diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22590 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "eval_steps": 500,
+  "global_step": 3760,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.849557522123894e-07,
+      "loss": 3.875,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7699115044247788e-06,
+      "loss": 3.8984,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6548672566371683e-06,
+      "loss": 3.8789,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5398230088495575e-06,
+      "loss": 3.8828,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.424778761061947e-06,
+      "loss": 3.9062,
+      "step": 5
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3097345132743365e-06,
+      "loss": 3.8984,
+      "step": 6
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.194690265486726e-06,
+      "loss": 3.8828,
+      "step": 7
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.079646017699115e-06,
+      "loss": 3.8711,
+      "step": 8
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.964601769911505e-06,
+      "loss": 3.8867,
+      "step": 9
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.849557522123894e-06,
+      "loss": 3.832,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.734513274336284e-06,
+      "loss": 3.8867,
+      "step": 11
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0619469026548673e-05,
+      "loss": 3.9023,
+      "step": 12
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1504424778761062e-05,
+      "loss": 3.8203,
+      "step": 13
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2389380530973452e-05,
+      "loss": 3.8477,
+      "step": 14
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3274336283185843e-05,
+      "loss": 3.8789,
+      "step": 15
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.415929203539823e-05,
+      "loss": 3.8125,
+      "step": 16
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.504424778761062e-05,
+      "loss": 3.7891,
+      "step": 17
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.592920353982301e-05,
+      "loss": 3.8086,
+      "step": 18
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6814159292035402e-05,
+      "loss": 3.7969,
+      "step": 19
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7699115044247787e-05,
+      "loss": 3.7852,
+      "step": 20
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.858407079646018e-05,
+      "loss": 3.8008,
+      "step": 21
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.946902654867257e-05,
+      "loss": 3.7734,
+      "step": 22
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.0353982300884957e-05,
+      "loss": 3.7695,
+      "step": 23
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1238938053097346e-05,
+      "loss": 3.7422,
+      "step": 24
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.2123893805309738e-05,
+      "loss": 3.7227,
+      "step": 25
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3008849557522124e-05,
+      "loss": 3.7812,
+      "step": 26
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3893805309734516e-05,
+      "loss": 3.7422,
+      "step": 27
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4778761061946905e-05,
+      "loss": 3.7148,
+      "step": 28
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5663716814159294e-05,
+      "loss": 3.75,
+      "step": 29
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.6548672566371686e-05,
+      "loss": 3.707,
+      "step": 30
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.743362831858407e-05,
+      "loss": 3.6484,
+      "step": 31
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.831858407079646e-05,
+      "loss": 3.6953,
+      "step": 32
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9203539823008852e-05,
+      "loss": 3.6562,
+      "step": 33
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.008849557522124e-05,
+      "loss": 3.6992,
+      "step": 34
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.097345132743363e-05,
+      "loss": 3.6484,
+      "step": 35
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.185840707964602e-05,
+      "loss": 3.582,
+      "step": 36
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.274336283185841e-05,
+      "loss": 3.7227,
+      "step": 37
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.3628318584070804e-05,
+      "loss": 3.6211,
+      "step": 38
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.451327433628319e-05,
+      "loss": 3.5781,
+      "step": 39
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.5398230088495574e-05,
+      "loss": 3.5195,
+      "step": 40
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.628318584070797e-05,
+      "loss": 3.5469,
+      "step": 41
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.716814159292036e-05,
+      "loss": 3.4492,
+      "step": 42
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8053097345132744e-05,
+      "loss": 3.4531,
+      "step": 43
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.893805309734514e-05,
+      "loss": 3.3281,
+      "step": 44
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.982300884955752e-05,
+      "loss": 3.4258,
+      "step": 45
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.0707964601769914e-05,
+      "loss": 3.4453,
+      "step": 46
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.15929203539823e-05,
+      "loss": 3.3008,
+      "step": 47
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.247787610619469e-05,
+      "loss": 3.2695,
+      "step": 48
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3362831858407084e-05,
+      "loss": 3.2188,
+      "step": 49
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4247787610619477e-05,
+      "loss": 3.3086,
+      "step": 50
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5132743362831855e-05,
+      "loss": 3.3945,
+      "step": 51
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.601769911504425e-05,
+      "loss": 3.1992,
