.125,
+      "step": 8900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 8901
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 8902
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0894,
+      "step": 8903
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.11,
+      "step": 8904
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1398,
+      "step": 8905
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 8906
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 8907
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 8908
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 8909
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 8910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 8911
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 8912
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 8913
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 8914
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 8915
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1429,
+      "step": 8916
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 8917
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 8918
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1543,
+      "step": 8919
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1325,
+      "step": 8920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 8921
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 8922
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 8923
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 8924
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 8925
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 8926
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1163,
+      "step": 8927
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1213,
+      "step": 8928
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 8929
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 8930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 8931
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 8932
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 8933
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 8934
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 8935
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 8936
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 8937
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 8938
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 8939
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 8940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 8941
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0998,
+      "step": 8942
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1256,
+      "step": 8943
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1194,
+      "step": 8944
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 8945
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 8946
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 8947
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 8948
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 8949
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 8950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 8951
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 8952
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 8953
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 8954
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 8955
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 8956
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 8957
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 8958
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 8959
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 8960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1246,
+      "step": 8961
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 8962
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 8963
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1397,
+      "step": 8964
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 8965
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 8966
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 8967
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 8968
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 8969
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 8970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 8971
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 8972
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 8973
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 8974
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 8975
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 8976
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 8977
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1317,
+      "step": 8978
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 8979
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 8980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 8981
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 8982
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0857,
+      "step": 8983
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 8984
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 8985
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1423,
+      "step": 8986
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 8987
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 8988
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 8989
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 8990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 8991
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 8992
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 8993
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 8994
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 8995
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1323,
+      "step": 8996
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 8997
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 8998
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 8999
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 9000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 9001
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 9002
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 9003
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 9004
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1369,
+      "step": 9005
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 9006
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1321,
+      "step": 9007
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 9008
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 9009
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 9010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1165,
+      "step": 9011
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 9012
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 9013
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 9014
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 9015
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 9016
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1206,
+      "step": 9017
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1408,
+      "step": 9018
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 9019
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1203,
+      "step": 9020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 9021
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 9022
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 9023
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 9024
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 9025
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9026
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 9027
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 9028
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 9029
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1091,
+      "step": 9030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 9031
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 9032
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 9033
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 9034
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1002,
+      "step": 9035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 9036
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 9037
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 9038
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 9039
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1259,
+      "step": 9040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.106,
+      "step": 9041
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 9042
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 9043
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 9044
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 9045
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 9046
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 9047
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 9048
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 9049
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 9050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 9051
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 9052
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1515,
+      "step": 9053
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 9054
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 9055
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 9056
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 9057
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 9058
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1019,
+      "step": 9059
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 9061
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 9062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 9063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1618,
+      "step": 9064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1178,
+      "step": 9065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 9066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1021,
+      "step": 9067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1055,
+      "step": 9068
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 9069
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 9070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1065,
+      "step": 9071
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 9072
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 9073
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1278,
+      "step": 9074
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 9075
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 9076
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 9077
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1347,
+      "step": 9078
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0991,
+      "step": 9079
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 9080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 9081
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 9082
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0923,
+      "step": 9083
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1274,
+      "step": 9084
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 9085
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 9086
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 9087
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1241,
+      "step": 9088
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 9089
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 9090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 9091
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 9092
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 9093
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9094
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 9095
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1358,
+      "step": 9096
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0924,
+      "step": 9097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 9098
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 9099
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 9100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 9101
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1071,
+      "step": 9102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 9103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 9104
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1075,
+      "step": 9105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 9106
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 9107
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 9108
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 9109
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 9110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 9111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 9112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 9113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 9114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 9115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 9116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1226,
+      "step": 9117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 9118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 9119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 9120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 9121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 9122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 9123
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 9124
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 9125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0982,
+      "step": 9126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 9127
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1266,
+      "step": 9128
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 9129
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 9130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 9131
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1493,
+      "step": 9132
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 9133
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 9134
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 9135
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1059,
+      "step": 9136
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 9137
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.112,
+      "step": 9138
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 9139
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 9140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 9141
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1118,
+      "step": 9142
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1287,
+      "step": 9143
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9144
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 9145
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0904,
+      "step": 9146
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 9147
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 9148
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 9149
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1249,
+      "step": 9150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 9151
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 9152
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 9153
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 9154
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 9155
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1201,
+      "step": 9156
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 9157
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1371,
+      "step": 9158
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1402,
+      "step": 9159
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 9160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 9161
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 9162
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0882,
+      "step": 9163
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0999,
+      "step": 9164
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 9165
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 9166
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 9167
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 9168
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 9169
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1208,
+      "step": 9170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 9171
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1226,
+      "step": 9172
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 9173
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 9174
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1334,
+      "step": 9175
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 9176
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1117,
+      "step": 9177
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 9178
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 9179
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 9180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 9181
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 9182
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.155,
+      "step": 9183
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 9184
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 9185
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 9186
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 9187
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 9188
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 9189
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 9190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.136,
+      "step": 9191
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 9192
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 9193
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 9194
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 9195
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1041,
+      "step": 9196
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0982,
+      "step": 9197
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 9198
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 9199
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 9200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 9201
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 9202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 9203
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 9204
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 9205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 9206
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 9207
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 9208
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 9209
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 9210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 9211
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 9212
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 9213
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.115,
+      "step": 9214
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 9215
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 9216
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 9217
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1106,
+      "step": 9218
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 9219
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 9220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 9221
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 9222
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 9223
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 9224
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 9225
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 9226
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 9227
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 9228
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 9229
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 9230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1296,
+      "step": 9231
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 9232
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 9233
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.12,
+      "step": 9234
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9235
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 9236
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9237
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1547,
+      "step": 9238
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.1101,
+      "step": 9239
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 9240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 9241
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 9242
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1409,
+      "step": 9243
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 9244
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 9245
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.118,
+      "step": 9246
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 9247
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 9248
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0938,
+      "step": 9249
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 9250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1412,
+      "step": 9251
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 9252
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 9253
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 9254
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 9255
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 9256
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 9257
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 9258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 9259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 9260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.119,
+      "step": 9261
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1806,
+      "step": 9262
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 9263
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1353,
+      "step": 9264
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 9265
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1263,
+      "step": 9266
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 9267
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1064,
+      "step": 9268
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 9269
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1183,
+      "step": 9270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 9271
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 9272
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 9273
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 9274
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 9275
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 9276
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 9277
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1175,
+      "step": 9278
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 9279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 9280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 9281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1769,
+      "step": 9282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 9283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1239,
+      "step": 9284
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 9285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 9286
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 9287
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 9288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 9289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 9290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 9291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1059,
+      "step": 9292
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 9293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 9294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 9295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1891,
+      "step": 9296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 9297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1331,
+      "step": 9298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1329,
+      "step": 9299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 9300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 9301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1479,
+      "step": 9302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 9303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 9304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 9305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 9306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 9307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 9308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 9309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1489,
+      "step": 9310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1361,
+      "step": 9311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1358,
+      "step": 9312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 9313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 9314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 9315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1119,
+      "step": 9316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 9317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1188,
+      "step": 9318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 9319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 9320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 9321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 9322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 9323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.132,
+      "step": 9324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 9325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 9326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 9327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 9328
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 9329
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1215,
+      "step": 9330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 9331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 9332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 9333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 9334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1313,
+      "step": 9335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1305,
+      "step": 9336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 9337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 9338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 9339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 9340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 9341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 9342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 9343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 9344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.126,
+      "step": 9345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 9346
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 9347
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 9348
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0883,
+      "step": 9349
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.128,
+      "step": 9350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 9351
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 9352
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 9353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 9354
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 9355
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 9356
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1312,
+      "step": 9357
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1137,
+      "step": 9358
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 9359
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 9360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1136,
+      "step": 9361
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 9362
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 9363
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 9364
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 9365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 9366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 9367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 9368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 9369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 9370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 9371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 9372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 9373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0975,
+      "step": 9374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 9375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 9376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 9377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 9378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 9379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 9380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 9381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 9382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1466,
+      "step": 9383
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 9384
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 9385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1212,
+      "step": 9386
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 9387
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 9388
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 9389
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 9390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 9391
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1021,
+      "step": 9392
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 9393
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 9394
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 9395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1373,
+      "step": 9396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 9397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 9398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 9399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 9400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 9401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 9403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 9404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 9405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 9406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 9407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 9408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 9409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 9410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.116,
+      "step": 9411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1335,
+      "step": 9412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 9413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 9414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 9415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 9416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1395,
+      "step": 9417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 9418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1273,
+      "step": 9419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 9420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0948,
+      "step": 9421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 9422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 9423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.123,
+      "step": 9424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1139,
+      "step": 9425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 9426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 9427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 9428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 9429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 9430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 9431
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 9432
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1029,
+      "step": 9433
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 9434
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 9435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 9436
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.126,
+      "step": 9437
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1051,
+      "step": 9438
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.131,
+      "step": 9439
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 9440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1291,
+      "step": 9441
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 9442
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 9443
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 9444
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 9445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 9446
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1546,
+      "step": 9447
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 9448
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 9449
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 9450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 9451
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 9452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 9453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 9454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 9455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 9456
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 9457
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 9458
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 9459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 9460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 9461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 9462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1209,
+      "step": 9463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 9464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 9465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 9466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 9467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1322,
+      "step": 9468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 9469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1595,
+      "step": 9470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 9471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 9472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 9473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 9474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 9475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 9476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 9477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 9478
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 9479
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 9480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 9481
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 9482
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 9483
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1269,
+      "step": 9484
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.108,
+      "step": 9485
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 9486
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 9487
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1285,
+      "step": 9488
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1283,
+      "step": 9489
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1362,
+      "step": 9490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 9491
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 9492
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 9493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 9494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1449,
+      "step": 9495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1035,
+      "step": 9496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 9497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 9498
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 9499
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 9500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 9501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 9502
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 9503
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 9504
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 9505
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 9506
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 9507
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 9508
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 9509
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 9510
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 9511
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 9512
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 9513
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 9514
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 9515
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 9516
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 9517
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 9518
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 9519
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 9520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0934,
+      "step": 9521
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 9522
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 9523
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 9524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 9525
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 9526
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 9527
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 9528
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 9529
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 9530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 9531
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 9532
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 9533
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 9534
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 9535
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 9536
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 9537
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 9538
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 9539
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 9540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 9541
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 9542
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 9543
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 9544
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 9545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 9546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 9547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 9548
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 9549
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 9550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 9551
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 9552
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 9553
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 9554
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 9555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 9556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 9557
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 9558
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 9559
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 9560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 9561
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 9562
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 9563
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 9564
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 9565
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 9566
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 9567
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 9568
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 9569
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 9570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 9571
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 9572
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 9573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 9574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 9575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 9576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 9577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 9578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 9579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 9580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 9581
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 9582
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 9583
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 9584
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 9585
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 9586
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 9587
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9588
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 9589
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 9590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 9591
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 9592
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 9593
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 9594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 9595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 9596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 9597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 9598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 9599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 9600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 9601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 9602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 9603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 9604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 9605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 9606
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 9607
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 9608
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 9609
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 9610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 9611
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 9612
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 9613
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 9614
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 9615
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 9616
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 9617
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 9618
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 9619
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 9620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 9621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 9622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 9623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 9624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 9625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 9626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 9627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 9628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 9629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 9630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 9631
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 9632
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 9633
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 9634
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 9635
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 9636
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 9637
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 9638
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 9639
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 9640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 9641
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 9642
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 9643
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 9644
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 9645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 9646
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 9647
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 9648
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 9649
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 9650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 9651
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 9652
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 9653
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 9654
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 9655
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 9656
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 9657
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 9658
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 9659
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 9660
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 9661
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 9662
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 9663
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 9664
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 9665
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 9666
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 9667
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 9668
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 9669
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 9670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 9671
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 9672
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 9673
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 9674
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 9675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 9676
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 9677
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 9678
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 9679
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 9680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0896,
+      "step": 9681
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 9682
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 9683
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 9684
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 9685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 9686
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 9687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 9688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 9689
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 9690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 9691
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 9692
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 9693
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 9694
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 9695
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 9696
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 9697
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 9698
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 9699
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 9700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 9701
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 9702
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 9703
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 9704
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 9705
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 9706
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 9707
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 9708
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 9709
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 9710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 9711
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 9712
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 9713
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 9714
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 9715
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 9716
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 9717
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 9718
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 9719
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 9720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.1126,
+      "step": 9721
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 9722
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 9723
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 9724
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 9725
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 9726
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 9727
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 9728
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 9729
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 9730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 9731
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 9732
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 9733
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 9734
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 9735
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 9736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 9737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 9738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 9739
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 9740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 9741
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 9742
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 9743
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 9744
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 9745
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 9746
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 9747
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 9748
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 9749
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 9750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 9751
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 9752
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 9753
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 9754
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 9755
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 9756
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 9757
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 9758
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9759
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 9760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 9761
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 9762
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 9763
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 9764
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 9765
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 9766
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 9767
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 9768
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 9769
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 9770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 9771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 9772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 9773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 9774
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 9775
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 9776
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 9777
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 9778
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 9779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 9780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 9781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 9782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 9783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 9784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 9785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 9786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 9787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 9788
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 9789
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 9790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 9791
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 9792
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 9793
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 9794
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 9795
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 9796
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 9797
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 9798
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 9799
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 9800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 9801
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9802
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 9803
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 9804
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 9805
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 9806
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 9807
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 9808
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 9809
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 9810
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 9811
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 9812
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 9813
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 9814
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 9815
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 9816
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 9817
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 9818
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 9819
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 9820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 9821
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 9822
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 9823
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 9824
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 9825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 9826
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 9827
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 9828
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 9829
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 9830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 9831
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 9832
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 9833
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 9834
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 9835
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 9836
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 9837
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 9838
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1022,
+      "step": 9839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 9840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 9841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 9842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 9843
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 9844
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 9845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 9846
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 9847
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 9848
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 9849
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 9850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 9851
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 9852
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 9853
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 9854
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 9855
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 9856
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 9857
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 9858
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 9859
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 9860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 9861
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 9862
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 9863
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 9864
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 9865
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 9866
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 9867
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 9868
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 9869
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 9870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 9871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 9872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 9873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 9874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 9875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 9876
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 9877
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 9878
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 9879
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 9880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 9881
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 9882
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 9883
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 9884
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 9885
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 9886
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 9887
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 9888
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 9889
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 9890
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 9891
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 9892
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 9893
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 9894
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 9895
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 9896
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 9897
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 9898
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 9899
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 9900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 9901
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 9902
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 9903
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 9904
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 9905
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 9906
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 9907
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 9908
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 9909
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 9910
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 9911
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 9912
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 9913
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 9914
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 9915
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 9916
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 9917
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 9918
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 9919
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0964,
+      "step": 9920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 9921
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1243,
+      "step": 9922
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 9923
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 9924
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 9925
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 9926
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 9927
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 9928
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 9929
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 9930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 9931
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 9932
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 9933
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 9934
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 9935
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 9936
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 9937
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 9938
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 9939
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 9940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 9941
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 9942
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 9943
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 9944
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 9945
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 9946
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 9947
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 9948
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 9949
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 9950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 9951
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 9952
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 9953
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 9954
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 9955
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 9956
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 9957
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 9958
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 9959
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 9960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 9961
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 9962
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 9963
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 9964
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 9965
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 9966
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 9967
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 9968
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 9969
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 9970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 9971
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 9972
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 9973
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 9974
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 9975
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 9976
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 9977
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 9978
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 9979
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 9980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 9981
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 9982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 9983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 9984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 9985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 9986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 9987
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 9988
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 9989
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 9990
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 9991
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 9992
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 9993
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 9994
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 9995
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 9996
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 9997
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 9998
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 9999
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 10000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 10001
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 10002
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 10003
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 10004
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 10005
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 10006
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 10007
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.1093,
+      "step": 10008
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 10009
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 10010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 10011
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 10012
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 10013
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 10014
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 10015
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 10016
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 10017
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10018
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 10019
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 10020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 10021
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 10022
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 10023
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 10024
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 10025
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 10026
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 10027
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 10028
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 10029
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 10030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 10031
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 10032
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 10033
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 10034
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 10035
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 10036
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 10037
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 10038
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 10039
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 10040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 10041
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 10042
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 10043
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 10044
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 10045
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 10046
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 10047
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 10048
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 10049
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 10050
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 10051
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 10052
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 10053
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 10054
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 10055
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 10056
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 10057
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 10058
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 10059
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 10060
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 10061
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 10062
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 10063
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 10064
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 10065
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 10066
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 10067
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 10068
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 10069
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 10070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 10071
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 10072
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 10073
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 10074
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 10075
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 10076
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 10077
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 10078
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 10079
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 10080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 10081
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 10082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 10083
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 10084
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 10085
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 10086
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 10087
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 10088
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 10089
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.1087,
+      "step": 10090
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 10091
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 10092
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 10093
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0775,
+      "step": 10094
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 10095
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 10096
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 10097
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 10098
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 10099
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 10100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 10101
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 10102
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 10103
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 10104
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 10105
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 10106
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 10107
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 10108
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 10109
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 10110
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 10111
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 10112
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 10113
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 10114
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 10115
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 10116
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 10117
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 10118
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0932,
+      "step": 10119
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 10120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 10121
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 10122
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 10123
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10124
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 10125
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 10126
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 10127
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 10128
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 10129
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 10130
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 10131
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 10132
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 10133
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 10134
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 10135
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10136
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 10137
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 10138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10139
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 10140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 10141
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 10142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 10143
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 10144
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 10145
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 10146
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 10147
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 10148
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 10149
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 10150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 10151
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 10152
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 10153
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 10154
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 10155
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 10156
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 10157
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 10158
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 10159
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 10160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 10161
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 10162
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 10163
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 10164
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 10165
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 10166
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 10167
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 10168
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 10169
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 10170
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 10171
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 10172
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 10173
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 10174
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 10175
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 10176
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 10177
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 10178
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10179
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 10180
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 10181
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10182
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 10183
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 10184