+      "step": 52
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.690265486725664e-05,
+      "loss": 3.1562,
+      "step": 53
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.778761061946903e-05,
+      "loss": 3.1719,
+      "step": 54
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.867256637168142e-05,
+      "loss": 3.2656,
+      "step": 55
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.955752212389381e-05,
+      "loss": 3.2461,
+      "step": 56
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.0442477876106195e-05,
+      "loss": 3.125,
+      "step": 57
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.132743362831859e-05,
+      "loss": 3.1836,
+      "step": 58
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.221238938053098e-05,
+      "loss": 3.043,
+      "step": 59
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.309734513274337e-05,
+      "loss": 3.1016,
+      "step": 60
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.398230088495575e-05,
+      "loss": 2.9727,
+      "step": 61
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.486725663716814e-05,
+      "loss": 2.9258,
+      "step": 62
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.575221238938053e-05,
+      "loss": 3.0664,
+      "step": 63
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.663716814159292e-05,
+      "loss": 2.9844,
+      "step": 64
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.752212389380531e-05,
+      "loss": 2.9336,
+      "step": 65
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8407079646017705e-05,
+      "loss": 2.8633,
+      "step": 66
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.92920353982301e-05,
+      "loss": 3.0039,
+      "step": 67
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.017699115044248e-05,
+      "loss": 2.9414,
+      "step": 68
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.106194690265487e-05,
+      "loss": 2.8281,
+      "step": 69
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.194690265486725e-05,
+      "loss": 2.8711,
+      "step": 70
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.283185840707965e-05,
+      "loss": 2.6836,
+      "step": 71
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.371681415929204e-05,
+      "loss": 2.8086,
+      "step": 72
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.460176991150442e-05,
+      "loss": 2.8477,
+      "step": 73
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.548672566371682e-05,
+      "loss": 2.8516,
+      "step": 74
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.637168141592921e-05,
+      "loss": 2.875,
+      "step": 75
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.725663716814161e-05,
+      "loss": 2.7422,
+      "step": 76
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.814159292035398e-05,
+      "loss": 2.6953,
+      "step": 77
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.902654867256638e-05,
+      "loss": 2.6289,
+      "step": 78
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.991150442477876e-05,
+      "loss": 2.793,
+      "step": 79
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.079646017699115e-05,
+      "loss": 2.7617,
+      "step": 80
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.168141592920355e-05,
+      "loss": 2.6758,
+      "step": 81
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.256637168141593e-05,
+      "loss": 2.6602,
+      "step": 82
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 7.345132743362832e-05,
+      "loss": 2.6797,
+      "step": 83
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.433628318584072e-05,
+      "loss": 2.8594,
+      "step": 84
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.522123893805309e-05,
+      "loss": 2.4375,
+      "step": 85
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.610619469026549e-05,
+      "loss": 2.668,
+      "step": 86
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.699115044247787e-05,
+      "loss": 2.7695,
+      "step": 87
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.787610619469027e-05,
+      "loss": 2.4297,
+      "step": 88
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.876106194690266e-05,
+      "loss": 2.582,
+      "step": 89
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.964601769911504e-05,
+      "loss": 2.6523,
+      "step": 90
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.053097345132744e-05,
+      "loss": 2.3672,
+      "step": 91
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.141592920353983e-05,
+      "loss": 2.3438,
+      "step": 92
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.230088495575221e-05,
+      "loss": 2.5156,
+      "step": 93
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.31858407079646e-05,
+      "loss": 2.5742,
+      "step": 94
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.4070796460177e-05,
+      "loss": 2.5898,
+      "step": 95
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.495575221238938e-05,
+      "loss": 2.5703,
+      "step": 96
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.584070796460177e-05,
+      "loss": 2.3242,
+      "step": 97
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.672566371681417e-05,
+      "loss": 2.5703,
+      "step": 98
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.761061946902655e-05,
+      "loss": 2.457,
+      "step": 99
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.849557522123895e-05,
+      "loss": 2.4336,
+      "step": 100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.938053097345133e-05,
+      "loss": 2.5742,
+      "step": 101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.026548672566371e-05,
+      "loss": 2.3691,
+      "step": 102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.115044247787611e-05,
+      "loss": 2.4102,
+      "step": 103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.20353982300885e-05,
+      "loss": 2.3906,