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 10185
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 10186
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 10187
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 10188
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 10189
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 10190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 10191
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 10192
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 10193
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 10194
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 10195
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 10196
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 10197
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 10198
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 10199
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 10200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 10201
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10202
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 10203
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 10204
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 10205
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 10206
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 10207
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 10208
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 10209
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 10210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 10211
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 10212
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 10213
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 10214
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 10215
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 10216
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 10217
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 10218
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 10219
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 10220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 10221
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 10222
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 10223
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 10224
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 10225
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 10226
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 10227
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 10228
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 10229
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 10230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 10231
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 10232
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 10233
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 10234
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 10235
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 10236
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 10237
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 10238
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 10239
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 10240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 10241
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 10242
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 10243
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 10244
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10245
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 10246
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 10247
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 10248
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 10249
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 10250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 10251
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 10252
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 10253
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 10254
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 10255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 10256
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 10257
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 10258
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 10259
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 10260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 10261
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 10262
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 10263
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 10264
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 10265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 10266
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 10267
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 10268
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 10269
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 10270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 10271
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 10272
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 10273
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 10274
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 10275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 10276
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 10277
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 10278
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 10279
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 10281
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 10282
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0179,
+      "step": 10283
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 10284
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 10285
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 10286
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 10287
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 10288
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 10289
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 10290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 10291
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 10292
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 10293
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 10294
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 10295
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 10296
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 10297
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 10298
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10299
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 10300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 10301
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 10302
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 10303
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 10304
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 10305
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 10306
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 10307
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 10308
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 10309
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 10310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 10311
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 10312
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10313
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.1162,
+      "step": 10314
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 10315
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 10316
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 10317
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 10318
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 10319
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 10320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 10321
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 10322
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 10323
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 10324
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 10325
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 10326
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 10327
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 10328
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 10329
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 10330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 10331
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 10332
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 10333
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 10334
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 10335
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 10336
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 10337
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 10338
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 10339
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 10340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10341
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 10342
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 10343
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 10344
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 10345
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 10346
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 10347
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 10348
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 10349
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 10350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 10351
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 10352
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 10353
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 10354
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 10355
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 10356
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1161,
+      "step": 10357
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 10358
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0848,
+      "step": 10359
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 10360
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 10361
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 10362
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 10363
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 10364
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 10365
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 10366
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 10367
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.1339,
+      "step": 10368
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 10369
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 10370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 10371
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 10372
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 10373
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 10374
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 10375
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 10376
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 10377
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 10378
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 10379
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 10380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 10381
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 10382
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 10383
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1088,
+      "step": 10384
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 10385
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 10386
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 10387
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 10388
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 10389
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 10390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 10391
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 10392
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 10393
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 10394
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 10395
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 10396
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 10397
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 10398
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 10399
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 10400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 10401
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 10402
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 10403
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 10404
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 10405
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 10406
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 10407
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 10408
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 10409
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 10410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 10411
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 10412
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0171,
+      "step": 10413
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 10414
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 10415
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 10416
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10417
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 10418
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 10419
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 10420
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 10421
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 10422
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 10423
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 10424
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 10425
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 10426
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 10427
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 10428
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 10429
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 10430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 10431
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 10432
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 10433
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 10434
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 10435
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 10436
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 10437
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 10438
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 10439
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 10440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 10441
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 10442
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 10443
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 10444
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 10445
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 10446
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 10447
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 10448
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 10449
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 10450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 10451
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 10452
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 10453
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 10454
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 10455
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 10456
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 10457
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 10458
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 10459
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 10460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 10461
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 10462
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 10463
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 10464
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 10465
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 10466
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 10467
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 10468
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 10469
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 10470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 10471
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 10472
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 10473
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 10474
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 10475
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 10476
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 10477
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 10478
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.1028,
+      "step": 10479
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 10480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 10481
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 10482
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 10483
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 10484
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 10485
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 10486
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 10487
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 10488
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 10489
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 10490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 10491
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 10492
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 10493
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 10494
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0843,
+      "step": 10495
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 10496
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 10497
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 10498
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 10499
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 10500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 10501
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 10502
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 10503
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 10504
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 10505
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 10506
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 10507
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 10508
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 10509
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 10510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10511
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 10512
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 10513
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 10514
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 10515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 10516
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 10517
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 10518
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10519
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 10520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 10521
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 10522
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 10523
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 10524
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 10525
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 10526
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 10527
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 10528
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 10529
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 10530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 10531
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 10532
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 10533
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 10534
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10535
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 10536
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 10537
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 10538
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 10539
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 10540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 10541
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 10542
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 10543
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 10544
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 10545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 10546
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 10547
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 10548
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 10549
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 10550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 10551
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 10552
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 10553
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 10554
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 10555
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 10556
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10557
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 10558
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 10559
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 10560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 10561
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 10562
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 10563
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10564
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 10565
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 10566
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 10567
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 10568
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 10569
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 10570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 10571
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 10572
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 10573
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 10574
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 10575
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10576
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 10577
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 10578
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 10579
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 10580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 10581
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 10582
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 10583
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 10584
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 10585
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 10586
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 10587
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 10588
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.1152,
+      "step": 10589
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 10590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 10591
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10592
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 10593
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 10594
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 10595
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 10596
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 10597
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10598
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 10599
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 10600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 10601
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.107,
+      "step": 10602
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 10603
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 10604
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 10605
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 10606
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 10607
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0958,
+      "step": 10608
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 10609
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 10610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 10611
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 10612
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 10613
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 10614
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 10615
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 10616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 10617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 10618
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 10620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 10621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 10622
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 10623
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 10624
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 10626
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 10627
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 10628
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 10629
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 10630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 10632
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 10633
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 10634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 10635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 10636
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 10637
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 10638
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 10639
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 10640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 10641
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 10642
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 10643
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 10644
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 10645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 10646
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 10647
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 10648
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 10649
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 10650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 10651
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 10652
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 10653
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 10654
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 10655
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 10656
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 10657
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 10658
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 10659
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 10660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 10661
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 10662
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 10663
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 10664
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 10665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 10666
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 10667
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 10668
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 10669
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 10670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 10671
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 10672
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 10673
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0928,
+      "step": 10674
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 10675
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 10676
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 10677
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 10678
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 10679
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 10680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 10681
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 10682
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 10683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 10684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 10685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 10687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 10688
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 10689
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 10690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 10691
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10692
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 10693
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 10694
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 10695
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 10696
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 10697
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10698
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 10699
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 10700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 10701
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 10702
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 10703
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 10704
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 10705
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 10706
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 10707
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 10708
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 10709
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 10710
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 10711
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 10712
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 10713
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 10714
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 10715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 10716
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 10717
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 10718
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 10719
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 10720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 10721
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 10722
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.1033,
+      "step": 10723
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 10724
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 10725
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 10726
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 10727
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 10728
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 10729
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 10731
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 10732
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 10733
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 10734
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 10735
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 10736
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 10737
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 10738
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 10739
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 10740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 10741
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 10742
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 10743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 10744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 10745
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 10746
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 10747
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 10748
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 10749
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 10750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 10751
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 10752
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 10753
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 10754
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 10755
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 10756
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 10757
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 10758
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 10759
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 10760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 10761
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 10762
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 10763
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 10764
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 10765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 10766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10767
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 10768
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 10769
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 10770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 10771
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 10772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 10773
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 10774
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 10775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 10776
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 10777
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 10778
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 10779
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 10780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 10781
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 10782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 10783
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 10784
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 10785
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 10786
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 10787
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 10788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 10789
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 10790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 10791
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 10792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 10793
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 10794
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 10795
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 10796
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 10797
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 10798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 10799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 10800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 10801
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 10802
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 10803
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 10804
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 10805
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 10806
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 10807
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 10808
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 10809
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 10810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 10811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 10812
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 10813
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 10814
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 10815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 10816
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 10817
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 10818
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 10819
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 10820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 10821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 10822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 10823
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 10824
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 10825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 10826
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 10827
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 10828
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10829
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 10830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 10831
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 10832
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 10833
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 10834
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 10835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 10836
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 10837
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 10838
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 10839
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 10840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 10841
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 10842
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 10843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 10844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 10845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 10846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 10847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.1129,
+      "step": 10848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 10849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0943,
+      "step": 10850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 10851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 10852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 10853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 10854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 10855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 10856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 10857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 10858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 10859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0843,
+      "step": 10860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 10861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 10862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 10863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 10864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 10865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 10866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 10867
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0977,
+      "step": 10868
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 10869
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 10870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 10871
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 10872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 10873
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 10874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 10875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 10876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 10877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 10878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 10879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 10880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 10881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 10882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 10883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 10884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 10885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 10886
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 10887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0939,
+      "step": 10888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 10889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 10890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 10891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 10892
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 10893
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 10894
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 10895
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 10896
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 10897
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.098,
+      "step": 10898
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 10899
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 10900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 10901
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 10902
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 10903
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 10904
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 10905
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10906
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 10907
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 10908
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 10909
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 10910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 10911
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 10912
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10913
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 10914
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 10915
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 10916
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 10917
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 10918
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 10919
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 10920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 10921
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 10922
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 10923
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 10924
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 10925
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 10926
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 10927
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 10928
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 10929
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 10930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 10931
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 10932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 10933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 10934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 10935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 10936
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 10937
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 10938
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 10939
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 10940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 10941
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 10942
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 10943
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 10944
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 10945
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 10946
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 10947
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 10948
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 10949
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 10950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 10951
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 10952
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 10953
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 10954
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 10955
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 10956
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 10957
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 10958
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 10959
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 10960
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 10961
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 10962
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 10963
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 10964
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 10965
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 10966
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 10967
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 10968
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 10969
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 10970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 10971
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 10972
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 10973
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 10974
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 10975
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 10976
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 10977
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 10978
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 10979
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 10980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 10981
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.1046,
+      "step": 10982
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 10983
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 10984
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 10985
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 10986
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 10987
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 10988
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 10989
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 10990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 10991
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 10992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 10993
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 10994
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 10995
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 10996
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 10997
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 10998
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 10999
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 11000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11001
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 11002
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0203,
+      "step": 11003
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 11004
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 11005
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 11006
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 11007
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 11008
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0881,
+      "step": 11009
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 11010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 11011
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0843,
+      "step": 11012
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 11013
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 11014
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 11015
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 11016
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 11017
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 11018
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 11019
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 11020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 11021
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 11022
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 11023
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 11024
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 11025
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 11026
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 11027
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 11028
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 11029
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 11030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 11031
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 11032
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 11033
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 11034
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 11035
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 11036
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 11037
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 11038
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 11039
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 11040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 11041
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 11042
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 11043
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 11044
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 11045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 11046
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 11047
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 11048
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 11049
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 11050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 11051
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 11052
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 11053
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 11054
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11055
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 11056
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 11057
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 11058
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 11059
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 11060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 11061
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 11062
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 11063
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 11064
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 11065
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 11066
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 11067
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 11068
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 11069
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 11070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 11071
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 11072
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 11074
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 11075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 11076
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.1176,
+      "step": 11077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 11079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 11080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 11081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 11082
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 11083
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 11084
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 11085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 11086
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 11087
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 11088
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 11089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 11090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 11091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 11092
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 11093
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 11094
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 11095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 11096
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 11097
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 11098
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 11099
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 11100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 11101
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 11102
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 11103
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 11104
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 11105
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 11106
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 11107
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11108
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 11109
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 11110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 11111
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 11112
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 11113
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 11114
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 11115
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 11116
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 11117
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 11118
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 11119
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 11120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 11121
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 11122
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 11123
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 11124
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 11125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 11126
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 11127
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 11128
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 11129
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 11130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 11131
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11132
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 11133
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 11134
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 11135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 11136
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 11137
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 11138
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 11139
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 11140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 11141
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 11142
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 11143
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 11144
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 11145
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 11146
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 11147
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 11148
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 11149
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 11150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0716,
+      "step": 11151
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 11152
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 11153
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 11154
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 11155
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 11156
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 11157
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11158
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 11159
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 11160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 11161
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 11162
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 11163
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 11164
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 11165
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 11166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.1005,
+      "step": 11167
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 11168
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 11169
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 11171
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11172
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 11173
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 11174
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 11175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 11176
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 11177
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 11178
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 11179
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 11180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 11181
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 11182
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 11183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 11184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 11185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 11187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11188
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 11189
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 11190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 11191
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 11192
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 11193
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 11194
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 11195
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 11196
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 11197
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 11198
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 11199
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 11200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 11201
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11202
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 11203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 11204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 11205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 11206
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 11207
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0908,
+      "step": 11208
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11209
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 11210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 11211
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 11212
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 11213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 11214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 11215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 11216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 11217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 11218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 11219
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11220
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 11221
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 11222
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 11223
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11224
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 11225
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 11226
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 11227
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 11228
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 11229
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 11230
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 11231
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 11232
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 11233
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 11234
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 11235
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 11236
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.099,
+      "step": 11237
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 11238
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 11239
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 11240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 11241
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 11242
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 11243
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 11244
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11245
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 11246
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 11247
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 11248
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 11249
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 11250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 11251
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 11252
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 11253
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 11254
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1078,
+      "step": 11255
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0906,
+      "step": 11256
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 11257
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 11258
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 11259
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 11260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 11261
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 11262
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 11263
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 11264
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 11265
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 11266
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 11267
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 11268
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11269
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 11270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 11271
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 11272
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 11273
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 11274
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 11275
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 11276
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 11277
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 11278
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 11279
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 11280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 11281
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 11282
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 11283
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 11284
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 11285
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 11286
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 11287
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 11288
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 11289
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0807,
+      "step": 11290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 11291
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 11292
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 11293
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 11294
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 11295
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 11296
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 11297
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 11298
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 11299
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 11300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 11301
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 11302
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 11303
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 11304
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 11305
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 11306
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 11307
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 11308
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 11309
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11310
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 11311
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 11312
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 11313
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 11314
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 11315
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 11316
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 11317
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 11318
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 11319
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 11320
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 11321
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 11322
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 11323
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 11324
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 11325
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 11326
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 11327
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 11328
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 11329
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 11330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 11331
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 11332
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 11333
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 11334
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 11335
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 11336
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 11337
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 11338
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 11339
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 11340
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11341
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 11342
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 11343
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 11344
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 11345
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 11346
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 11347
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 11348
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 11349
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 11350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 11351
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 11352
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 11353
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 11354
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 11355
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 11356
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11357
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 11358
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 11359
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 11360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 11361
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 11362
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 11363
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 11364
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 11365
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 11366
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 11367
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 11368
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 11369
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 11371
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 11372
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 11373
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 11374
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 11375
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11376
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 11377
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 11378
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 11379
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 11380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 11381
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 11382
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 11383
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 11384
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 11385
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 11386
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 11387
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 11388
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 11389
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 11390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 11391
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 11392
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 11393
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 11394
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 11395
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 11396
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 11397
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 11398
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 11399
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 11400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 11401
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 11402
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 11403
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11404
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 11405
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 11406
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 11407
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 11408
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 11409
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 11410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 11411
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 11412
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 11413
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 11414
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 11415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 11416
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 11417
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 11418
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 11419
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 11420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 11421
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 11422
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 11423
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 11424
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 11425
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 11426
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 11427
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 11428
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 11429
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 11430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 11431
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 11432
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 11433
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 11434
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 11435
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 11436
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 11437
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 11438
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 11439
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 11440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 11441
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 11442
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 11443
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 11444
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 11445
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 11446
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 11447
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 11448
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 11449
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 11450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 11451
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0971,
+      "step": 11452
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 11453
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 11454
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 11455
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 11456
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 11457
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 11458
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 11459
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 11460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 11461
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 11462
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 11463
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0974,
+      "step": 11464
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 11465
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 11466
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 11467
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 11468
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 11469
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 11470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 11471
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11472
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 11473
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11474
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 11475
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11476
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 11477
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 11478
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 11479
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 11480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 11481
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 11482
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 11483
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11484
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 11485
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 11486
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 11487
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 11488
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 11489
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 11490
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 11491
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 11492
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 11493
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 11494
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 11495
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 11496
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 11497
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 11498
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 11499
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 11500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 11501
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 11502
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 11503
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 11504
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11505
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 11506
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 11507
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 11508
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 11509
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 11510
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 11511
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 11512
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 11513
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 11514
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 11515
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 11516
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11517
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 11518
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 11519
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 11520
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 11521
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 11522
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 11523
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 11524
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 11525
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 11526
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 11527
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 11528
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11529
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 11530
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 11531
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 11532
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 11533
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 11534
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 11535
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.1034,
+      "step": 11536
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 11537
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 11538
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 11539
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 11540
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 11541
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 11542
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 11543
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 11544
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 11545
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0806,
+      "step": 11546
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 11547
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 11548
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 11549
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 11550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 11551
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 11552
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 11553
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 11554
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 11555
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 11556
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 11557
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 11558
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 11559
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 11560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 11561
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 11562
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 11563
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 11564
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 11565
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 11566
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 11567
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 11568
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 11569
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 11570
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0988,
+      "step": 11571
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 11572
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 11573
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 11574
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 11575
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 11576
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 11577
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 11578
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 11579
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 11580
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 11581
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 11582
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 11583
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 11584
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 11585
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 11586
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 11587
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 11588
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 11589
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 11590
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 11591
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 11592
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 11593
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 11594
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 11595
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 11596
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 11597
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 11598
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 11599
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 11600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 11601
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 11602
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 11603
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 11604
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 11605
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 11606
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 11607
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 11608
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11609
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 11610
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 11611
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 11612
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 11613
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 11614
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 11615
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11616
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 11617
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 11618
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 11619
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 11620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 11621
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 11622
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 11623
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 11624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 11625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 11626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 11627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 11628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 11629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 11630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 11631
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 11632
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11633
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 11634
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 11635
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 11636
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 11637
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 11638
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 11639
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0675,
+      "step": 11640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 11641
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 11642
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11643
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 11644
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11645
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 11646
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 11647
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 11648
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 11649
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 11650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 11651
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 11652
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 11653
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 11654
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 11655
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 11656
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 11657
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 11658
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 11659
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 11660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 11661
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 11662
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11663
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 11664
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 11665
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 11666
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 11667
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 11668
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 11669
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 11670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 11671
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 11672
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 11673
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 11674
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 11675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 11676
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 11677
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 11678
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 11679
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 11680
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 11681
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 11682
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 11683
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 11684
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 11685
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0918,
+      "step": 11686
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 11687
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 11688
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 11689
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 11690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 11691
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 11692
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 11693
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 11694
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 11695
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 11696
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 11697
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 11698
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 11699
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 11700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 11701
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 11702
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 11703
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 11704
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 11705
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 11706
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 11707
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 11708
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 11709
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 11710
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 11711
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 11712
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 11713
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 11714
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 11715
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 11716
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 11717
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 11718
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 11719
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 11720
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 11721
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 11722
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 11723
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1204,
+      "step": 11724
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 11725
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 11726
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 11727
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 11728
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 11729
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 11730
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 11731
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 11732
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 11733
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 11734
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 11735
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 11736
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 11737
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 11738
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 11739
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 11740
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 11741
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 11742
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11743
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 11744
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 11745
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 11746
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 11747
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 11748
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 11749
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 11750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11751
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 11752
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 11753
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 11754
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.1306,
+      "step": 11755
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 11756
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 11757
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 11758
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 11759
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 11760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 11761
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 11762
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 11763
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 11764
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 11765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 11766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 11767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 11768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 11769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 11770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 11771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 11772
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 11773
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 11774
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 11775
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 11776
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 11777
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 11778
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 11779
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 11780
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 11781
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 11782
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 11783
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 11784
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 11785
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 11786
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 11787
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 11788
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 11789
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 11790
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 11791
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 11792
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 11793
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11794
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 11795
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 11796
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 11797
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 11798
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 11799
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.1037,
+      "step": 11800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 11801
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 11802
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 11803
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 11804
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 11805
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 11806
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 11807
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 11808
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 11809
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 11810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 11811
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 11812
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 11813
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 11814
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 11815
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 11816
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 11817
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11818
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 11819
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 11820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 11821
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 11822
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 11823
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 11824
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 11825
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 11826
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 11827
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 11828
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 11829
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 11830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 11831
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 11832
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 11833
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 11834
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 11835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 11836
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 11837
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 11838
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 11839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 11840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 11841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 11842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 11843
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 11844
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 11845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 11846
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 11847
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.1083,
+      "step": 11848
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 11849
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 11850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 11851
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 11852
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 11853
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 11854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 11855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1014,
+      "step": 11856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 11857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 11858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 11859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 11860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 11861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 11862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 11863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 11864
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 11865
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 11866
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 11867
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 11868
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 11869
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 11870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 11871
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 11872
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 11873
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 11874
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 11875
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 11876
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 11877
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 11878
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 11879
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 11880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 11881
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 11882
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 11883
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 11884
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.1232,
+      "step": 11885
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 11886
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 11887
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 11888
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 11889
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 11890
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 11891
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0937,
+      "step": 11892
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 11893
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 11894
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 11895
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 11896
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 11897
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 11898
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 11899
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 11900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 11901
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 11902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 11903
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 11904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 11905
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 11906
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 11907
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 11908
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 11909
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 11910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 11911
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 11912
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 11913
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 11914
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 11915
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1067,
+      "step": 11916
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0983,
+      "step": 11917
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 11918
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0879,
+      "step": 11919
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 11920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 11921
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 11922
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 11923
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 11924
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 11925
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 11926
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 11927
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 11928
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 11929
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 11930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 11931
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 11932
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 11933
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 11934
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 11935
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 11936
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 11937
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 11938
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0927,
+      "step": 11939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 11940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 11941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 11942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 11943
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 11944
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 11945
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 11946
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.1006,
+      "step": 11947
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 11948
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 11949
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 11950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 11951
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 11952
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 11953
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 11954
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 11955
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 11956
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 11957
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 11958
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 11959
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 11960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 11961
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 11962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 11963
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 11964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 11965
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.09,
+      "step": 11966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0917,
+      "step": 11967
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 11968
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11969
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 11970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 11971
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 11972
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 11973
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 11974
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 11975
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 11976
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 11977
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 11978
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 11979
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 11980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 11981
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 11982
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 11983
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 11984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 11985
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 11986
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 11987
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 11988
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 11989
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 11990
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 11991
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 11992
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 11993
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 11994
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 11995
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 11996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 11997
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 11998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 11999
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 12000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 12001
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 12002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 12003
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 12004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 12005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 12006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.1213,
+      "step": 12007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 12008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 12009
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 12010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 12011
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 12012
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 12013
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 12014
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 12015
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 12016
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 12017
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 12018
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 12019
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 12020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 12021
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0785,
+      "step": 12022
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 12023
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 12024
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 12025
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 12026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 12027
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 12028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 12029
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 12030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 12031
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 12032
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 12033
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 12034
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 12035
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 12036
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 12037
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 12038
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 12039
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 12040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 12041
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 12042
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12043
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 12044
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 12045
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 12046
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 12047
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 12048
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 12049
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 12050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 12051
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 12052
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 12053
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 12054
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 12055
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 12056
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 12057
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 12058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 12059
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 12060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 12061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.1113,
+      "step": 12062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 12063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 12064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 12065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 12066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 12067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 12068
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0756,
+      "step": 12069
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 12070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 12071
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 12072
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 12073
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 12074
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 12075
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 12076
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 12077
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 12078
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 12079
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 12080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 12081
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 12082
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 12083
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 12084
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 12085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 12086
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 12087
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 12088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 12089
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 12090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 12091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 12092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 12093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 12094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 12095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 12096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1001,
+      "step": 12097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 12098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 12099
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 12100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 12101
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 12102
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 12103
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 12104
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 12105
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 12106
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12107
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 12108
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 12109
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 12110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 12111
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 12112
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12113
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12114
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 12115
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 12116
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0897,
+      "step": 12117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 12118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 12119
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 12120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 12121
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 12122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 12123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 12124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0995,
+      "step": 12125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 12126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 12127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 12128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 12129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 12130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 12131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 12132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 12133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 12134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 12135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 12136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 12137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 12138
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 12139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 12140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 12141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 12142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 12143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1024,
+      "step": 12144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 12145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 12146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 12147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 12148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 12149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 12150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 12151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 12152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 12153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 12154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 12155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 12156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 12157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 12158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 12159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 12160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 12162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 12163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 12164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 12165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 12166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 12167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.1127,
+      "step": 12168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0876,
+      "step": 12169
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 12170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 12171
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 12172
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 12173
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 12174
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 12175
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 12176
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 12177
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 12178
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 12179
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 12180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 12181
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 12182
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 12183
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 12184
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 12185
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 12186
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 12187
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1057,
+      "step": 12188
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0777,
+      "step": 12189
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 12190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 12191
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 12192
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12193
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 12194
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 12195
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 12196
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 12197
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 12198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 12199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 12200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 12201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 12202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 12203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 12204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 12205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 12206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 12207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 12208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 12209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 12210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 12211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 12212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 12213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 12214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1016,
+      "step": 12215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 12216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 12217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 12218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 12219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 12220
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 12221
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 12222
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 12223
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 12224
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 12225
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 12226
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.102,
+      "step": 12227
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 12228
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 12229
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 12230
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 12231
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 12232
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 12233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 12234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 12235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 12236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.113,
+      "step": 12237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 12238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 12239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 12240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 12241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 12242
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 12243
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 12244
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 12245
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0891,
+      "step": 12246
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12247
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 12248
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 12249
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 12250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 12251
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 12252
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 12253
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 12254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 12255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 12256
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 12257
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 12258
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 12259
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 12260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 12261
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 12262
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 12263
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 12264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 12265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 12266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 12267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 12268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 12269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 12270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 12271
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 12272
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 12273
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 12274
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 12275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 12276
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 12277
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 12278
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 12279
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 12280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 12281
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 12282
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 12283
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 12284
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 12285
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 12286
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 12287
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 12288
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0979,
+      "step": 12289
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 12290
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 12291
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 12292
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 12293
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 12294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 12295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 12296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 12297
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 12298
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 12299
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 12300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 12301
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 12302
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 12303
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 12304
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 12305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 12306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 12307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 12308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 12309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 12310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 12311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 12312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 12313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 12314
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 12315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 12316
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 12317
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 12318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0888,
+      "step": 12319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 12320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 12321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 12322
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 12323
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 12324
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.1011,
+      "step": 12325
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 12326
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 12327
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 12328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 12329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 12330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 12331
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0989,
+      "step": 12332
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 12333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 12334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 12335
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 12336
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 12337
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 12338
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 12339
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 12340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 12341
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 12342
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 12343
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 12344
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 12345
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 12346
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 12347
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 12348
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 12349
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 12350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 12351
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 12352
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 12353
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 12354
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 12355
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 12356
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 12357
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 12358
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 12359
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 12360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 12361
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 12362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 12363
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0849,
+      "step": 12364
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0813,
+      "step": 12365
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 12366
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 12367
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 12368
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 12369
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 12370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 12371
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 12372
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 12373
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 12374
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 12375
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 12376
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 12377
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 12378
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 12379
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 12380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 12381
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 12382
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 12383
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 12384
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 12385
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 12386
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 12387
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 12388
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0942,
+      "step": 12389
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0997,
+      "step": 12390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 12391
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 12392
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 12393
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 12394
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 12395
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0869,
+      "step": 12396
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 12397
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 12398
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 12399
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 12400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 12401
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 12402
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 12403
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.1218,
+      "step": 12404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 12405
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 12406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 12407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 12408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 12409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 12410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 12411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 12412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 12413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 12414
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 12415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 12416
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 12417
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 12418
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 12419
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 12420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 12421
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 12422
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 12423
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 12424
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 12425
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 12426
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 12427
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 12428
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0978,
+      "step": 12429
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0926,
+      "step": 12430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 12431
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 12432
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 12433
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 12434
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 12435
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 12436
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 12437
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 12438
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 12439
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 12440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 12441
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 12442
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 12443
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.085,
+      "step": 12444
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 12445
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 12446
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 12447
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 12448
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 12449
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 12450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 12451
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0893,
+      "step": 12452
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 12453
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 12454
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 12455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 12456
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 12457
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0919,
+      "step": 12458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 12459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 12460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 12461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 12462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 12463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0951,
+      "step": 12464
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 12465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12466
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 12467
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 12468
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 12469
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 12470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 12471
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 12472
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 12473
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 12474
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12475
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 12476
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 12477
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 12478
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 12479
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 12480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 12481
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 12482
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 12483
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1301,
+      "step": 12484
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 12485
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 12486
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 12487
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 12488
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 12489
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 12490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 12491
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1036,
+      "step": 12492
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 12493
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.1003,
+      "step": 12494
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.101,
+      "step": 12495
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 12496
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 12497
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 12498
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 12499
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 12500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 12501
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 12502
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 12503
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 12504
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 12505
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 12506
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 12507
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 12508
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0734,
+      "step": 12509
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 12510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 12511
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 12512
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 12513
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 12514
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 12515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 12516
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 12517
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 12518
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 12519
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 12521
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0911,
+      "step": 12522
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 12523
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 12524
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 12525
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 12526
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 12527
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 12528
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 12529
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 12530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 12531
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 12532
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 12533
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 12534
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1153,
+      "step": 12535
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0554,
+      "step": 12536
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 12537
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0787,
+      "step": 12538
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1012,
+      "step": 12539
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 12540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 12541
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0818,
+      "step": 12542
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 12543
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 12544
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 12545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 12546
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 12547
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 12548
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 12549
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 12550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 12551
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 12552
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 12553
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 12554
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0907,
+      "step": 12555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.1282,
+      "step": 12556
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 12557
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 12558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 12559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 12560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 12561
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 12562
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 12563
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 12564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0804,
+      "step": 12565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 12566
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 12567
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 12568
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 12569
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 12570
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1079,
+      "step": 12571
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 12572
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 12573
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 12574
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 12575
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 12576
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 12577
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 12578
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 12579
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 12580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 12581
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 12582
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0936,
+      "step": 12583
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 12584
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 12585
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 12586
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 12587
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 12588
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 12589
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 12590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 12591
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 12592
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12593
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 12594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 12595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 12596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 12597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 12598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 12599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 12600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 12601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0969,
+      "step": 12602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 12603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 12604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 12605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 12606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 12607
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 12608
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.1,
+      "step": 12609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 12610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 12611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 12612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 12613
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 12614
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 12615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 12616
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12617
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 12618
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 12619
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 12620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 12621
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1128,
+      "step": 12622
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12623
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 12624
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.113,
+      "step": 12625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 12626
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 12627
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 12628
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 12629
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 12630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 12631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 12632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 12633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0843,
+      "step": 12634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 12635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 12636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 12637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1223,
+      "step": 12638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.094,
+      "step": 12639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.1146,
+      "step": 12640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 12641
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 12642
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 12643
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 12644
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0886,
+      "step": 12645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 12646
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 12647
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 12648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 12649
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 12650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 12651
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 12652
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0992,
+      "step": 12653
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 12654
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 12655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 12656
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 12657
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 12658
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 12659
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 12660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 12661
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 12662
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0984,
+      "step": 12663
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 12664
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0847,
+      "step": 12665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 12666
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 12667
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0699,
+      "step": 12668
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 12669
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 12670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 12671
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 12672
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 12673
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 12674
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 12675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 12676
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 12677
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0889,
+      "step": 12678
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 12679
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 12680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 12681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 12682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.1131,
+      "step": 12683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 12684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 12685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 12686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0987,
+      "step": 12687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0758,
+      "step": 12688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 12689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 12690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 12691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 12692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 12693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 12694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 12695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 12696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 12697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 12698
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 12699
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 12700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12701
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 12702
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 12703
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 12704
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 12705
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 12706
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 12707
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 12708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 12709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 12710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 12711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1236,
+      "step": 12712
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 12713
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 12714
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 12715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 12716
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 12717
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 12718
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 12719
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 12720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 12721
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 12722
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 12723
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 12724
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 12725
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 12726
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 12727
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0895,
+      "step": 12728
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12729
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 12730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 12731
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 12732
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 12733
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 12734
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 12735
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 12736
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 12737
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 12738
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 12739
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 12740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 12741
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 12742
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 12743
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 12744
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 12745
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 12746
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 12747
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 12748
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 12749
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 12750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 12751
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 12752
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 12753
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 12754
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 12755
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 12756
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 12757
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12758
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0877,
+      "step": 12759
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 12760
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12761
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 12762
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 12763
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 12764
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 12765
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 12766
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 12767
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 12768
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 12769
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 12770
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 12771
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 12772
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 12773
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 12774
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 12775
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 12776
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 12777
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 12778
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 12779