+      "step": 104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.29203539823009e-05,
+      "loss": 2.4336,
+      "step": 105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.380530973451328e-05,
+      "loss": 2.3984,
+      "step": 106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.469026548672566e-05,
+      "loss": 2.4492,
+      "step": 107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.557522123893806e-05,
+      "loss": 2.5391,
+      "step": 108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.646017699115044e-05,
+      "loss": 2.3711,
+      "step": 109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.734513274336283e-05,
+      "loss": 2.4102,
+      "step": 110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.823008849557522e-05,
+      "loss": 2.2168,
+      "step": 111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.911504424778762e-05,
+      "loss": 2.3477,
+      "step": 112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001,
+      "loss": 2.4766,
+      "step": 113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.999998144896262e-05,
+      "loss": 2.5234,
+      "step": 114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.999992579586424e-05,
+      "loss": 2.3711,
+      "step": 115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.999983304074616e-05,
+      "loss": 2.3086,
+      "step": 116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.99997031836772e-05,
+      "loss": 2.0645,
+      "step": 117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.999953622475372e-05,
+      "loss": 2.1738,
+      "step": 118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.999933216409963e-05,
+      "loss": 2.2051,
+      "step": 119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.999909100186632e-05,
+      "loss": 2.2344,
+      "step": 120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.999881273823277e-05,
+      "loss": 2.2383,
+      "step": 121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.999849737340546e-05,
+      "loss": 2.3203,
+      "step": 122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.999814490761838e-05,
+      "loss": 2.1758,
+      "step": 123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.999775534113309e-05,
+      "loss": 2.1406,
+      "step": 124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.999732867423868e-05,
+      "loss": 2.0957,
+      "step": 125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.999686490725173e-05,
+      "loss": 2.3516,
+      "step": 126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.999636404051638e-05,
+      "loss": 2.1953,
+      "step": 127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.99958260744043e-05,
+      "loss": 2.3516,
+      "step": 128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.999525100931469e-05,
+      "loss": 2.4883,
+      "step": 129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.999463884567426e-05,
+      "loss": 2.0664,
+      "step": 130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.999398958393726e-05,
+      "loss": 2.2227,
+      "step": 131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.999330322458547e-05,
+      "loss": 2.1719,
+      "step": 132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.999257976812819e-05,
+      "loss": 2.1523,
+      "step": 133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.999181921510229e-05,
+      "loss": 2.127,
+      "step": 134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.99910215660721e-05,
+      "loss": 2.1094,
+      "step": 135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.99901868216295e-05,
+      "loss": 1.9336,
+      "step": 136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.998931498239394e-05,
+      "loss": 1.9434,
+      "step": 137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.998840604901233e-05,
+      "loss": 2.1074,
+      "step": 138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.998746002215915e-05,
+      "loss": 2.1387,
+      "step": 139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.998647690253638e-05,
+      "loss": 2.1641,
+      "step": 140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.998545669087355e-05,
+      "loss": 2.0,
+      "step": 141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.99843993879277e-05,
+      "loss": 1.8047,
+      "step": 142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.998330499448339e-05,
+      "loss": 2.0352,
+      "step": 143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.99821735113527e-05,
+      "loss": 2.2598,
+      "step": 144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.998100493937523e-05,
+      "loss": 2.0762,
+      "step": 145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.997979927941812e-05,
+      "loss": 2.1035,
+      "step": 146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.997855653237602e-05,
+      "loss": 2.0781,
+      "step": 147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.99772766991711e-05,
+      "loss": 1.7441,
+      "step": 148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.997595978075305e-05,
+      "loss": 2.0684,
+      "step": 149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.997460577809906e-05,
+      "loss": 1.998,
+      "step": 150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.997321469221388e-05,
+      "loss": 2.2109,
+      "step": 151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.997178652412976e-05,
+      "loss": 1.9258,
+      "step": 152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.997032127490642e-05,
+      "loss": 2.0898,
+      "step": 153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.996881894563117e-05,
+      "loss": 1.9258,
+      "step": 154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.99672795374188e-05,
+      "loss": 2.1484,
+      "step": 155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.99657030514116e-05,
+      "loss": 1.877,
+      "step": 156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.996408948877939e-05,
+      "loss": 1.9785,
+      "step": 157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.996243885071949e-05,
+      "loss": 1.9355,
+      "step": 158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.996075113845678e-05,
+      "loss": 1.6895,
+      "step": 159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.995902635324359e-05,
+      "loss": 1.9355,
+      "step": 160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.995726449635977e-05,
+      "loss": 1.7539,
+      "step": 161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.995546556911271e-05,
+      "loss": 1.9062,
+      "step": 162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.995362957283727e-05,
+      "loss": 1.9336,
+      "step": 163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.995175650889586e-05,
+      "loss": 1.9785,
+      "step": 164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.994984637867835e-05,
+      "loss": 1.9961,
+      "step": 165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.994789918360214e-05,
+      "loss": 1.8594,
+      "step": 166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.994591492511213e-05,
+      "loss": 1.6953,
+      "step": 167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.994389360468073e-05,
+      "loss": 2.0898,
+      "step": 168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.994183522380784e-05,
+      "loss": 2.0059,
+      "step": 169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.993973978402085e-05,
+      "loss": 1.998,
+      "step": 170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99376072868747e-05,
+      "loss": 2.1152,
+      "step": 171
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.993543773395174e-05,
+      "loss": 2.0039,
+      "step": 172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99332311268619e-05,
+      "loss": 1.8105,
+      "step": 173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.993098746724259e-05,
+      "loss": 2.0469,
+      "step": 174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.992870675675866e-05,
+      "loss": 2.2676,
+      "step": 175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.992638899710252e-05,
+      "loss": 1.6582,
+      "step": 176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.992403418999403e-05,
+      "loss": 1.7148,
+      "step": 177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.992164233718055e-05,
+      "loss": 1.7988,
+      "step": 178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.991921344043696e-05,
+      "loss": 1.959,
+      "step": 179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.991674750156557e-05,
+      "loss": 2.1133,
+      "step": 180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.991424452239622e-05,
+      "loss": 1.7246,
+      "step": 181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.991170450478624e-05,
+      "loss": 1.8496,
+      "step": 182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99091274506204e-05,
+      "loss": 1.9121,
+      "step": 183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.9906513361811e-05,
+      "loss": 1.8047,
+      "step": 184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.990386224029782e-05,
+      "loss": 1.8359,
+      "step": 185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.990117408804807e-05,
+      "loss": 1.6738,
+      "step": 186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.989844890705649e-05,
+      "loss": 1.7676,
+      "step": 187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.989568669934526e-05,
+      "loss": 1.875,
+      "step": 188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.989288746696406e-05,
+      "loss": 1.373,
+      "step": 189
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.989005121199005e-05,
+      "loss": 1.4688,
+      "step": 190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.988717793652782e-05,
+      "loss": 1.4453,
+      "step": 191
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.988426764270949e-05,
+      "loss": 1.5801,
+      "step": 192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.98813203326946e-05,
+      "loss": 1.4453,
+      "step": 193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.98783360086702e-05,
+      "loss": 1.3984,
+      "step": 194
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.987531467285074e-05,
+      "loss": 1.5117,
+      "step": 195
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.987225632747821e-05,
+      "loss": 1.4316,
+      "step": 196
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.986916097482204e-05,
+      "loss": 1.5566,
+      "step": 197
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.986602861717907e-05,
+      "loss": 1.3047,
+      "step": 198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.986285925687366e-05,
+      "loss": 1.3398,
+      "step": 199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.985965289625763e-05,
+      "loss": 1.4062,
+      "step": 200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.985640953771018e-05,
+      "loss": 1.5215,
+      "step": 201
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.985312918363806e-05,
+      "loss": 1.5273,
+      "step": 202
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.984981183647542e-05,
+      "loss": 1.5781,
+      "step": 203
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.984645749868385e-05,
+      "loss": 1.4609,
+      "step": 204
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.984306617275243e-05,
+      "loss": 1.207,
+      "step": 205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.983963786119765e-05,
+      "loss": 1.4492,
+      "step": 206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.983617256656348e-05,
+      "loss": 1.3906,
+      "step": 207
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.983267029142129e-05,
+      "loss": 1.7832,
+      "step": 208
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.982913103836993e-05,
+      "loss": 1.6309,
+      "step": 209