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 12780
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 12781
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 12782
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 12783
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 12784
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 12785
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 12786
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 12787
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1166,
+      "step": 12788
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 12789
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 12790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 12791
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 12792
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 12793
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 12794
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0825,
+      "step": 12795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 12796
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 12797
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 12798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 12799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0789,
+      "step": 12800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 12801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 12802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 12803
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 12804
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 12805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 12806
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 12807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 12808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1047,
+      "step": 12809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 12810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 12811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 12812
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 12813
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 12814
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.076,
+      "step": 12815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 12816
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 12817
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 12818
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 12819
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 12820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 12821
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1066,
+      "step": 12822
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 12823
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 12824
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 12825
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 12826
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 12827
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 12828
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 12829
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1062,
+      "step": 12830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 12831
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0967,
+      "step": 12832
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 12833
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 12834
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 12835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0688,
+      "step": 12836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1585,
+      "step": 12837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 12838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0783,
+      "step": 12839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 12840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 12841
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 12842
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0791,
+      "step": 12843
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0943,
+      "step": 12844
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 12845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 12846
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 12847
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 12848
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 12849
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 12850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 12851
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 12852
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 12853
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 12854
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 12855
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 12856
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 12857
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 12858
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 12859
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0746,
+      "step": 12860
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 12861
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 12862
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 12863
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 12864
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 12865
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 12866
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1097,
+      "step": 12867
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 12868
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 12869
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 12870
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 12871
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1124,
+      "step": 12872
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 12873
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 12874
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 12875
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0672,
+      "step": 12876
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 12877
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 12878
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 12879
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 12880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 12881
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 12882
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 12883
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 12884
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 12885
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 12886
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1123,
+      "step": 12887
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 12888
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 12889
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 12890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 12891
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 12892
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 12893
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 12894
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 12895
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 12896
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.1156,
+      "step": 12897
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 12898
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 12899
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 12900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 12901
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 12902
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 12903
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.114,
+      "step": 12904
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 12905
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 12906
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 12907
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 12908
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 12909
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 12910
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 12911
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 12912
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 12913
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 12914
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0963,
+      "step": 12915
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 12916
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 12917
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0863,
+      "step": 12918
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 12919
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 12920
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 12921
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 12922
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.1143,
+      "step": 12923
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 12924
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 12925
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 12926
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 12927
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 12928
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0747,
+      "step": 12929
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 12930
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 12931
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 12932
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 12933
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 12934
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 12935
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 12936
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 12937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 12938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 12939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 12940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 12941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 12942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 12943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 12944
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 12945
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 12946
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 12947
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 12948
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 12949
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 12950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0993,
+      "step": 12951
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 12952
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 12953
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 12954
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 12955
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 12956
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 12957
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 12958
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1008,
+      "step": 12959
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 12960
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 12961
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 12962
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 12963
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 12964
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 12965
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 12966
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 12967
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 12968
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 12969
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 12970
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 12971
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 12972
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 12973
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 12974
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 12975
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 12976
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 12977
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 12978
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 12979
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 12980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 12981
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 12982
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 12983
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 12984
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0887,
+      "step": 12985
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 12986
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 12987
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 12988
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 12989
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 12990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 12991
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 12992
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 12993
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 12994
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 12995
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 12996
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 12997
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 12998
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 12999
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 13000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 13001
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 13002
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 13003
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0842,
+      "step": 13004
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 13005
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 13006
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 13007
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0809,
+      "step": 13008
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 13009
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 13010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 13011
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 13012
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 13013
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 13014
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 13015
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 13016
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 13017
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 13018
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 13019
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 13020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 13021
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 13022
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 13023
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 13024
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 13025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 13026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1053,
+      "step": 13027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 13028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 13029
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 13030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 13031
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 13032
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 13033
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 13034
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 13035
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 13036
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 13037
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 13038
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.1132,
+      "step": 13039
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 13040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 13041
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 13042
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 13043
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 13044
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 13045
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 13046
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 13047
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 13048
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 13049
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 13050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 13051
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 13052
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 13053
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 13054
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 13055
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 13056
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 13057
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 13058
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 13059
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0824,
+      "step": 13060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 13061
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1364,
+      "step": 13062
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 13063
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 13064
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 13065
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1085,
+      "step": 13066
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 13067
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 13068
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 13069
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 13070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 13071
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 13072
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 13073
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 13074
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 13075
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 13076
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.1016,
+      "step": 13077
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 13078
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 13079
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 13080
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 13081
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 13082
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0182,
+      "step": 13083
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 13084
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 13085
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 13086
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 13087
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 13088
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 13089
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 13090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.1237,
+      "step": 13091
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 13092
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 13093
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 13094
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0793,
+      "step": 13095
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 13096
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.078,
+      "step": 13097
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 13098
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 13099
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 13100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 13101
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.089,
+      "step": 13102
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 13103
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 13104
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 13105
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 13106
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 13107
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 13108
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 13109
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0935,
+      "step": 13110
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 13111
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 13112
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 13113
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 13114
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 13115
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 13116
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 13117
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 13118
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13119
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 13120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0765,
+      "step": 13121
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 13122
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 13123
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 13124
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 13125
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0753,
+      "step": 13126
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 13127
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 13128
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0708,
+      "step": 13129
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 13130
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 13131
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 13132
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 13133
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 13134
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 13135
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 13136
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0862,
+      "step": 13137
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 13138
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 13139
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0853,
+      "step": 13140
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 13141
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 13142
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 13143
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 13144
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 13145
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 13146
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1068,
+      "step": 13147
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 13148
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 13149
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 13150
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 13151
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 13152
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 13153
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 13154
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 13155
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 13156
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 13157
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1056,
+      "step": 13158
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1288,
+      "step": 13159
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 13160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0831,
+      "step": 13161
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 13162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 13163
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 13164
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 13165
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 13166
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 13167
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 13168
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 13169
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 13170
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 13171
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 13172
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 13173
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 13174
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1151,
+      "step": 13175
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.1485,
+      "step": 13176
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 13177
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 13178
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 13179
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 13180
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 13181
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 13182
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 13183
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 13184
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 13185
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 13186
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 13187
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0797,
+      "step": 13188
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 13189
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 13190
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 13191
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 13192
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 13193
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 13194
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 13195
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 13196
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0742,
+      "step": 13197
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 13198
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 13199
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 13200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.079,
+      "step": 13201
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 13202
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 13203
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 13204
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0723,
+      "step": 13205
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 13206
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 13207
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 13208
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 13209
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 13210
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 13211
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 13212
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 13213
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 13214
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 13215
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 13216
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 13217
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 13218
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 13219
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 13220
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 13221
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 13222
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0933,
+      "step": 13223
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 13224
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 13225
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 13226
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0739,
+      "step": 13227
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 13228
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 13229
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0965,
+      "step": 13230
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2e-05,
+      "loss": 0.097,
+      "step": 13231
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 13232
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 13233
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 13234
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 13235
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 13236
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 13237
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1049,
+      "step": 13238
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 13239
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 13240
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 13241
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 13242
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 13243
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 13244
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 13245
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 13246
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 13247
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 13248
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 13249
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1142,
+      "step": 13250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 13251
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1082,
+      "step": 13252
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 13253
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 13254
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0717,
+      "step": 13255
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 13256
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 13257
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 13258
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 13259
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 13260
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 13261
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 13262
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 13263
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 13264
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 13265
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.1052,
+      "step": 13266
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 13267
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 13268
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 13269
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 13270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 13271
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 13272
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0841,
+      "step": 13273
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 13274
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 13275
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 13276
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 13277
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 13278
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 13279
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0845,
+      "step": 13280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 13281
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 13282
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 13283
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 13284
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 13285
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 13286
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 13287
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 13288
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 13289
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 13290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 13291
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 13292
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 13293
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 13294
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0726,
+      "step": 13295
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 13296
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 13297
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 13298
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 13299
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 13300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 13301
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0957,
+      "step": 13302
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 13303
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 13304
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 13305
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 13306
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 13307
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 13308
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.1109,
+      "step": 13309
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 13310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 13311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 13312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 13313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 13314
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 13315
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 13316
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 13317
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0799,
+      "step": 13318
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 13319
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 13320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0867,
+      "step": 13321
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 13322
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 13323
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 13324
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 13325
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 13326
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 13327
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 13328
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 13329
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 13330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 13331
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 13332
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 13333
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 13334
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 13335
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.063,
+      "step": 13336
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 13337
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 13338
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 13339
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0829,
+      "step": 13340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 13341
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 13342
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 13343
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 13344
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 13345
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 13346
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 13347
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 13348
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 13349
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 13350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 13351
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 13352
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 13353
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 13354
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 13355
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 13356
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 13357
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 13358
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 13359
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 13360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 13361
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.1158,
+      "step": 13362
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 13363
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 13364
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 13365
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 13366
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 13367
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 13368
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 13369
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 13370
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 13371
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 13372
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 13373
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2e-05,
+      "loss": 0.087,
+      "step": 13374
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 13375
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 13376
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 13377
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 13378
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 13379
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 13380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 13381
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 13382
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 13383
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0931,
+      "step": 13384
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 13385
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 13386
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 13387
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 13388
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 13389
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 13390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 13391
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 13392
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 13393
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 13394
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 13395
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 13396
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 13397
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 13398
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 13399
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 13400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 13401
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 13402
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 13403
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.1154,
+      "step": 13404
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 13405
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 13406
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0729,
+      "step": 13407
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 13408
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 13409
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 13410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 13411
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 13412
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0796,
+      "step": 13413
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 13414
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 13415
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 13416
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 13417
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 13418
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 13419
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 13420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 13421
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 13422
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1073,
+      "step": 13423
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 13424
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 13425
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 13426
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 13427
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 13428
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 13429
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0762,
+      "step": 13430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 13431
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 13432
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 13433
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 13434
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 13435
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.088,
+      "step": 13436
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.1348,
+      "step": 13437
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 13438
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 13439
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 13440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 13441
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 13442
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 13443
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 13444
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0922,
+      "step": 13445
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0878,
+      "step": 13446
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 13447
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0871,
+      "step": 13448
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 13449
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 13450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 13451
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 13452
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 13453
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 13454
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 13455
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13456
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 13457
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 13458
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 13459
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 13460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 13461
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 13462
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.081,
+      "step": 13463
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0169,
+      "step": 13464
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 13465
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 13466
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 13467
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 13468
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 13469
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0751,
+      "step": 13470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 13471
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 13472
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 13473
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 13474
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 13475
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 13476
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 13477
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 13478
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 13479
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 13480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 13481
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0949,
+      "step": 13482
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 13483
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 13484
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.1017,
+      "step": 13485
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 13486
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13487
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 13488
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 13489
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 13490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 13491
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 13492
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 13493
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 13494
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 13495
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 13496
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 13497
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 13498
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 13499
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 13500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 13501
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 13502
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 13503
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 13504
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 13505
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 13506
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 13507
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 13508
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 13509
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0776,
+      "step": 13510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 13511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 13512
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.072,
+      "step": 13513
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 13514
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 13515
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2e-05,
+      "loss": 0.084,
+      "step": 13516
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 13517
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 13518
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 13519
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 13520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 13521
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 13522
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 13523
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 13524
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 13525
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 13526
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 13527
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 13528
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 13529
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 13530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0864,
+      "step": 13531
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 13532
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 13533
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 13534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 13535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 13536
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 13537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 13538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 13539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0683,
+      "step": 13540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 13541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 13542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1018,
+      "step": 13543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 13544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 13545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 13546
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 13547
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 13548
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0646,
+      "step": 13549
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0954,
+      "step": 13550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 13551
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.083,
+      "step": 13552
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 13553
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 13554
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0899,
+      "step": 13555
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 13556
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0673,
+      "step": 13557
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 13558
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0768,
+      "step": 13559
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 13560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.1015,
+      "step": 13561
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 13562
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 13563
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0817,
+      "step": 13564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 13565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 13566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 13567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 13568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 13569
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 13570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 13571
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 13572
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 13573
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 13574
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0725,
+      "step": 13575
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 13576
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 13577
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 13578
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 13579
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 13580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 13581
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 13582
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 13583
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 13584
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 13585
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 13586
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0661,
+      "step": 13587
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 13588
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 13589
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 13590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0981,
+      "step": 13591
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 13592
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 13593
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 13594
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 13595
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1026,
+      "step": 13596
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 13597
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 13598
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 13599
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 13600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0764,
+      "step": 13601
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.103,
+      "step": 13602
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0972,
+      "step": 13603
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 13604
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.096,
+      "step": 13605
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 13606
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.1039,
+      "step": 13607
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 13608
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 13609
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 13610
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 13611
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 13612
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 13613
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 13614
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 13615
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 13616
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 13617
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 13618
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 13619
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 13620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 13621
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 13622
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13623
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0757,
+      "step": 13624
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 13625
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 13626
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 13627
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 13628
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 13629
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 13630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 13631
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 13632
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 13633
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 13634
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 13635
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 13636
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 13637
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 13638
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0782,
+      "step": 13639
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 13640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 13641
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0743,
+      "step": 13642
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 13643
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 13644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 13645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 13646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 13647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 13648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 13649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 13650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 13651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 13652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0823,
+      "step": 13653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 13654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0837,
+      "step": 13655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 13656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 13657
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 13658
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 13659
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 13660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 13661
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 13662
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 13663
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 13664
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0671,
+      "step": 13665
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 13666
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 13667
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 13668
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 13669
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 13670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 13671
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0884,
+      "step": 13672
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 13673
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 13674
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.07,
+      "step": 13675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 13676
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 13677
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 13678
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 13679
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 13680
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 13681
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 13682
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 13683
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0946,
+      "step": 13684
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 13685
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 13686
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 13687
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 13688
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 13689
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 13690
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 13691
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 13692
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 13693
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 13694
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1318,
+      "step": 13695
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 13696
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0728,
+      "step": 13697
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 13698
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0866,
+      "step": 13699
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 13700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 13701
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 13702
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 13703
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 13704
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 13705
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 13706
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 13707
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 13708
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.091,
+      "step": 13709
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 13710
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 13711
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 13712
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 13713
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 13714
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 13715
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0781,
+      "step": 13716
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 13717
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 13718
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 13719
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 13720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 13721
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 13722
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 13723
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0875,
+      "step": 13724
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 13725
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 13726
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 13727
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1145,
+      "step": 13728
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 13729
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 13730
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0814,
+      "step": 13731
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 13732
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 13733
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 13734
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 13735
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 13736
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0735,
+      "step": 13737
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.1027,
+      "step": 13738
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 13739
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 13740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 13741
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 13742
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 13743
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 13744
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 13745
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0749,
+      "step": 13746
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 13747
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 13748
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 13749
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 13750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 13751
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 13752
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 13753
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 13754
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 13755
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 13756
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 13757
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 13758
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1076,
+      "step": 13759
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 13760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 13761
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 13762
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 13763
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 13764
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 13765
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 13766
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 13767
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 13768
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0822,
+      "step": 13769
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 13770
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 13771
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 13772
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0855,
+      "step": 13773
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 13774
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 13775
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 13776
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 13777
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 13778
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 13779
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 13780
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 13781
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.095,
+      "step": 13782
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 13783
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 13784
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 13785
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 13786
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 13787
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 13788
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 13789
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 13790
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 13791
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0678,
+      "step": 13792
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 13793
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0792,
+      "step": 13794
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 13795
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 13796
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 13797
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 13798
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.1554,
+      "step": 13799
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 13800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0915,
+      "step": 13801
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 13802
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 13803
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 13804
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 13805
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 13806
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 13807
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 13808
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 13809
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 13810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 13811
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 13812
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 13813
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 13814
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 13815
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0898,
+      "step": 13816
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 13817
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 13818
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 13819
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 13820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.1044,
+      "step": 13821
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0961,
+      "step": 13822
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0851,
+      "step": 13823
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 13824
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0856,
+      "step": 13825
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 13826
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0852,
+      "step": 13827
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 13828
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 13829
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 13830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 13831
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 13832
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 13833
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 13834
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 13835
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 13836
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0816,
+      "step": 13837
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 13838
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0718,
+      "step": 13839
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0685,
+      "step": 13840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 13841
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 13842
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0639,
+      "step": 13843
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0754,
+      "step": 13844
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0834,
+      "step": 13845
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 13846
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 13847
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 13848
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 13849
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 13850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13851
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 13852
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 13853
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 13854
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 13855
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 13856
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 13857
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 13858
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 13859
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1116,
+      "step": 13860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0663,
+      "step": 13861
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 13862
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 13863
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 13864
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 13865
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 13866
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 13867
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 13868
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 13869
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1086,
+      "step": 13870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 13871
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0858,
+      "step": 13872
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 13873
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1239,
+      "step": 13874
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 13875
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1074,
+      "step": 13876
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 13877
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 13878
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.1009,
+      "step": 13879
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 13880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 13881
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0651,
+      "step": 13882
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 13883
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 13884
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 13885
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 13886
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0748,
+      "step": 13887
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 13888
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0865,
+      "step": 13889
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0771,
+      "step": 13890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0689,
+      "step": 13891
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 13892
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 13893
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 13894
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 13895
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 13896
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 13897
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 13898
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0808,
+      "step": 13899
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 13900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 13901
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 13902
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 13903
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 13904
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 13905
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 13906
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 13907
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 13908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 13909
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0945,
+      "step": 13910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 13911
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 13912
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0811,
+      "step": 13913
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 13914
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 13915
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 13916
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0982,
+      "step": 13917
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 13918
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 13919
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 13920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0608,
+      "step": 13921
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 13922
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 13923
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0668,
+      "step": 13924
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 13925
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0986,
+      "step": 13926
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 13927
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 13928
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 13929
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 13930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0953,
+      "step": 13931
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0786,
+      "step": 13932
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 13933
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1247,
+      "step": 13934
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 13935
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0624,
+      "step": 13936
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 13937
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.1007,
+      "step": 13938
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0916,
+      "step": 13939
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 13940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 13941
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 13942
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0873,
+      "step": 13943
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 13944
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 13945
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0722,
+      "step": 13946
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 13947
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 13948
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 13949
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0912,
+      "step": 13950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 13951
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 13952
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 13953
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0941,
+      "step": 13954
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 13955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 13956
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 13957
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 13958
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 13959
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 13960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 13961
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 13962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 13963
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 13964
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 13965
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 13966
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 13967
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0763,
+      "step": 13968
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 13969
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 13970
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 13971
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 13972
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0902,
+      "step": 13973
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 13974
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 13975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 13976
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 13977
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0905,
+      "step": 13978
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0924,
+      "step": 13979
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 13980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 13981
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0909,
+      "step": 13982
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 13983
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 13984
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 13985
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 13986
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 13987
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 13988
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 13989
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 13990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 13991
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 13992
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 13993
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 13994
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 13995
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 13996
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 13997
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1048,
+      "step": 13998
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 13999
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0872,
+      "step": 14000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0698,
+      "step": 14001
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 14002
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 14003
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 14004
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 14005
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1405,
+      "step": 14006
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14007
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 14008
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 14009
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 14010
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 14011
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0784,
+      "step": 14012
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0944,
+      "step": 14013
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 14014
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 14015
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 14016
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 14017
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 14018
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 14019
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0795,
+      "step": 14020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 14021
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 14022