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.982555481003566e-05,
+      "loss": 1.4199,
+      "step": 210
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.982194160907219e-05,
+      "loss": 1.6143,
+      "step": 211
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.981829143816067e-05,
+      "loss": 1.5059,
+      "step": 212
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.981460430000968e-05,
+      "loss": 1.5605,
+      "step": 213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.981088019735522e-05,
+      "loss": 1.4805,
+      "step": 214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.980711913296075e-05,
+      "loss": 1.2227,
+      "step": 215
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.980332110961712e-05,
+      "loss": 1.5176,
+      "step": 216
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.979948613014262e-05,
+      "loss": 1.6289,
+      "step": 217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.979561419738296e-05,
+      "loss": 1.3965,
+      "step": 218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.97917053142113e-05,
+      "loss": 1.4121,
+      "step": 219
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.978775948352815e-05,
+      "loss": 1.2812,
+      "step": 220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.978377670826153e-05,
+      "loss": 1.5312,
+      "step": 221
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.977975699136677e-05,
+      "loss": 1.2168,
+      "step": 222
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.977570033582672e-05,
+      "loss": 1.4805,
+      "step": 223
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.977160674465155e-05,
+      "loss": 1.3281,
+      "step": 224
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.97674762208789e-05,
+      "loss": 1.2754,
+      "step": 225
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.976330876757376e-05,
+      "loss": 1.4336,
+      "step": 226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.97591043878286e-05,
+      "loss": 1.373,
+      "step": 227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.97548630847632e-05,
+      "loss": 1.1455,
+      "step": 228
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.975058486152478e-05,
+      "loss": 1.3477,
+      "step": 229
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.9746269721288e-05,
+      "loss": 1.3828,
+      "step": 230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.974191766725485e-05,
+      "loss": 1.2441,
+      "step": 231
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.973752870265472e-05,
+      "loss": 1.4434,
+      "step": 232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.973310283074442e-05,
+      "loss": 1.6016,
+      "step": 233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.972864005480815e-05,
+      "loss": 1.2871,
+      "step": 234
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.972414037815743e-05,
+      "loss": 1.1426,
+      "step": 235
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.971960380413126e-05,
+      "loss": 1.2031,
+      "step": 236
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.97150303360959e-05,
+      "loss": 1.1543,
+      "step": 237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.971041997744512e-05,
+      "loss": 1.3379,
+      "step": 238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.970577273159996e-05,
+      "loss": 1.0518,
+      "step": 239
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.970108860200887e-05,
+      "loss": 1.4883,
+      "step": 240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.969636759214768e-05,
+      "loss": 1.3867,
+      "step": 241
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.969160970551957e-05,
+      "loss": 1.3828,
+      "step": 242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.968681494565509e-05,
+      "loss": 1.584,
+      "step": 243
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.968198331611217e-05,
+      "loss": 1.1562,
+      "step": 244
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.967711482047603e-05,
+      "loss": 1.5879,
+      "step": 245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.967220946235934e-05,
+      "loss": 1.2246,
+      "step": 246
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.966726724540206e-05,
+      "loss": 1.2715,
+      "step": 247
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.966228817327154e-05,
+      "loss": 1.4414,
+      "step": 248
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.965727224966244e-05,
+      "loss": 1.4277,
+      "step": 249
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.965221947829677e-05,
+      "loss": 1.4395,
+      "step": 250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.964712986292393e-05,
+      "loss": 1.377,
+      "step": 251
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.964200340732059e-05,
+      "loss": 1.2441,
+      "step": 252
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.963684011529084e-05,
+      "loss": 1.6387,
+      "step": 253
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.963163999066601e-05,
+      "loss": 1.1982,
+      "step": 254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.962640303730484e-05,
+      "loss": 1.2891,
+      "step": 255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.962112925909334e-05,
+      "loss": 1.2617,
+      "step": 256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.961581865994489e-05,
+      "loss": 1.4062,
+      "step": 257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.961047124380017e-05,
+      "loss": 1.3594,
+      "step": 258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.960508701462719e-05,
+      "loss": 1.2109,
+      "step": 259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.959966597642126e-05,
+      "loss": 1.3652,
+      "step": 260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.959420813320503e-05,
+      "loss": 1.1426,
+      "step": 261
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.958871348902843e-05,
+      "loss": 1.457,
+      "step": 262
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.958318204796874e-05,
+      "loss": 1.5605,
+      "step": 263
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.957761381413049e-05,
+      "loss": 1.2441,
+      "step": 264
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.957200879164554e-05,
+      "loss": 1.3164,
+      "step": 265
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.956636698467308e-05,
+      "loss": 1.3203,
+      "step": 266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.956068839739954e-05,
+      "loss": 1.2969,
+      "step": 267
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.955497303403867e-05,
+      "loss": 1.2041,
+      "step": 268
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.954922089883151e-05,
+      "loss": 1.3965,
+      "step": 269
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.954343199604639e-05,
+      "loss": 1.4199,
+      "step": 270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.95376063299789e-05,
+      "loss": 1.3223,
+      "step": 271
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.953174390495194e-05,
+      "loss": 1.2676,
+      "step": 272
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.952584472531568e-05,
+      "loss": 1.5605,
+      "step": 273
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.951990879544753e-05,
+      "loss": 1.291,
+      "step": 274
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.951393611975221e-05,
+      "loss": 1.3867,
+      "step": 275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.950792670266169e-05,
+      "loss": 1.2197,
+      "step": 276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.950188054863522e-05,
+      "loss": 1.1074,
+      "step": 277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.949579766215929e-05,
+      "loss": 1.3389,
+      "step": 278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.948967804774764e-05,
+      "loss": 1.3164,
+      "step": 279
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.948352170994129e-05,
+      "loss": 1.1553,
+      "step": 280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.94773286533085e-05,
+      "loss": 1.4785,
+      "step": 281
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.947109888244478e-05,
+      "loss": 1.3574,
+      "step": 282
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.946483240197287e-05,
+      "loss": 1.3125,
+      "step": 283
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.945852921654275e-05,
+      "loss": 1.4199,
+      "step": 284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.945218933083166e-05,
+      "loss": 1.3691,
+      "step": 285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.944581274954405e-05,
+      "loss": 1.2578,
+      "step": 286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.943939947741161e-05,
+      "loss": 1.0391,
+      "step": 287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.943294951919326e-05,
+      "loss": 1.1943,
+      "step": 288
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.942646287967515e-05,
+      "loss": 1.3848,
+      "step": 289
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.941993956367058e-05,
+      "loss": 1.4199,
+      "step": 290
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941337957602017e-05,
+      "loss": 1.3496,
+      "step": 291
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940678292159171e-05,
+      "loss": 1.209,
+      "step": 292
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.940014960528015e-05,
+      "loss": 1.1357,
+      "step": 293
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.939347963200772e-05,
+      "loss": 1.2988,
+      "step": 294
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.938677300672382e-05,
+      "loss": 1.2949,
+      "step": 295
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.938002973440502e-05,
+      "loss": 1.3691,
+      "step": 296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.937324982005509e-05,
+      "loss": 1.3613,
+      "step": 297
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.936643326870508e-05,
+      "loss": 1.2568,
+      "step": 298
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.935958008541308e-05,
+      "loss": 1.375,
+      "step": 299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.935269027526447e-05,
+      "loss": 1.2305,
+      "step": 300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.934576384337178e-05,
+      "loss": 1.2734,
+      "step": 301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.933880079487468e-05,
+      "loss": 1.2939,
+      "step": 302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.93318011349401e-05,
+      "loss": 1.1328,
+      "step": 303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.932476486876201e-05,
+      "loss": 1.1602,
+      "step": 304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.931769200156164e-05,
+      "loss": 1.2617,
+      "step": 305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.931058253858736e-05,
+      "loss": 1.3555,
+      "step": 306
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.930343648511469e-05,
+      "loss": 1.2617,
+      "step": 307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.929625384644627e-05,
+      "loss": 1.4531,
+      "step": 308
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.928903462791195e-05,
+      "loss": 1.1748,
+      "step": 309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.928177883486865e-05,
+      "loss": 1.1328,
+      "step": 310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.92744864727005e-05,
+      "loss": 1.207,
+      "step": 311
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.926715754681874e-05,
+      "loss": 1.2617,
+      "step": 312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.925979206266172e-05,
+      "loss": 1.2109,
+      "step": 313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.925239002569493e-05,