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 14023
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0839,
+      "step": 14024
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0819,
+      "step": 14025
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 14026
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 14027
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 14028
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 14029
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 14030
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0956,
+      "step": 14031
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 14032
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 14033
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 14034
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 14035
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 14036
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.1042,
+      "step": 14037
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0827,
+      "step": 14038
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 14039
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 14040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 14041
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.082,
+      "step": 14042
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14043
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 14044
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 14045
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 14046
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0994,
+      "step": 14047
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0885,
+      "step": 14048
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0715,
+      "step": 14049
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 14050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1025,
+      "step": 14051
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0836,
+      "step": 14052
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 14053
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0868,
+      "step": 14054
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0968,
+      "step": 14055
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 14056
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 14057
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 14058
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 14059
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 14060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 14061
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 14062
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0914,
+      "step": 14063
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 14064
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 14065
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0773,
+      "step": 14066
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1292,
+      "step": 14067
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 14068
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0859,
+      "step": 14069
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0619,
+      "step": 14070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0846,
+      "step": 14071
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 14072
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.1135,
+      "step": 14073
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0772,
+      "step": 14074
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0844,
+      "step": 14075
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 14076
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 14077
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14078
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 14079
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 14080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 14081
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 14082
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0706,
+      "step": 14083
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 14084
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.092,
+      "step": 14085
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0861,
+      "step": 14086
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 14087
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 14088
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0704,
+      "step": 14089
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0952,
+      "step": 14090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 14091
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 14092
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0788,
+      "step": 14093
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0901,
+      "step": 14094
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 14095
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 14096
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 14097
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 14098
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 14099
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.074,
+      "step": 14100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 14101
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0732,
+      "step": 14102
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0696,
+      "step": 14103
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 14104
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 14105
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0802,
+      "step": 14106
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 14107
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 14108
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0766,
+      "step": 14109
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 14110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 14111
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 14112
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0667,
+      "step": 14113
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14114
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 14115
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0738,
+      "step": 14116
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 14117
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 14118
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.104,
+      "step": 14119
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0947,
+      "step": 14120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 14121
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 14122
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 14123
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 14124
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 14125
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 14126
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 14127
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.1234,
+      "step": 14128
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 14129
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 14130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 14131
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 14132
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 14133
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 14134
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0632,
+      "step": 14135
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 14136
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 14137
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14138
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0832,
+      "step": 14139
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.086,
+      "step": 14140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 14141
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0903,
+      "step": 14142
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0959,
+      "step": 14143
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 14144
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 14145
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 14146
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 14147
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 14148
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 14149
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 14150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 14151
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 14152
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 14153
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 14154
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 14155
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 14156
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0821,
+      "step": 14157
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0826,
+      "step": 14158
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0779,
+      "step": 14159
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0636,
+      "step": 14160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 14161
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 14162
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 14163
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.093,
+      "step": 14164
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 14165
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0874,
+      "step": 14166
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 14167
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0921,
+      "step": 14168
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 14169
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 14170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.105,
+      "step": 14171
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 14172
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 14173
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1099,
+      "step": 14174
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 14175
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 14176
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 14177
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 14178
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1193,
+      "step": 14179
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.1031,
+      "step": 14180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0778,
+      "step": 14181
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 14182
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0761,
+      "step": 14183
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0838,
+      "step": 14184
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0769,
+      "step": 14185
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 14186
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0737,
+      "step": 14187
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0707,
+      "step": 14188
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0687,
+      "step": 14189
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 14190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 14191
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 14192
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 14193
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 14194
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 14195
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 14196
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 14197
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 14198
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0833,
+      "step": 14199
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 14200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 14201
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.069,
+      "step": 14202
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 14203
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 14204
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 14205
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 14206
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0724,
+      "step": 14207
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0835,
+      "step": 14208
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0962,
+      "step": 14209
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 14210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 14211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1045,
+      "step": 14212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 14213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0854,
+      "step": 14214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 14215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0815,
+      "step": 14216
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 14217
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 14218
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 14219
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 14220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 14221
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 14222
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 14223
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0711,
+      "step": 14224
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0658,
+      "step": 14225
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14226
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 14227
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1448,
+      "step": 14228
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2e-05,
+      "loss": 0.1032,
+      "step": 14229
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 14230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0727,
+      "step": 14231
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 14232
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0684,
+      "step": 14233
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1115,
+      "step": 14234
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0741,
+      "step": 14235
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.071,
+      "step": 14236
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 14237
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 14238
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0731,
+      "step": 14239
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0812,
+      "step": 14240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 14241
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0694,
+      "step": 14242
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.1094,
+      "step": 14243
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0805,
+      "step": 14244
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0626,
+      "step": 14245
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 14246
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 14247
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0774,
+      "step": 14248
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0892,
+      "step": 14249
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 14250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 14251
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 14252
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0693,
+      "step": 14253
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0203,
+      "step": 14254
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 14255
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 14256
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 14257
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 14258
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 14259
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 14260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 14261
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 14262
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 14263
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 14264
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 14265
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 14266
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 14267
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 14268
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 14269
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 14270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 14271
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 14272
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 14273
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 14274
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 14275
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 14276
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 14277
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 14278
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0721,
+      "step": 14279
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 14280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 14281
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 14282
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 14283
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 14284
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 14285
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 14286
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14287
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 14288
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 14289
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14290
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 14291
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 14292
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 14293
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 14294
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 14295
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 14296
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0094,
+      "step": 14297
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14298
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 14299
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0152,
+      "step": 14301
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 14302
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 14303
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 14304
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 14305
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 14306
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 14307
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 14308
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 14309
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14310
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 14311
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 14312
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 14313
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14314
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 14315
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14316
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14317
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0123,
+      "step": 14318
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14319
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 14320
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 14321
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14322
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 14323
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 14324
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 14325
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 14326
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 14327
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 14328
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 14329
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0163,
+      "step": 14331
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 14332
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 14333
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 14334
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 14335
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14336
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 14337
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14338
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 14339
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 14340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14341
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 14342
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0145,
+      "step": 14343
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 14344
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 14345
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 14346
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14347
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 14348
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14349
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 14350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 14351
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 14352
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 14353
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 14354
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 14355
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 14356
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14357
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14358
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0149,
+      "step": 14359
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 14360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 14361
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 14362
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 14363
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14364
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 14365
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 14366
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14367
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 14368
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 14369
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 14370
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 14371
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 14372
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 14373
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 14374
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14375
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 14376
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14377
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 14378
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 14379
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 14380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 14381
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 14382
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 14383
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 14384
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0129,
+      "step": 14385
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 14386
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 14387
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 14388
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 14389
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 14390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 14391
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 14392
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 14393
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 14394
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 14395
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 14396
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 14397
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 14398
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 14399
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 14400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 14401
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 14402
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14403
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 14404
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 14405
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 14406
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 14407
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 14408
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 14409
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 14410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14411
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0653,
+      "step": 14412
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 14413
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 14414
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 14415
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14416
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 14417
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 14418
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 14419
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 14420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 14421
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 14422
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 14423
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 14424
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 14425
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 14426
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14427
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 14428
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 14429
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 14430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 14431
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 14432
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 14433
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 14434
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 14435
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 14436
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14437
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 14438
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 14439
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 14440
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 14441
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 14442
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 14443
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 14444
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 14445
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 14446
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 14447
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 14448
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 14449
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 14450
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 14451
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 14452
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14453
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 14454
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 14455
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 14456
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 14457
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 14458
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 14459
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 14460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14461
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.015,
+      "step": 14462
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 14463
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 14464
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 14465
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 14466
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0142,
+      "step": 14467
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 14468
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 14469
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 14470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 14471
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 14472
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 14473
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14474
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 14475
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 14476
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14477
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 14478
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 14479
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 14480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14481
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 14482
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 14483
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14484
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 14485
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 14486
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 14487
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 14488
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 14489
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0193,
+      "step": 14490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 14491
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 14492
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.018,
+      "step": 14493
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 14494
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 14495
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 14496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 14497
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 14498
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 14499
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 14500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 14501
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0108,
+      "step": 14502
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 14503
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14504
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14505
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 14506
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 14507
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 14508
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 14509
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 14510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 14511
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 14512
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 14513
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 14514
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0132,
+      "step": 14515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0166,
+      "step": 14516
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14517
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0147,
+      "step": 14518
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 14519
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 14521
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 14522
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 14523
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 14524
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0176,
+      "step": 14525
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 14526
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 14527
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 14528
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14529
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 14530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 14531
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 14532
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 14533
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 14534
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 14535
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 14536
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 14537
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 14538
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 14539
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0162,
+      "step": 14540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 14541
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 14542
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 14543
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 14544
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 14545
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 14546
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 14547
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 14548
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 14549
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 14551
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 14552
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 14553
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 14554
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 14555
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 14556
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 14557
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 14558
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 14559
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 14560
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 14561
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 14562
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 14563
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14564
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14565
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 14566
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 14567
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 14568
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 14569
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0179,
+      "step": 14570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 14571
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 14572
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 14573
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0151,
+      "step": 14574
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 14575
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 14576
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 14577
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14578
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 14579
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 14581
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 14582
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 14583
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14584
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 14585
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 14586
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 14587
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 14588
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 14589
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 14590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 14591
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 14592
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 14593
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 14594
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14595
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14596
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 14597
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 14598
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 14599
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0604,
+      "step": 14600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 14601
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14602
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 14603
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 14604
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14605
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14606
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14607
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 14608
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2e-05,
+      "loss": 0.0138,
+      "step": 14609
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 14610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 14611
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 14612
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 14613
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 14614
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 14615
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 14616
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14617
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 14618
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 14619
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0117,
+      "step": 14620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0129,
+      "step": 14621
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 14622
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 14623
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 14624
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14625
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 14626
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 14627
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 14628
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 14629
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 14630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 14631
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 14632
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 14633
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 14634
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 14635
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 14636
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 14637
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0135,
+      "step": 14638
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14639
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 14641
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 14642
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 14643
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 14644
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 14645
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 14646
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 14647
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 14648
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 14649
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 14650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 14651
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 14652
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 14653
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 14654
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 14655
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 14656
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 14657
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 14658
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0159,
+      "step": 14659
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 14660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 14661
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 14662
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 14663
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 14664
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 14665
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 14666
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 14667
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14668
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 14669
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 14670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 14671
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 14672
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 14673
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 14674
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 14675
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 14676
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14677
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 14678
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 14679
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 14681
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 14682
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14683
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 14684
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 14685
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 14686
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 14687
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 14688
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 14689
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 14690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 14691
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 14692
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 14693
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 14694
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 14695
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 14696
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 14697
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 14698
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 14699
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 14700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 14701
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 14702
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 14703
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14704
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 14705
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14706
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 14707
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 14708
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 14709
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 14710
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 14711
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 14712
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 14713
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 14714
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14715
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 14716
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 14717
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 14718
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 14719
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 14720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 14721
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 14722
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 14723
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 14724
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14725
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 14726
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0151,
+      "step": 14727
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 14728
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 14729
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 14730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 14731
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 14732
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14733
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14734
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 14735
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 14736
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 14737
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 14738
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 14739
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 14740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 14741
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 14742
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 14743
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 14744
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 14745
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 14746
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0159,
+      "step": 14747
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.017,
+      "step": 14748
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 14749
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 14750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2e-05,
+      "loss": 0.0152,
+      "step": 14751
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 14752
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 14753
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 14754
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 14755
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 14756
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 14757
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 14758
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 14759
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 14760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 14761
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 14762
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 14763
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 14764
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 14765
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14766
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 14767
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 14768
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14769
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 14770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 14771
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 14772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14773
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 14774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 14775
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 14776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 14777
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 14778
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 14779
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 14780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0177,
+      "step": 14781
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 14782
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 14783
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 14784
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 14785
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 14786
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 14787
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0582,
+      "step": 14788
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 14789
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 14791
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 14792
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 14793
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 14794
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 14795
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 14796
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14797
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 14798
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 14799
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 14800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 14801
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 14802
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 14803
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 14804
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0137,
+      "step": 14805
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 14806
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 14807
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 14808
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 14809
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0159,
+      "step": 14810
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 14811
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 14812
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 14813
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 14814
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 14815
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 14816
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0146,
+      "step": 14817
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 14818
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 14819
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 14820
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 14821
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0184,
+      "step": 14822
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0233,
+      "step": 14823
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 14824
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 14825
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 14826
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 14827
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.013,
+      "step": 14828
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14829
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 14830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 14831
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 14832
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 14833
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 14834
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 14835
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 14836
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 14837
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 14838
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 14839
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 14840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 14841
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 14842
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 14843
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 14844
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 14845
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 14846
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 14847
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 14848
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 14849
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 14850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 14851
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0156,
+      "step": 14852
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14853
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 14854
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 14855
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 14856
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 14857
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14858
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 14859
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 14860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 14861
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 14862
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 14863
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14864
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 14865
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 14866
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 14867
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 14868
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 14869
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 14870
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 14871
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 14872
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 14873
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0161,
+      "step": 14874
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 14875
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 14876
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 14877
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14878
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 14879
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 14880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 14881
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 14882
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 14883
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 14884
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 14885
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 14886
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0154,
+      "step": 14887
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 14888
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0702,
+      "step": 14889
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 14890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 14891
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 14892
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 14893
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14894
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14895
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 14896
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 14897
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 14898
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 14899
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 14900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 14901
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 14902
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 14903
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 14904
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 14905
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 14906
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 14907
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 14908
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 14909
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 14910
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 14911
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 14912
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 14913
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 14914
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 14915
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 14916
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 14917
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 14918
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 14919
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 14920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 14921
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 14922
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 14923
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 14924
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 14925
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 14926
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 14927
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 14928
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 14929
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 14930
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 14931
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 14932
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 14933
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 14934
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 14935
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 14936
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 14937
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 14938
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14939
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 14940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 14941
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 14942
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 14943
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14944
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 14945
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 14946
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 14947
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0182,
+      "step": 14948
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14949
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 14950
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 14951
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 14952
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 14953
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 14954
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 14955
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 14956
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 14957
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0168,
+      "step": 14958
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 14959
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 14960
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 14961
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 14962
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 14963
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 14964
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 14965
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 14966
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 14967
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 14968
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 14969
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 14970
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 14971
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 14972
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 14973
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 14974
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 14975
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 14976
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 14977
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 14978
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 14979
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 14980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 14981
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 14982
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 14983
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 14984
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 14985
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 14986
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 14987
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 14988
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 14989
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 14990
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 14991
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 14992
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 14993
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 14994
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 14995
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 14996
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 14997
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 14998
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 14999
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 15000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 15001
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 15002
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 15003
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15004
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15005
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 15006
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 15007
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15008
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 15009
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 15010
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 15011
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 15012
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15013
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 15014
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15015
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 15016
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 15017
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 15018
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 15019
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 15020
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 15021
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 15022
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 15023
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 15024
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15025
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 15026
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 15027
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 15028
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 15029
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 15030
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0155,
+      "step": 15031
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 15032
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 15033
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 15034
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 15035
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2e-05,
+      "loss": 0.019,
+      "step": 15036
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0119,
+      "step": 15037
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15038
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15039
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 15040
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15041
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 15042
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 15043
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 15044
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15045
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 15046
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0169,
+      "step": 15047
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 15048
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 15049
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 15050
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 15051
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 15052
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15053
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 15054
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 15055
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 15056
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 15057
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15058
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 15059
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0188,
+      "step": 15060
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 15061
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 15062
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 15063
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 15064
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 15065
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 15066
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 15067
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 15068
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 15069
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15070
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 15071
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 15072
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 15073
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15074
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 15075
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 15076
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 15077
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 15078
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 15079
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 15080
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 15081
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0193,
+      "step": 15082
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 15083
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 15084
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 15085
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 15086
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 15087
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 15088
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0188,
+      "step": 15089
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 15090
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 15091
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 15092
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 15093
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 15094
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 15095
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 15096
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0174,
+      "step": 15097
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 15098
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 15099
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 15100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 15101
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 15102
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 15103
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0138,
+      "step": 15104
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 15105
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 15106
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 15107
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 15108
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15109
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 15110
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 15111
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 15112
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 15113
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 15114
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 15115
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 15116
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 15117
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 15118
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 15119
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 15120
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 15121
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0127,
+      "step": 15122
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 15123
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 15124
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 15125
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 15126
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 15127
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 15128
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 15129
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15130
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 15131
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 15132
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.009,
+      "step": 15133
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 15134
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 15135
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 15136
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 15137
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 15138
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15139
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 15140
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 15141
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 15142
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0154,
+      "step": 15143
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 15144
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 15145
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 15146
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 15147
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 15148
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 15149
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 15150
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 15151
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15152
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15153
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15154
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 15155
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 15156
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15157
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 15158
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 15159
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 15160
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 15161
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0151,
+      "step": 15162
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0177,
+      "step": 15163
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 15164
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 15165
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15166
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 15167
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 15168
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 15169
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 15170
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 15171
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 15172
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 15173
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 15174
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 15175
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 15176
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15177
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 15178
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 15179
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 15180
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 15181
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 15182
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 15183
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 15184
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 15185
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15186
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 15187
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 15188
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 15189
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15190
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 15191
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 15192
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 15193
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 15194
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 15195
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 15196
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15197
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15198
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 15199
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15200
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 15201
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0179,
+      "step": 15202
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 15203
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 15204
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0233,
+      "step": 15205
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 15206
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15207
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 15208
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 15209
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 15210
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 15211
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 15212
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 15213
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15214
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15215
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 15216
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 15217
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 15218
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 15219
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 15220
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 15221
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 15222
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 15223
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 15224
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 15225
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2e-05,
+      "loss": 0.0122,
+      "step": 15226
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 15227
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 15228
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 15229
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 15230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 15231
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 15232
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 15233
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 15234
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 15235
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 15236
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 15237
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 15238
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 15239
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0141,
+      "step": 15240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15241
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 15242
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0151,
+      "step": 15243
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 15244
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15245
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 15246
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 15247
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0109,
+      "step": 15248
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15249
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 15250
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 15251
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 15252
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0184,
+      "step": 15253
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 15254
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 15255
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 15256
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 15257
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0174,
+      "step": 15258
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 15259
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 15260
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 15261
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15262
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 15263
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 15264
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 15265
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 15266
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15267
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 15268
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 15269
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 15270
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15271
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 15272
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 15273
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 15274
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 15275
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 15276
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 15277
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 15278
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 15279
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 15280
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0136,
+      "step": 15281
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 15282
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 15283
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 15284
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 15285