+      "loss": 1.2568,
+      "step": 314
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.924495144141102e-05,
+      "loss": 1.2461,
+      "step": 315
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.923747631532968e-05,
+      "loss": 1.1758,
+      "step": 316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.92299646529978e-05,
+      "loss": 1.2422,
+      "step": 317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.922241645998933e-05,
+      "loss": 1.2383,
+      "step": 318
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.921483174190536e-05,
+      "loss": 1.3828,
+      "step": 319
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.920721050437405e-05,
+      "loss": 1.3633,
+      "step": 320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.919955275305067e-05,
+      "loss": 1.7188,
+      "step": 321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.91918584936176e-05,
+      "loss": 1.2656,
+      "step": 322
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.91841277317843e-05,
+      "loss": 1.3027,
+      "step": 323
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.917636047328732e-05,
+      "loss": 1.293,
+      "step": 324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.916855672389026e-05,
+      "loss": 1.0117,
+      "step": 325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.916071648938387e-05,
+      "loss": 1.3203,
+      "step": 326
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.915283977558589e-05,
+      "loss": 1.2598,
+      "step": 327
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.91449265883412e-05,
+      "loss": 1.1455,
+      "step": 328
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.913697693352169e-05,
+      "loss": 1.168,
+      "step": 329
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.912899081702633e-05,
+      "loss": 1.1953,
+      "step": 330
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.912096824478116e-05,
+      "loss": 1.2666,
+      "step": 331
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.911290922273928e-05,
+      "loss": 1.2559,
+      "step": 332
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.910481375688081e-05,
+      "loss": 1.3496,
+      "step": 333
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.909668185321289e-05,
+      "loss": 1.4355,
+      "step": 334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.908851351776977e-05,
+      "loss": 1.2373,
+      "step": 335
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.908030875661266e-05,
+      "loss": 1.1172,
+      "step": 336
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.907206757582986e-05,
+      "loss": 1.3496,
+      "step": 337
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.906378998153666e-05,
+      "loss": 1.3848,
+      "step": 338
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.905547597987537e-05,
+      "loss": 1.5371,
+      "step": 339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.904712557701534e-05,
+      "loss": 1.2383,
+      "step": 340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.903873877915291e-05,
+      "loss": 1.1572,
+      "step": 341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.903031559251141e-05,
+      "loss": 1.1328,
+      "step": 342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.902185602334123e-05,
+      "loss": 1.1338,
+      "step": 343
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.90133600779197e-05,
+      "loss": 1.0635,
+      "step": 344
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.900482776255116e-05,
+      "loss": 1.3066,
+      "step": 345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.899625908356695e-05,
+      "loss": 1.2012,
+      "step": 346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.898765404732539e-05,
+      "loss": 1.2285,
+      "step": 347
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.897901266021179e-05,
+      "loss": 1.0059,
+      "step": 348
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.897033492863837e-05,
+      "loss": 1.2793,
+      "step": 349
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.89616208590444e-05,
+      "loss": 1.2119,
+      "step": 350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.895287045789607e-05,
+      "loss": 1.1123,
+      "step": 351
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.894408373168656e-05,
+      "loss": 1.2891,
+      "step": 352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.893526068693597e-05,
+      "loss": 1.2871,
+      "step": 353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.892640133019136e-05,
+      "loss": 1.0283,
+      "step": 354
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.891750566802675e-05,
+      "loss": 1.0234,
+      "step": 355
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.890857370704309e-05,
+      "loss": 1.2812,
+      "step": 356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.889960545386826e-05,
+      "loss": 1.1045,
+      "step": 357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.889060091515707e-05,
+      "loss": 1.2148,
+      "step": 358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.888156009759129e-05,
+      "loss": 1.457,
+      "step": 359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.887248300787954e-05,
+      "loss": 1.2344,
+      "step": 360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.886336965275744e-05,
+      "loss": 0.9482,
+      "step": 361
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.885422003898745e-05,
+      "loss": 1.2539,
+      "step": 362
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.884503417335897e-05,
+      "loss": 1.1035,
+      "step": 363
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.88358120626883e-05,
+      "loss": 1.1143,
+      "step": 364
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.882655371381862e-05,
+      "loss": 1.1562,
+      "step": 365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.881725913362001e-05,