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 15286
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 15287
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 15288
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 15289
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 15290
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 15291
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 15292
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 15293
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 15294
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15295
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0169,
+      "step": 15296
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 15297
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 15298
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 15299
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15300
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 15301
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 15302
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 15303
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 15304
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0188,
+      "step": 15305
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.015,
+      "step": 15306
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 15307
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 15308
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 15309
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 15310
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 15311
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 15312
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 15313
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 15314
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 15315
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 15316
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 15317
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 15318
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 15319
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 15320
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 15321
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 15322
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 15323
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 15324
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15325
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 15326
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15327
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 15328
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 15329
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 15330
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 15331
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 15332
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 15333
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 15334
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 15335
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 15336
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 15337
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 15338
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15339
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 15340
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 15341
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0111,
+      "step": 15342
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 15343
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 15344
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 15345
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 15346
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15347
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 15348
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 15349
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 15350
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15351
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 15352
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 15353
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 15354
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 15355
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 15356
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 15357
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 15358
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 15359
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 15360
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 15361
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 15362
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 15363
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 15364
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 15365
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 15366
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 15367
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 15368
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15369
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0171,
+      "step": 15370
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 15371
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15372
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 15373
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 15374
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.06,
+      "step": 15375
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15376
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 15377
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 15378
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0182,
+      "step": 15379
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 15380
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 15381
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 15382
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 15383
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 15384
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 15385
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 15386
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0114,
+      "step": 15387
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 15388
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 15389
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 15390
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15391
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 15392
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15393
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 15394
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0129,
+      "step": 15395
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 15396
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 15397
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 15398
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 15399
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 15400
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15401
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 15402
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15403
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 15404
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 15405
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0627,
+      "step": 15406
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 15407
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 15408
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 15409
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 15410
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.015,
+      "step": 15411
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 15412
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 15413
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 15414
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 15415
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15416
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 15417
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 15418
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 15419
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15420
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 15421
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 15422
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 15423
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 15424
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15425
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 15426
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 15427
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 15428
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 15429
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 15430
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0179,
+      "step": 15431
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 15432
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 15433
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 15434
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 15435
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 15436
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 15437
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15438
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15439
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 15440
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 15441
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15442
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 15443
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 15444
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 15445
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15446
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 15447
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 15448
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15449
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 15450
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 15451
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15452
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 15453
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 15454
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 15455
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 15456
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 15457
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 15458
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 15459
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 15460
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 15461
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 15462
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 15463
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 15464
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 15465
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0224,
+      "step": 15466
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 15467
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15468
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15469
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 15470
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 15471
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 15472
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 15473
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 15474
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 15475
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 15476
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 15477
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 15478
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15479
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 15481
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 15482
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 15483
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 15484
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 15485
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 15486
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 15487
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 15488
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 15489
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0654,
+      "step": 15490
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 15491
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 15492
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 15493
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 15494
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 15495
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 15496
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 15497
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 15498
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 15499
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15501
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15502
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 15503
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0163,
+      "step": 15504
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 15505
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 15506
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 15507
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 15508
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 15509
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 15510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 15511
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 15512
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 15513
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 15514
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.064,
+      "step": 15515
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 15516
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 15517
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 15518
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 15519
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 15520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 15521
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 15522
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 15523
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 15524
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 15525
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15526
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 15527
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15528
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 15529
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 15530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15531
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 15532
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 15533
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 15534
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 15535
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 15536
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 15537
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0155,
+      "step": 15538
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 15539
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 15540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15541
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 15542
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 15543
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 15544
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 15545
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 15546
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 15547
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 15548
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 15549
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 15550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 15551
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 15552
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 15553
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 15554
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 15555
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 15556
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 15557
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 15558
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 15559
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 15560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 15561
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 15562
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 15563
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15564
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 15565
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 15566
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 15567
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 15568
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15569
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 15570
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 15571
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 15572
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 15573
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 15574
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0168,
+      "step": 15575
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 15576
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 15577
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 15578
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 15579
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 15580
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0168,
+      "step": 15581
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 15582
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 15583
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 15584
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 15585
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 15586
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 15587
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 15588
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 15589
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 15590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15591
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 15592
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 15593
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 15594
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15595
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 15596
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0572,
+      "step": 15597
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 15598
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 15599
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 15600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 15601
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15602
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 15603
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 15604
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15605
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0177,
+      "step": 15606
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 15607
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 15608
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 15609
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 15610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15611
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15612
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 15613
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.017,
+      "step": 15614
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15615
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 15616
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 15617
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 15618
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 15619
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 15620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0585,
+      "step": 15621
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15622
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 15623
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 15624
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 15625
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 15626
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15627
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 15628
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 15629
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 15630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 15631
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 15632
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15633
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15634
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15635
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 15636
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 15637
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 15638
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 15639
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 15640
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 15641
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 15642
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0147,
+      "step": 15643
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 15644
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 15645
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 15646
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 15647
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 15648
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 15649
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 15650
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15651
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 15652
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 15653
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2e-05,
+      "loss": 0.0119,
+      "step": 15654
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 15655
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 15656
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 15657
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 15658
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 15659
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 15660
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 15661
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 15662
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 15663
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 15664
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 15665
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0154,
+      "step": 15666
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 15667
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15668
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 15669
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15670
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 15671
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 15672
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 15673
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 15674
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 15675
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 15676
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 15677
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 15678
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 15679
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 15680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15681
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 15682
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 15683
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 15684
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 15685
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 15686
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 15687
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 15688
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 15689
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15690
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15691
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 15692
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0573,
+      "step": 15693
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 15694
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15695
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 15696
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 15697
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 15698
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15699
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 15700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 15701
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2e-05,
+      "loss": 0.018,
+      "step": 15702
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 15703
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 15704
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 15705
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 15706
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 15707
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15708
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 15709
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 15710
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 15711
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 15712
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 15713
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 15714
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 15715
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15716
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 15717
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 15718
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 15719
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 15720
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 15721
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 15722
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 15723
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 15724
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 15725
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 15726
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 15727
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15728
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 15729
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0155,
+      "step": 15730
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 15731
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 15732
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 15733
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 15734
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 15735
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 15736
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 15737
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 15738
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 15739
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 15740
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 15741
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 15742
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 15743
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 15744
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 15745
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 15746
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15747
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15748
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 15749
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15750
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 15751
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 15752
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 15753
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 15754
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 15755
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 15756
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15757
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 15758
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 15759
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 15760
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15761
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 15762
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 15763
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 15764
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 15765
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0153,
+      "step": 15766
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 15767
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.061,
+      "step": 15768
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 15769
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 15770
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15771
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 15772
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 15773
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 15774
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 15775
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 15776
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15777
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 15778
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 15779
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15780
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15781
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 15782
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 15783
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 15784
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 15785
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 15786
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.011,
+      "step": 15787
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 15788
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 15789
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 15790
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 15791
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 15792
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 15793
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 15794
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 15795
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 15796
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 15797
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 15798
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 15799
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 15800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15801
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 15802
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 15803
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 15804
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 15805
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15806
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15807
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0155,
+      "step": 15808
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 15809
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 15810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 15811
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 15812
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 15813
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 15814
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 15815
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 15816
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15817
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 15818
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15819
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 15820
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 15821
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 15822
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 15823
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0138,
+      "step": 15824
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.018,
+      "step": 15825
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 15826
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 15827
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 15828
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 15829
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 15830
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 15831
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 15832
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 15833
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 15834
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15835
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 15836
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 15837
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15838
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 15839
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 15840
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 15841
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 15842
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 15843
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 15844
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 15845
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 15846
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 15847
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 15848
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 15849
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 15850
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 15851
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 15852
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 15853
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 15854
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 15855
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 15856
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 15857
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 15858
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 15859
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 15860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15861
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15862
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 15863
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 15864
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 15865
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 15866
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 15867
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 15868
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 15869
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 15870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 15871
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 15872
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 15873
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 15874
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 15875
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 15876
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15877
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 15878
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0131,
+      "step": 15879
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 15880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 15881
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 15882
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 15883
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 15884
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 15885
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 15886
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 15887
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 15888
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 15889
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 15890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 15891
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 15892
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 15893
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 15894
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 15895
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0153,
+      "step": 15896
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 15897
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 15898
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 15899
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 15900
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 15901
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 15902
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 15903
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 15904
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 15905
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 15906
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 15907
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 15908
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 15909
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 15910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 15911
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 15912
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 15913
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 15914
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 15915
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 15916
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 15917
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 15918
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 15919
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0714,
+      "step": 15920
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 15921
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 15922
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 15923
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 15924
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 15925
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 15926
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 15927
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 15928
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 15929
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 15930
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 15931
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 15932
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 15933
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 15934
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 15935
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 15936
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 15937
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 15938
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 15939
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 15940
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 15941
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 15942
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 15943
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 15944
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 15945
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 15946
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 15947
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 15948
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 15949
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 15950
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 15951
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 15952
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0166,
+      "step": 15953
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 15954
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 15955
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.065,
+      "step": 15956
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 15957
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 15958
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 15959
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 15960
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 15961
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0115,
+      "step": 15962
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 15963
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 15964
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 15965
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 15966
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0144,
+      "step": 15967
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 15968
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 15969
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 15970
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 15971
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 15972
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 15973
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 15974
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 15975
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 15976
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 15977
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0144,
+      "step": 15978
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 15979
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 15980
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 15981
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 15982
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 15983
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 15984
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0641,
+      "step": 15985
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 15986
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 15987
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 15988
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 15989
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 15990
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 15991
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15992
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0118,
+      "step": 15993
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 15994
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 15995
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 15996
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 15997
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 15998
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 15999
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 16001
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 16002
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16003
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 16004
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 16005
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0107,
+      "step": 16006
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 16007
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 16008
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 16009
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16010
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 16011
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 16012
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 16013
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0598,
+      "step": 16014
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0633,
+      "step": 16015
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 16016
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16017
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16018
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 16019
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 16020
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0124,
+      "step": 16021
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16022
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 16023
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 16024
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 16025
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 16026
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 16027
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 16028
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16029
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 16030
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 16031
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 16032
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 16033
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 16034
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 16035
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 16036
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 16037
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16038
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 16039
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 16040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 16041
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 16042
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 16043
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 16044
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.019,
+      "step": 16045
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 16046
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 16047
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 16048
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 16049
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0669,
+      "step": 16050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 16051
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 16052
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 16053
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16054
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 16055
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0719,
+      "step": 16056
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16057
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 16058
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 16059
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 16060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 16061
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 16062
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 16063
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 16064
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 16065
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 16066
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 16067
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 16068
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 16069
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0156,
+      "step": 16070
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 16071
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 16072
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 16073
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 16074
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16075
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 16076
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16077
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 16078
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0171,
+      "step": 16079
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 16080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 16081
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 16082
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 16083
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 16084
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 16085
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 16086
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 16087
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 16088
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16089
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 16090
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 16091
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 16092
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 16093
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 16094
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16095
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 16096
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 16097
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 16098
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0193,
+      "step": 16099
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 16100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 16101
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 16102
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 16103
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 16104
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 16105
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 16106
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16107
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 16108
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 16109
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 16110
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 16111
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16112
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 16113
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 16114
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 16115
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 16116
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 16117
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 16118
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 16119
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16120
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 16121
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0203,
+      "step": 16122
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 16123
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 16124
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16125
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 16126
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 16127
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 16128
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 16129
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 16130
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 16131
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 16132
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 16133
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0139,
+      "step": 16134
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 16135
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16136
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0168,
+      "step": 16137
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 16138
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 16139
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 16140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 16141
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 16142
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 16143
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16144
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 16145
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 16146
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 16147
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16148
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16149
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16150
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 16151
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16152
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 16153
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 16154
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16155
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 16156
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 16157
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16158
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 16159
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 16160
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 16161
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 16162
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 16163
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 16164
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16165
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 16166
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 16167
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 16168
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 16169
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 16170
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 16171
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16172
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 16173
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 16174
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 16175
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 16176
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 16177
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 16178
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0132,
+      "step": 16179
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 16180
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 16181
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 16182
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 16183
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16184
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 16185
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 16186
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 16187
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 16188
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 16189
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 16190
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 16191
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 16192
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 16193
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 16194
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 16195
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 16196
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0114,
+      "step": 16197
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 16198
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 16199
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16201
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 16202
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 16203
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 16204
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0645,
+      "step": 16205
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 16206
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 16207
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 16208
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 16209
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 16210
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16211
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 16212
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 16213
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 16214
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 16215
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 16216
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 16217
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16218
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 16219
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16220
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 16221
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 16222
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 16223
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 16224
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 16225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 16226
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16227
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 16228
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 16229
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 16230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 16231
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 16232
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 16233
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16234
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 16235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 16236
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 16237
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16238
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16239
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 16240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 16241
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16242
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 16243
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 16244
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 16245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 16246
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16247
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 16248
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 16249
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 16250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 16251
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 16252
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0233,
+      "step": 16253
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 16254
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0156,
+      "step": 16255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 16256
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 16257
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 16258
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 16259
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 16260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 16261
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 16262
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 16263
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16264
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 16265
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 16266
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 16267
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 16268
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 16269
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 16270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 16271
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 16272
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 16273
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 16274
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 16275
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0681,
+      "step": 16276
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 16277
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16278
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 16279
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 16280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 16281
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 16282
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 16283
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 16284
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 16285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 16286
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 16287
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 16288
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 16289
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 16291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 16292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 16293
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 16294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 16295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 16296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 16298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 16299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 16300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16301
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 16302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 16303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 16304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 16305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0106,
+      "step": 16306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 16307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 16308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 16309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 16310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 16311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 16312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 16313
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 16314
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 16315
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 16316
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16317
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 16318
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16319
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 16320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 16321
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16322
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 16323
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 16324
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 16326
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 16327
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 16328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0162,
+      "step": 16329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0153,
+      "step": 16331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0709,
+      "step": 16332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 16334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 16336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 16338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 16339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 16340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 16341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 16342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16344
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 16345
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 16346
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 16347
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0638,
+      "step": 16348
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16349
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 16350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 16351
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 16352
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 16353
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 16354
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 16355
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 16356
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 16357
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 16358
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 16359
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 16360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 16361
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 16362
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 16363
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 16364
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16365
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 16366
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 16367
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 16368
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16369
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16370
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 16371
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 16372
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16373
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 16374
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 16375
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 16376
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 16377
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 16378
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 16379
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 16380
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 16381
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 16382
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 16383
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 16384
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.067,
+      "step": 16385
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16386
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 16387
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 16388
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 16389
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 16390
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 16391
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 16392
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 16393
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16394
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16395
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 16396
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 16397
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 16398
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 16399
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 16400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0123,
+      "step": 16401
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 16402
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 16403
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 16404
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 16405
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 16406
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 16407
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 16408
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16409
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 16410
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 16411
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 16412
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 16413
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2e-05,
+      "loss": 0.0609,
+      "step": 16414
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0142,
+      "step": 16415
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 16416
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 16417
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 16418
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 16419
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 16420
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 16421
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 16422
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 16423
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 16424
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 16425
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 16426
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 16427
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 16428
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 16429
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16430
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 16431
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0659,
+      "step": 16432
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 16433
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 16434
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16435
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 16436
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0171,
+      "step": 16437
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 16438
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 16439
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 16440
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16441
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 16442
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.013,
+      "step": 16443
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 16444
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16445
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 16446
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 16447
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 16448
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 16449
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 16450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 16451
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 16452
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 16453
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16454
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 16455
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16456
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 16457
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 16458
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0182,
+      "step": 16459
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 16460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16461
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 16462
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16463
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 16464
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 16465
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 16466
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 16467
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 16468
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 16469
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 16470
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 16471
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 16472
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 16473
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 16474
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 16475
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 16476
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 16477
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 16478
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 16479
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 16480
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16481
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 16482
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 16483
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16484
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16485
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 16486
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 16487
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 16488
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 16489
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 16490
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16491
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16492
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 16493
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 16494
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 16495
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 16496
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 16497
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 16498
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 16499
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 16500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 16501
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16502
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16503
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16504
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 16505
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.019,
+      "step": 16506
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16507
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 16508
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2e-05,
+      "loss": 0.0143,
+      "step": 16509
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 16510
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 16511
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 16512
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 16513
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 16514
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 16515
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 16516
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16517
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 16518
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 16519
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 16520
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 16521
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16522
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 16523
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 16524
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0143,
+      "step": 16525
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 16526
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 16527
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 16528
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 16529
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 16530
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 16531
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16532
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 16533
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 16534
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 16535
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 16536
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 16537
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 16538
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 16539
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16540
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16541
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 16542
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16543
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 16544
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 16545
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16546
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 16547
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 16548
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 16549
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 16550
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 16551
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0692,
+      "step": 16552
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16553
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 16554
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 16555
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0193,
+      "step": 16556
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16557
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 16558
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 16559
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 16560
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 16561
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 16562
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 16563
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 16564
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 16565
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 16566
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 16567
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 16568
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 16569
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0117,
+      "step": 16570
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 16571
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0124,
+      "step": 16572
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 16573
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 16574
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 16575
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16576
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.016,
+      "step": 16577
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 16578
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 16579
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 16580
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 16581
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 16582
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 16583
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0153,
+      "step": 16584
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 16585
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 16586
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 16587
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 16588
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0179,
+      "step": 16589
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 16590
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 16591
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 16592
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 16593
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16594
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 16595
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 16596
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 16597
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 16598
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 16599
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 16600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 16601
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.075,
+      "step": 16602
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16603
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 16604
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 16605
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16606
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16607
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 16608
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 16609
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 16610
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 16611
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16612
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 16613
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0112,
+      "step": 16614
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 16615
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.014,
+      "step": 16616
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 16617
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16618
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 16619
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 16620
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 16621
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 16622
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 16623
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 16624
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 16625
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 16626
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 16627
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16628
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16629
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 16630
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 16631
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 16632
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 16633
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 16634
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 16635
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16636
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 16637
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 16638
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 16639
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 16640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 16641
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 16642
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 16643
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 16644
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 16645
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 16646
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0174,
+      "step": 16647
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0184,
+      "step": 16648
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 16649
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 16650
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 16651
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 16652
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 16653
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 16654
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 16655
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 16656
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 16657
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 16658
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 16659
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 16660
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 16661
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16662
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 16663
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 16664
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16665
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 16666
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 16667
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16668
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 16669
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16670
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 16671
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 16672
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 16673
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16674
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 16675
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 16676
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0112,
+      "step": 16677
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0153,
+      "step": 16678
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 16679
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 16680
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 16681
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 16682
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 16683
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 16684
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 16685
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 16686
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 16687
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 16688
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 16689
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16690
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 16691
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 16692
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 16693
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16694
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 16695
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16696
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 16697
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 16698
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 16699
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 16700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 16701
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 16702
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 16703
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 16704
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16705
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 16706
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 16707
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 16708
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 16709
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 16710
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 16711
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 16712
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 16713
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 16714
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 16715
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 16716
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0594,
+      "step": 16717
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16718
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 16719
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 16720
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 16721
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 16722
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16723
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 16724
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 16725
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 16726
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 16727
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 16728
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 16729
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 16730
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 16731
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 16732
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 16733
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 16734
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 16735
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16736
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 16737
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16738
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 16739
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16740
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 16741
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 16742
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 16743
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 16744
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 16745
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 16746
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 16747
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 16748
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 16749
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 16750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 16751
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0144,
+      "step": 16752
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0162,
+      "step": 16753
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 16754
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 16755
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 16756
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 16757
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 16758
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 16759
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 16760
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 16761
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0625,
+      "step": 16762
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 16763
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 16764
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16765
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0112,
+      "step": 16766
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16767
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 16768
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 16769
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 16770
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 16771
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 16772
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16773
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 16774
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16775
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 16776
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16777
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 16778
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 16779
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 16780
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 16781
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 16782
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 16783
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 16784
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 16785
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 16786
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 16787
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 16788
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16789
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 16790
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 16791
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 16792
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 16793
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 16794
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0647,
+      "step": 16795
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 16796
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 16797
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 16798
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 16799
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 16800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 16801
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 16802
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 16803
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 16804
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16805
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 16806
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0657,
+      "step": 16807
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 16808
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 16809
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 16810
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 16811
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 16812
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 16813
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 16814
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 16815
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 16816
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 16817
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 16818
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 16819
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 16820
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 16821
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 16822
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 16823
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 16824
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 16825
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 16826
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 16827
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 16828
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0514,
+      "step": 16829
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 16830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 16831
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 16832
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 16833
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 16834
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 16835
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 16836
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 16837
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 16838
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 16839
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16840
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 16841
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16842
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 16843
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 16844
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 16845
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 16846
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 16847
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0589,
+      "step": 16848
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0145,
+      "step": 16849
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 16850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 16851
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 16852
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 16853
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 16854
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 16855
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 16856
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 16857
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 16858
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 16859
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 16860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 16861
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 16862
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 16863
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 16864
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16865
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 16866
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 16867
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 16868
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 16869
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 16870
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 16871
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 16872
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 16873
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 16874
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 16875
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 16876
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 16877
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 16878
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 16879
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 16880
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 16881
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 16882
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 16883
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 16884
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16885
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 16886
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 16887
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 16888
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16889
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 16890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 16891
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 16892
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 16893
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 16894
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 16895
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 16896
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 16897
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 16898
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 16899
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 16900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 16901
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 16902
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 16903
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 16904
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 16905
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 16906
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 16907
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 16908
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 16909
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 16910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 16911
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 16912
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 16913
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 16914
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.016,
+      "step": 16915
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 16916
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 16917
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 16918
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 16919
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 16920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 16921
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 16922
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 16923
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 16924
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 16925
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 16926
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 16927
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 16928
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 16929
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 16930
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 16931
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 16932
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 16933
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 16934
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 16935
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 16936
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 16937
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 16938
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 16939
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 16940
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 16941
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 16942
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 16943
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 16944
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 16945
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 16946
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0177,
+      "step": 16947
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 16948
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 16949
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 16950
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 16951
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 16952
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 16953
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 16954
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 16955
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 16956
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 16957
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 16958
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 16959
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 16960
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 16961
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16962
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 16963
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 16964
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 16965
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 16966
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 16967
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 16968
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 16969
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 16970
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 16971
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 16972
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 16973
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 16974
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 16975
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 16976
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 16977
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 16978
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 16979
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 16980
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 16981
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 16982
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0649,
+      "step": 16983
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 16984
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 16985
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 16986
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 16987
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 16988
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 16989
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 16990
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 16991
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 16992
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0397,
+      "step": 16993
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 16994
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 16995
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 16996
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 16997
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 16998
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 16999
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 17000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 17001
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 17002
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 17003
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 17004
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 17005
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 17006
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 17007
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 17008
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 17009
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 17010
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 17011
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 17012
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 17013
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 17014
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 17015
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 17016
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 17017
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 17018
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 17019
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0146,
+      "step": 17020
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 17021
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 17022
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 17023
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17024
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 17025
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 17026
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 17027
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17028
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 17029
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 17030
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 17031
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2e-05,
+      "loss": 0.0156,
+      "step": 17032
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 17033
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 17034
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 17035
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 17036
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 17037
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 17038
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 17039
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 17040
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 17041
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 17042
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 17043
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 17044
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 17045
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 17046
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 17047
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0159,
+      "step": 17048
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0148,
+      "step": 17049
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 17050
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 17051
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 17052
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 17053
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 17054
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 17055
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 17056
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0145,
+      "step": 17057
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 17058
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 17059
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 17060
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0602,
+      "step": 17061
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 17062
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 17063
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17064
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 17065
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 17066
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 17067
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.018,
+      "step": 17068
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 17069
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 17070
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 17071
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0168,
+      "step": 17072
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 17073
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17074
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 17075
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 17076
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 17077
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 17078
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 17079
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.035,
+      "step": 17080
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 17081
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 17082
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 17083
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 17084
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 17085
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 17086
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 17087
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 17088
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 17089
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0551,
+      "step": 17090
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0117,
+      "step": 17091
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 17092
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17093
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 17094
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 17095
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 17096
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0508,
+      "step": 17097
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 17098
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 17099
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 17100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 17101
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 17102
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0188,
+      "step": 17103
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 17104
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 17105
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17106
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 17107
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 17108
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 17109
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 17110
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 17111
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 17112
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0664,
+      "step": 17113
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 17114
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0631,
+      "step": 17115
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 17116
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17117
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 17118
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 17119
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 17120
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17121
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 17122
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 17123
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 17124
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 17125
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17126
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 17127
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 17128
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 17129
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 17130
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17131
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0225,
+      "step": 17132
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 17133
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17134
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 17135
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 17136
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17137
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 17138
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0183,
+      "step": 17139
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 17140
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17141
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17142
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17143
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 17144
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0093,
+      "step": 17145
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 17146
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17147
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 17148
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 17149
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 17150
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17151
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 17152
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 17153
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 17154
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 17155
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 17156
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 17157
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 17158
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17159
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17160
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 17161
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 17162
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 17163
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 17164
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 17165
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 17166
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 17167
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 17168
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 17169
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 17170
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0173,
+      "step": 17171
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17172
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 17173
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 17174
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17175
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 17176
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17177
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 17178
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 17179
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17180
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 17181
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17182
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 17183
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 17184
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17185
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17186
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 17187
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17188
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 17189
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 17190
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 17191
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 17192
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 17193
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 17194
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 17195
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 17196
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 17197
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0174,
+      "step": 17198
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 17199
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0235,
+      "step": 17200
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 17201
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 17202
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 17203
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 17204
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 17205
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 17206
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 17207
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 17208
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 17209
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 17210
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17211
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 17212
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 17213
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 17214
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17215
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17216
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 17217
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 17218
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 17219
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 17221
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 17222
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 17223
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 17224
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 17225
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 17226
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 17227
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 17228
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17229
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 17230
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0442,
+      "step": 17231
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 17232
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 17233
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 17234
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 17235
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 17236
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 17237
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 17238
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 17239
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17240
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 17241
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17242
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 17243
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 17244
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17245
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 17246
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 17247
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 17248
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 17249
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 17250
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17251
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 17252
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 17253
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 17254
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17255
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0592,
+      "step": 17256
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17257
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 17258
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 17259
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 17260
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 17261
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 17262
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 17263
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 17264
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 17265
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0263,
+      "step": 17266
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 17267
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17268
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17269
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 17270
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0703,
+      "step": 17271
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 17272
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17273
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 17274
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 17275
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 17276
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 17277
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 17278
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 17279
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 17280
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17281
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 17282
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 17283
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 17284
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 17285
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 17286
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 17287
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 17288
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 17289
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 17290
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 17291
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0165,
+      "step": 17292
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17293
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 17294
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 17295
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 17296
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 17297
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 17298
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 17299
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 17300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 17301
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0575,
+      "step": 17302
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 17303
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 17304
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 17305
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 17306
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17307
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 17308
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 17309
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17310
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 17311
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 17312
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 17313
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0123,
+      "step": 17314
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17315
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 17316
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 17317
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 17318
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 17319
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 17320
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 17321
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 17322
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17323
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 17324
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 17325
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 17326
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 17327
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 17328
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 17329
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 17330
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 17331
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 17332
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 17333
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 17334
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 17335
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 17336
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 17337
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 17338
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 17339
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 17340
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 17341
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 17342
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 17343
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0159,
+      "step": 17344
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 17345
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 17346
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 17347
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17348
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 17349
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 17350
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 17351
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17352
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0193,
+      "step": 17353
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 17354
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 17355
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 17356
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.077,
+      "step": 17357
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 17358
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 17359
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 17360
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 17361
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 17362
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 17363
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 17364
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 17365
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 17366
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 17367
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 17368
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 17369
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 17370
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 17371
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 17372
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 17373
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 17374
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 17375
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 17376
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 17377
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 17378
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 17379
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 17380
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0182,
+      "step": 17381
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 17382
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 17383
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 17384
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17385
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 17386
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 17387
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0166,
+      "step": 17388
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0459,
+      "step": 17389
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 17390
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 17391
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0666,
+      "step": 17392
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17393
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17394
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17395
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 17396
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17397
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17398
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 17399
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 17401
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 17402
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 17403
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 17404
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 17405
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 17406
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17407
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 17408
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 17409
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 17410
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 17411
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 17412
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 17413
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17414
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 17415
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 17416
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 17417
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 17418
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 17419
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 17420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 17421
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 17422
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0593,
+      "step": 17423
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 17424
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17425
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.017,
+      "step": 17426
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17427
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17428
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0241,
+      "step": 17429
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 17431
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0132,
+      "step": 17432
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 17433
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 17434
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 17435
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 17436
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 17437
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 17438
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17439
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 17440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 17441
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17442
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17443
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 17444
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 17445
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0736,
+      "step": 17446
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 17447
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 17448
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 17449
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17450
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 17451
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 17452
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 17453
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 17454
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0152,
+      "step": 17455
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.013,
+      "step": 17456
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 17457
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17458
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2e-05,
+      "loss": 0.0161,
+      "step": 17459
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 17460
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 17461
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0234,
+      "step": 17462
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17463
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 17464
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 17465
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 17466
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 17467
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 17468
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 17469
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17470
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 17471
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 17472
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0571,
+      "step": 17473
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0174,
+      "step": 17474
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 17475
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 17476
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 17477
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17478
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 17479
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0564,
+      "step": 17481
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 17482
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 17483
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17484
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0701,
+      "step": 17485
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 17486
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0253,
+      "step": 17487
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 17488
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 17489
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 17490
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17491
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 17492
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 17493
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 17494
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17495
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0553,
+      "step": 17496
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 17497
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 17498
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 17499
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 17500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 17501
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 17502
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17503
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 17504
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 17505
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 17506
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 17507
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17508
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0149,
+      "step": 17509
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0365,
+      "step": 17510
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 17511
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17512
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 17513
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0474,
+      "step": 17514
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17515
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 17516
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 17517
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 17518
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 17519
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 17520
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 17521
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 17522
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0147,
+      "step": 17523
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 17524
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 17525
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 17526
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0419,
+      "step": 17527
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.011,
+      "step": 17528
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 17529
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 17530
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 17531
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 17532
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17533
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 17534
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 17535
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 17536
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 17537
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 17538
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 17539
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 17540
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 17541
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.018,
+      "step": 17542
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 17543
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 17544
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17545
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 17546
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 17547
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17548
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 17549
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 17550
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 17551
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17552
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0499,
+      "step": 17553
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 17554
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 17555
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0595,
+      "step": 17556
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17557
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 17558
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 17559
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 17560
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 17561
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 17562
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17563
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17564
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 17565
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17566
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 17567
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 17568
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 17569
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 17570
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.048,
+      "step": 17571
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 17572
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17573
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0752,
+      "step": 17574
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 17575
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 17576
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 17577
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 17578
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 17579
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0634,
+      "step": 17580
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 17581
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 17582
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 17583
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 17584
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 17585
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 17586
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 17587
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17588
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 17589
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17590
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 17591
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 17592
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 17593
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 17594
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 17595
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 17596
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0127,
+      "step": 17597
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 17598
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 17599
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 17600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 17601
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2e-05,
+      "loss": 0.0616,
+      "step": 17602
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 17603
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0913,
+      "step": 17604
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 17605
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 17606
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 17607
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 17608
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 17609
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 17610
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 17611
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17612
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 17613
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 17614
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 17615
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 17616
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 17617
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 17618
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 17619
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 17620
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 17621
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 17622
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 17623
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 17624
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17625
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 17626
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 17627
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 17628
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 17629
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 17630
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 17631
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 17632
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17633
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 17634
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 17635
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 17636
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 17637
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 17638
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 17639
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 17640
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 17641
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 17642
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 17643
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17644
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17645
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 17646
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 17647
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0439,
+      "step": 17648
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 17649
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0579,
+      "step": 17650
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0175,
+      "step": 17651
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 17652
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 17653
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0644,
+      "step": 17654
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 17655
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 17656
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 17657
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0143,
+      "step": 17658
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 17659
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0549,
+      "step": 17660
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 17661
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17662
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 17663
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 17664
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 17665
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 17666
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 17667
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 17668
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 17669
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 17670
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 17671
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0745,
+      "step": 17672
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17673
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17674
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 17675
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 17676
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 17677
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 17678
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 17679
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 17680
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17681
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 17682
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 17683
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17684
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 17685
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 17686
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 17687
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 17688
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 17689
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 17690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 17691
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 17692
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 17693
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17694
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17695
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 17696
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 17697
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 17698
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 17699
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 17700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17701
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17702
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 17703
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 17704
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0803,
+      "step": 17705
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 17706
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 17707
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 17708
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0465,
+      "step": 17709
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 17710
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 17711
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0643,
+      "step": 17712
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0297,
+      "step": 17713
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 17714
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 17715
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 17716
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 17717
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17718
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 17719
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 17720
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 17721
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 17722
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 17723
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 17724
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17725
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 17726
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17727
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 17728
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 17729
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 17730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 17731
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0375,
+      "step": 17732
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 17733
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 17734
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17735
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 17736
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 17737
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 17738
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0583,
+      "step": 17739
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17740
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 17741
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 17742
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 17743
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 17744
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 17745
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 17746
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17747
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 17748
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0271,
+      "step": 17749
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 17750
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 17751
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17752
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 17753
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 17754
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17755
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 17756
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17757
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 17758
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 17759
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 17760
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 17761
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17762
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 17763
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 17764
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 17765
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17766
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 17767
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 17768
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 17769
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 17770
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 17771
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 17772
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 17773
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 17774
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 17775
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0605,
+      "step": 17776
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 17777
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 17778
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17779
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 17780
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 17781
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17782
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 17783
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 17784
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 17785
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17786
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 17787
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 17788
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17789
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 17790
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 17791
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 17792
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 17793
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17794
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 17795
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 17796
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 17797
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 17798
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 17799
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 17800
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 17801
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 17802
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 17803
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 17804
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17805
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 17806
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 17807
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 17808
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.058,
+      "step": 17809
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 17810
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 17811
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17812
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0324,
+      "step": 17813
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 17814
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0677,
+      "step": 17815
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 17816
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 17817
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 17818
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17819
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 17820
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 17821
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 17822
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 17823
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0192,
+      "step": 17824
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 17825
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 17826
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 17827
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.017,
+      "step": 17828
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17829
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 17830
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 17831
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 17832
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 17833
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 17834
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 17835
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 17836
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17837
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 17838
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 17839
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 17840
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17841
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 17842
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 17843
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 17844
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 17845
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 17846
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17847
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0591,
+      "step": 17848
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 17849
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 17850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 17851
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 17852
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 17853
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 17854
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17855
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 17856
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 17857
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17858
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 17859
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 17860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 17861
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 17862
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 17863
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0581,
+      "step": 17864
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17865
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 17866
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 17867
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 17868
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 17869
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 17870
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0565,
+      "step": 17871
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 17872
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 17873
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 17874
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 17875
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 17876
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 17877
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 17878
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0622,
+      "step": 17879
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 17880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 17881
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 17882
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 17883
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 17884
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17885
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 17886
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17887
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 17888
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.029,
+      "step": 17889
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 17890
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 17891
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 17892
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 17893
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 17894
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 17895
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 17896
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 17897
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 17898
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 17899
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 17900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 17901
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 17902
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 17903
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 17904
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 17905
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 17906
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 17907
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 17908
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0134,
+      "step": 17909
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 17910
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17911
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 17912
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 17913
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17914
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.024,
+      "step": 17915
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 17916
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 17917
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0578,
+      "step": 17918
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 17919
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 17920
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 17921
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0577,
+      "step": 17922
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17923
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 17924
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 17925
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 17926
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 17927
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 17928
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0392,
+      "step": 17929
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 17930
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17931
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 17932
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 17933
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 17934
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 17935
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 17936
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 17937
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0476,
+      "step": 17938
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 17939
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 17940
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 17941
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 17942
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 17943
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 17944
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 17945
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0256,
+      "step": 17946
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 17947
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 17948
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 17949
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 17950
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0327,
+      "step": 17951
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 17952
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 17953
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 17954
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0202,
+      "step": 17955
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 17956
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 17957
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 17958
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 17959
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 17960
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 17961
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 17962
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 17963
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 17964
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 17965
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 17966
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 17967
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 17968
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0423,
+      "step": 17969
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 17970
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 17971
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0674,
+      "step": 17972
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 17973
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 17974
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 17975
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 17976
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 17977
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 17978
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 17979
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 17980
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 17981
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 17982
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 17983
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 17984
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 17985
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 17986
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 17987
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 17988
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 17989
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0335,
+      "step": 17990
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 17991
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 17992
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0366,
+      "step": 17993
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 17994
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 17995
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 17996
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 17997
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 17998
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 17999
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0332,
+      "step": 18000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 18001
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 18002
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 18003
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 18004
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18005
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 18006
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18007
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 18008
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 18009
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 18010
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 18011
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0161,
+      "step": 18012
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 18013
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 18014
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0128,
+      "step": 18015
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 18016
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 18017
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 18018
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 18019
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 18020
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 18021
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 18022
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0152,
+      "step": 18023
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 18024
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18025
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 18026
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 18027
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0523,
+      "step": 18028
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 18029
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18030
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 18031
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 18032
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18033
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 18034
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 18035
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 18036
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 18037
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 18038
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 18039
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0478,
+      "step": 18040
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 18041
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18042
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 18043
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 18044
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 18045
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18046
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 18047
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 18048
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18049
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 18050
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0652,
+      "step": 18051
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18052
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 18053
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 18054
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 18055
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 18056
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0139,
+      "step": 18057
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.055,
+      "step": 18058
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 18059
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 18060
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 18061
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18062
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 18063
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 18064
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 18065
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 18066
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 18067
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 18068
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0655,
+      "step": 18069
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 18070
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 18071
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 18072
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0562,
+      "step": 18073
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 18074
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.015,
+      "step": 18075
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0145,
+      "step": 18076
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 18077
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 18078
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 18079
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 18080
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0213,
+      "step": 18081
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 18082
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18083
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 18084
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 18085
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 18086
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 18087
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18088
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 18089
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 18090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 18091
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18092
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 18093
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 18094
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 18095
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 18096
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 18097
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 18098
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 18099
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 18100
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 18101
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 18102
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 18103
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 18104
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0305,
+      "step": 18105
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18106
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 18107
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 18108
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 18109
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 18110
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0587,
+      "step": 18111
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0209,
+      "step": 18112
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 18113
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 18114
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 18115
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 18116
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 18117
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 18118
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0449,
+      "step": 18119
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 18120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 18121
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0185,
+      "step": 18122
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 18123
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 18124
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 18125
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 18126
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18127
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 18128
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 18129
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0543,
+      "step": 18130
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0146,
+      "step": 18131
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 18132
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 18133
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 18134
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 18135
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 18136
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 18137
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 18138
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18139
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.021,
+      "step": 18140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18141
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 18142
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 18143
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18144
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 18145
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 18146
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 18147
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 18148
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 18149
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.02,
+      "step": 18150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 18151
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0705,
+      "step": 18152
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 18153
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0223,
+      "step": 18154
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 18155
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18156
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 18157
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 18158
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 18159
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 18160
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 18161
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 18162
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0088,
+      "step": 18163
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 18164
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 18165
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18166
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18167
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 18168
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 18169
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 18170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 18171
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 18172
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18173
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 18174
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 18175
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 18176
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 18177
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18178
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 18179
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 18180
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 18181
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0306,
+      "step": 18182
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 18183
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 18184
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 18185
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 18186
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 18187
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 18188
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 18189
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 18190
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 18191
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18192
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0712,
+      "step": 18193
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0532,
+      "step": 18194
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.03,
+      "step": 18195
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.052,
+      "step": 18196
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18197
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18198
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18199
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.049,
+      "step": 18200
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 18201
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18202
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 18203
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 18204
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 18205
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 18206
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 18207
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 18208
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 18209
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0257,
+      "step": 18210
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 18211
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0357,
+      "step": 18212
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 18213
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 18214
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 18215
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 18216
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 18217
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0262,
+      "step": 18218
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 18219
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 18220
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 18221
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 18222
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0233,
+      "step": 18223
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 18224
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0204,
+      "step": 18225
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 18226
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.045,
+      "step": 18227
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0452,
+      "step": 18228
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 18229
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0642,
+      "step": 18230
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 18231
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 18232
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 18233
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 18234
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 18235
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 18236
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 18237
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 18238
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 18239
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18240
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0194,
+      "step": 18241
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 18242
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 18243
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0637,
+      "step": 18244
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 18245
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 18246
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 18247
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0531,
+      "step": 18248
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0493,
+      "step": 18249
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0307,
+      "step": 18250
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0503,
+      "step": 18251
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0152,
+      "step": 18252
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 18253
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 18254
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 18255
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 18256
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18257
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18258
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 18259
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0427,
+      "step": 18260
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 18261
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0131,
+      "step": 18262
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0284,
+      "step": 18263
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 18264
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0488,
+      "step": 18265
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.0618,
+      "step": 18266
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 18267
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18268
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0468,
+      "step": 18269
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0515,
+      "step": 18270
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18271
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 18272
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 18273
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 18274
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 18275
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18276
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 18277
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18278
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 18279
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0502,
+      "step": 18280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 18281
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 18282
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 18283
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18284
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0451,
+      "step": 18285
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0477,
+      "step": 18286
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 18287
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 18288
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 18289
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0161,
+      "step": 18291
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 18292
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18293
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 18294
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 18295
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 18296
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0462,
+      "step": 18297
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0489,
+      "step": 18298
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 18299
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 18300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 18301
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0279,
+      "step": 18302
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 18303
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 18304
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 18305
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0471,
+      "step": 18306
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0243,
+      "step": 18307
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 18308
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0352,
+      "step": 18309
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 18310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 18311
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0801,
+      "step": 18312
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 18313
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 18314
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 18315
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 18316
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18317
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.015,
+      "step": 18318
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0607,
+      "step": 18319
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 18320
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0283,
+      "step": 18321
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18322
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 18323
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18324
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 18325
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 18326
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 18327
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.023,
+      "step": 18328
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 18329
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 18330
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 18331
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0369,
+      "step": 18332
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0494,
+      "step": 18333
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0454,
+      "step": 18334
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 18335
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 18336
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0172,
+      "step": 18337
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 18338
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 18339
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 18340
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 18341
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0157,
+      "step": 18342
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0189,
+      "step": 18343
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0138,
+      "step": 18344
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 18345
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 18346
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 18347
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 18348
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0267,
+      "step": 18349
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0218,
+      "step": 18350
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18351
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 18352
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 18353
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18354
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0555,
+      "step": 18355
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 18356
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0528,
+      "step": 18357
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 18358
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0321,
+      "step": 18359
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 18360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 18361
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18362
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0323,
+      "step": 18363
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0391,
+      "step": 18364
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18365
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 18366
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 18367
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 18368
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 18369
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0429,
+      "step": 18370
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 18371
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 18372
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 18373
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0197,
+      "step": 18374
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 18375
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 18376
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.047,
+      "step": 18377
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0615,
+      "step": 18378
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 18379
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18380
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0498,
+      "step": 18381
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0612,
+      "step": 18382
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18383
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 18384
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 18385
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 18386
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.04,
+      "step": 18387
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0574,
+      "step": 18388
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0466,
+      "step": 18389
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 18390
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 18391
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0542,
+      "step": 18392
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18393
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0546,
+      "step": 18394
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 18395
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 18396
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 18397
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 18398
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0513,
+      "step": 18399
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0229,
+      "step": 18400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 18401
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0516,
+      "step": 18402
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0529,
+      "step": 18403
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18404
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0527,
+      "step": 18405
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0505,
+      "step": 18406
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 18407
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18408
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0384,
+      "step": 18409
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18410
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18411
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 18412
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 18413
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 18414
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0358,
+      "step": 18415
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 18416
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 18417
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18418
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 18419
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 18420
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 18421
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18422
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 18423
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 18424
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18425
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0446,
+      "step": 18426
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18427
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 18428
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 18429
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 18430
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0695,
+      "step": 18431
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 18432
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0309,
+      "step": 18433
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18434
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18435
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 18436
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0227,
+      "step": 18437
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 18438
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.073,
+      "step": 18439
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 18440
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 18441
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 18442
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 18443
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 18444
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 18445
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 18446
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0428,
+      "step": 18447
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18448
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 18449
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 18450
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 18451
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 18452
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 18453
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0416,
+      "step": 18454
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0308,
+      "step": 18455
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 18456
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 18457
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 18458
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 18459
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18460
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18461
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0568,
+      "step": 18462
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 18463
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0755,
+      "step": 18464
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 18465
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0379,
+      "step": 18466
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18467
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0539,
+      "step": 18468
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 18469
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 18470
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 18471
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 18472
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 18473
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0545,
+      "step": 18474
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 18475
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 18476
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 18477
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18478
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 18479
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18480
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 18481
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18482
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18483
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 18484
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0426,
+      "step": 18485
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0623,
+      "step": 18486
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0432,
+      "step": 18487
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 18488
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 18489
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0444,
+      "step": 18490
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0211,
+      "step": 18491
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0346,
+      "step": 18492
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 18493
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 18494
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.068,
+      "step": 18495
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 18496
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.056,
+      "step": 18497
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0536,
+      "step": 18498
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0596,
+      "step": 18499
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0691,
+      "step": 18500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 18501
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.022,
+      "step": 18502
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0457,
+      "step": 18503
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18504
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 18505
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0328,
+      "step": 18506
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0275,
+      "step": 18507
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0151,
+      "step": 18508
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0378,
+      "step": 18509
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18510
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 18511
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 18512
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 18513
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 18514
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.057,
+      "step": 18515
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0334,
+      "step": 18516
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 18517
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0482,
+      "step": 18518
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0686,
+      "step": 18519
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 18520
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.025,
+      "step": 18521
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0338,
+      "step": 18522
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 18523
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18524
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0359,
+      "step": 18525
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 18526
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0544,
+      "step": 18527
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 18528
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0569,
+      "step": 18529
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18530
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 18531
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 18532
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.032,
+      "step": 18533
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18534
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 18535
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 18536
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 18537
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0298,
+      "step": 18538
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0364,
+      "step": 18539
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 18540
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0401,
+      "step": 18541
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0437,
+      "step": 18542
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.062,
+      "step": 18543
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 18544
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0376,
+      "step": 18545
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 18546
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 18547
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 18548
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0268,
+      "step": 18549
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18550
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0208,
+      "step": 18551
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2e-05,
+      "loss": 0.0295,
+      "step": 18552
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 18553
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0302,
+      "step": 18554
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 18555
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 18556
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 18557
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 18558
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 18559
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18560
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 18561
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 18562
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18563
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 18564
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0258,
+      "step": 18565
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 18566
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 18567
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0385,
+      "step": 18568
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 18569
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 18570
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 18571
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18572
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0588,
+      "step": 18573
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0231,
+      "step": 18574
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0798,
+      "step": 18575
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0281,
+      "step": 18576
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 18577
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 18578
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 18579
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0169,
+      "step": 18580
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18581
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0245,
+      "step": 18582
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0469,
+      "step": 18583
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 18584
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0187,
+      "step": 18585
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0351,
+      "step": 18586
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18587
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18588
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0538,
+      "step": 18589
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 18590
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18591
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 18592
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 18593
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.066,
+      "step": 18594
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0517,
+      "step": 18595
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 18596
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 18597
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0408,
+      "step": 18598
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18599
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2e-05,
+      "loss": 0.0237,
+      "step": 18600
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0557,
+      "step": 18601
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0606,
+      "step": 18602
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0411,
+      "step": 18603
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18604
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.051,
+      "step": 18605
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 18606
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0163,
+      "step": 18607
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0198,
+      "step": 18608
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18609
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 18610
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18611
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0135,
+      "step": 18612
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 18613
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0463,
+      "step": 18614
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 18615
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18616
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18617
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 18618
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0239,
+      "step": 18619
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 18620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0507,
+      "step": 18621
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 18622
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18623
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0312,
+      "step": 18624
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.044,
+      "step": 18625
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0697,
+      "step": 18626
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 18627
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0226,
+      "step": 18628
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0389,
+      "step": 18629
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.041,
+      "step": 18630
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 18631
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0377,
+      "step": 18632
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0201,
+      "step": 18633
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 18634
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0491,
+      "step": 18635
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0303,
+      "step": 18636
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 18637
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18638
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 18639
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 18640
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0318,
+      "step": 18641
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0367,
+      "step": 18642
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 18643
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0259,
+      "step": 18644
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18645
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0759,
+      "step": 18646
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2e-05,
+      "loss": 0.0393,
+      "step": 18647
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18648
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0566,
+      "step": 18649
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 18650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18651
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0733,
+      "step": 18652
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 18653
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0261,
+      "step": 18654
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 18655
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0648,
+      "step": 18656
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0109,
+      "step": 18657
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0561,
+      "step": 18658
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0436,
+      "step": 18659
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 18660
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0304,
+      "step": 18661
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0559,
+      "step": 18662
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 18663
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 18664
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18665
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0329,
+      "step": 18666
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 18667
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 18668
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0195,
+      "step": 18669
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 18670
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18671
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 18672
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18673
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18674
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.013,
+      "step": 18675
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18676
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 18677
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0186,
+      "step": 18678
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0286,
+      "step": 18679
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 18680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0504,
+      "step": 18681
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.028,
+      "step": 18682
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 18683
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0767,
+      "step": 18684
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 18685
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0558,
+      "step": 18686
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 18687
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0534,
+      "step": 18688
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0435,
+      "step": 18689
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18690
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.017,
+      "step": 18691
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18692
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0405,
+      "step": 18693
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18694
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 18695
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0431,
+      "step": 18696
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 18697
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0215,
+      "step": 18698
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 18699
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0301,
+      "step": 18700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0326,
+      "step": 18701
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0232,
+      "step": 18702
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0453,
+      "step": 18703
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18704
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 18705
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 18706
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 18707
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0222,
+      "step": 18708
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0461,
+      "step": 18709
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 18710
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0541,
+      "step": 18711
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0368,
+      "step": 18712
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0387,
+      "step": 18713
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0394,
+      "step": 18714
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0247,
+      "step": 18715
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0255,
+      "step": 18716
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 18717
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 18718
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18719
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 18720
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 18721
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18722
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0443,
+      "step": 18723
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0396,
+      "step": 18724
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0525,
+      "step": 18725
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0144,
+      "step": 18726
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 18727
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0381,
+      "step": 18728
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0274,
+      "step": 18729
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0214,
+      "step": 18730
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0277,
+      "step": 18731
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 18732
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0142,
+      "step": 18733
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 18734
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0299,
+      "step": 18735
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 18736
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 18737
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 18738
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0611,
+      "step": 18739
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0339,
+      "step": 18740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0353,
+      "step": 18741
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2e-05,
+      "loss": 0.0178,
+      "step": 18742
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0199,
+      "step": 18743
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0404,
+      "step": 18744
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18745
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18746
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 18747
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 18748
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0519,
+      "step": 18749
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 18750
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0524,
+      "step": 18751
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0337,
+      "step": 18752
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 18753
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0506,
+      "step": 18754
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.031,
+      "step": 18755
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0412,
+      "step": 18756
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 18757
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0547,
+      "step": 18758
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0601,
+      "step": 18759
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 18760
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 18761
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0665,
+      "step": 18762
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 18763
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 18764
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 18765
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0386,
+      "step": 18766
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0244,
+      "step": 18767
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0355,
+      "step": 18768
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 18769
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.054,
+      "step": 18770
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18771
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0486,
+      "step": 18772
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18773
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 18774
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0464,
+      "step": 18775
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0496,
+      "step": 18776
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0206,
+      "step": 18777
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0445,
+      "step": 18778
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0597,
+      "step": 18779
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0217,
+      "step": 18780
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0207,
+      "step": 18781
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0285,
+      "step": 18782
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0455,
+      "step": 18783
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0456,
+      "step": 18784
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18785
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18786
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0422,
+      "step": 18787
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0361,
+      "step": 18788
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.053,
+      "step": 18789
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2e-05,
+      "loss": 0.0413,
+      "step": 18790
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 18791
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0522,
+      "step": 18792
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 18793
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0521,
+      "step": 18794
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0447,
+      "step": 18795
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 18796
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 18797
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0166,
+      "step": 18798
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0238,
+      "step": 18799
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0586,
+      "step": 18800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 18801
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18802
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0563,
+      "step": 18803
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0345,
+      "step": 18804
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 18805
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 18806
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 18807
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0676,
+      "step": 18808
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 18809
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0248,
+      "step": 18810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0313,
+      "step": 18811
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0347,
+      "step": 18812
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18813
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 18814
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0472,
+      "step": 18815
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0167,
+      "step": 18816
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0448,
+      "step": 18817
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.034,
+      "step": 18818
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 18819
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 18820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18821
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 18822
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 18823
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0682,
+      "step": 18824
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 18825
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18826
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0395,
+      "step": 18827
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0497,
+      "step": 18828
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 18829
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0628,
+      "step": 18831
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 18832
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 18833
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18834
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0349,
+      "step": 18835
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0272,
+      "step": 18836
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2e-05,
+      "loss": 0.0441,
+      "step": 18837
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0296,
+      "step": 18838
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18839
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0342,
+      "step": 18840
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.026,
+      "step": 18841
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18842
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0495,
+      "step": 18843
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0383,
+      "step": 18844
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0548,
+      "step": 18845
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0656,
+      "step": 18846
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18847
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18848
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0254,
+      "step": 18849
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 18850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 18851
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.042,
+      "step": 18852
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 18853
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0424,
+      "step": 18854
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 18855
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 18856
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0336,
+      "step": 18857
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0288,
+      "step": 18858
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0509,
+      "step": 18859
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0236,
+      "step": 18860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0794,
+      "step": 18861
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0576,
+      "step": 18862
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18863
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0291,
+      "step": 18864
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0264,
+      "step": 18865
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0599,
+      "step": 18866
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0635,
+      "step": 18867
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0713,
+      "step": 18868
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0414,
+      "step": 18869
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0216,
+      "step": 18870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 18871
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0341,
+      "step": 18872
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.043,
+      "step": 18873
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.05,
+      "step": 18874
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0458,
+      "step": 18875
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18876
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0331,
+      "step": 18877
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 18878
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0492,
+      "step": 18879
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 18880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0319,
+      "step": 18881
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0273,
+      "step": 18882
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 18883
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0617,
+      "step": 18884
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 18885
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 18886
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0603,
+      "step": 18887
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0181,
+      "step": 18888
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0406,
+      "step": 18889
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0481,
+      "step": 18890
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 18891
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0511,
+      "step": 18892
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0567,
+      "step": 18893
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0613,
+      "step": 18894
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0501,
+      "step": 18895
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0467,
+      "step": 18896
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 18897
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0399,
+      "step": 18898
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18899
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0418,
+      "step": 18900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0354,
+      "step": 18901
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0415,
+      "step": 18902
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0115,
+      "step": 18903
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0484,
+      "step": 18904
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0287,
+      "step": 18905
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18906
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0191,
+      "step": 18907
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0228,
+      "step": 18908
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0158,
+      "step": 18909
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0372,
+      "step": 18910
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18911
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0407,
+      "step": 18912
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0679,
+      "step": 18913
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0196,
+      "step": 18914
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18915
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0292,
+      "step": 18916
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0266,
+      "step": 18917
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.038,
+      "step": 18918
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18919
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0246,
+      "step": 18920
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0348,
+      "step": 18921
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18922
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.036,
+      "step": 18923
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0311,
+      "step": 18924
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0242,
+      "step": 18925
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0629,
+      "step": 18926
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0293,
+      "step": 18927
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0621,
+      "step": 18928
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0289,
+      "step": 18929
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18930
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0221,
+      "step": 18931
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2e-05,
+      "loss": 0.0537,
+      "step": 18932
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0409,
+      "step": 18933
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0526,
+      "step": 18934
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0425,
+      "step": 18935
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0269,
+      "step": 18936
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0343,
+      "step": 18937
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 18938
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0485,
+      "step": 18939
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0205,
+      "step": 18940
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.027,
+      "step": 18941
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0356,
+      "step": 18942
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.037,
+      "step": 18943
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 18944
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0512,
+      "step": 18945
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18946
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0322,
+      "step": 18947
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0434,
+      "step": 18948
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0219,
+      "step": 18949
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.046,
+      "step": 18950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0533,
+      "step": 18951
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0363,
+      "step": 18952
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0276,
+      "step": 18953
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0662,
+      "step": 18954
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0535,
+      "step": 18955
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18956
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0398,
+      "step": 18957
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0487,
+      "step": 18958
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0483,
+      "step": 18959
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0614,
+      "step": 18960
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.039,
+      "step": 18961
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0233,
+      "step": 18962
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0314,
+      "step": 18963
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 18964
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0362,
+      "step": 18965
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0325,
+      "step": 18966
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0421,
+      "step": 18967
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 18968
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0479,
+      "step": 18969
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0252,
+      "step": 18970
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0552,
+      "step": 18971
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0402,
+      "step": 18972
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0518,
+      "step": 18973
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0156,
+      "step": 18974
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0212,
+      "step": 18975
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0374,
+      "step": 18976
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0169,
+      "step": 18977
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0373,
+      "step": 18978
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0438,
+      "step": 18979
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2e-05,
+      "loss": 0.0382,
+      "step": 18980
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0282,
+      "step": 18981
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0278,
+      "step": 18982
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0403,
+      "step": 18983
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0744,
+      "step": 18984
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0473,
+      "step": 18985
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0164,
+      "step": 18986
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0249,
+      "step": 18987
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18988
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0388,
+      "step": 18989
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0344,
+      "step": 18990
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0584,
+      "step": 18991
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0475,
+      "step": 18992
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0251,
+      "step": 18993
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0317,
+      "step": 18994
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0265,
+      "step": 18995
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0417,
+      "step": 18996
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0556,
+      "step": 18997
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0315,
+      "step": 18998
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.059,
+      "step": 18999
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 19000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0294,
+      "step": 19001
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0371,
+      "step": 19002
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0316,
+      "step": 19003
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2e-05,
+      "loss": 0.0333,
+      "step": 19004
+    },
+    {
+      "epoch": 4.0,
+      "step": 19004,
+      "total_flos": 3979050976542720.0,
+      "train_loss": 0.09924012448907626,
+      "train_runtime": 105322.553,
+      "train_samples_per_second": 23.095,
+      "train_steps_per_second": 0.18
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 19004,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 3979050976542720.0,
+  "trial_name": null,
+  "trial_params": null
+}