+      "loss": 1.2402,
+      "step": 366
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.880792832898944e-05,
+      "loss": 1.3359,
+      "step": 367
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.879856130685074e-05,
+      "loss": 1.1416,
+      "step": 368
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.878915807415464e-05,
+      "loss": 0.9775,
+      "step": 369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.877971863787874e-05,
+      "loss": 1.0752,
+      "step": 370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.877024300502747e-05,
+      "loss": 1.332,
+      "step": 371
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.876073118263216e-05,
+      "loss": 1.0273,
+      "step": 372
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.875118317775097e-05,
+      "loss": 1.2559,
+      "step": 373
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.87415989974689e-05,
+      "loss": 1.2598,
+      "step": 374
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.873197864889786e-05,
+      "loss": 1.1348,
+      "step": 375
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.87223221391765e-05,
+      "loss": 0.8408,
+      "step": 376
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.871262947547035e-05,
+      "loss": 0.8271,
+      "step": 377
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.870290066497181e-05,
+      "loss": 0.8223,
+      "step": 378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.869313571490002e-05,
+      "loss": 0.8906,
+      "step": 379
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.8683334632501e-05,
+      "loss": 0.8486,
+      "step": 380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.867349742504754e-05,
+      "loss": 0.9004,
+      "step": 381
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.866362409983927e-05,
+      "loss": 0.7471,
+      "step": 382
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.86537146642026e-05,
+      "loss": 0.8037,
+      "step": 383
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.864376912549078e-05,
+      "loss": 0.6748,
+      "step": 384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.863378749108377e-05,
+      "loss": 0.75,
+      "step": 385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.862376976838835e-05,
+      "loss": 0.7461,
+      "step": 386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.861371596483813e-05,
+      "loss": 0.8594,
+      "step": 387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.86036260878934e-05,
+      "loss": 0.8428,
+      "step": 388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.859350014504129e-05,
+      "loss": 0.6963,
+      "step": 389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.858333814379568e-05,
+      "loss": 0.7939,
+      "step": 390
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.857314009169719e-05,
+      "loss": 0.7295,
+      "step": 391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.856290599631319e-05,
+      "loss": 0.7432,
+      "step": 392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.855263586523782e-05,
+      "loss": 0.707,
+      "step": 393
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.854232970609192e-05,
+      "loss": 0.6841,
+      "step": 394
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.853198752652309e-05,
+      "loss": 0.7295,
+      "step": 395
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.852160933420568e-05,
+      "loss": 0.8789,
+      "step": 396
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.851119513684072e-05,
+      "loss": 0.7109,
+      "step": 397
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.850074494215597e-05,
+      "loss": 0.8135,
+      "step": 398
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.849025875790594e-05,
+      "loss": 0.8682,
+      "step": 399
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.84797365918718e-05,
+      "loss": 0.7642,
+      "step": 400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.846917845186141e-05,
+      "loss": 0.7231,
+      "step": 401
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.845858434570936e-05,
+      "loss": 0.771,
+      "step": 402
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.844795428127694e-05,
+      "loss": 0.7178,
+      "step": 403
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.843728826645207e-05,
+      "loss": 0.9004,
+      "step": 404
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.842658630914938e-05,
+      "loss": 0.7061,
+      "step": 405
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.84158484173102e-05,
+      "loss": 0.9033,
+      "step": 406
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.840507459890243e-05,
+      "loss": 0.7656,
+      "step": 407
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.839426486192075e-05,
+      "loss": 0.6836,
+      "step": 408
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.83834192143864e-05,
+      "loss": 0.8135,
+      "step": 409
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.83725376643473e-05,
+      "loss": 0.6113,
+      "step": 410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.836162021987802e-05,
+      "loss": 0.8242,
+      "step": 411
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.835066688907977e-05,
+      "loss": 0.7568,
+      "step": 412
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.833967768008035e-05,
+      "loss": 0.8232,
+      "step": 413
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.832865260103423e-05,
+      "loss": 0.7549,
+      "step": 414
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.831759166012247e-05,
+      "loss": 0.9541,
+      "step": 415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.830649486555274e-05,
+      "loss": 0.6953,
+      "step": 416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.829536222555933e-05,
+      "loss": 0.7827,
+      "step": 417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.828419374840311e-05,
+      "loss": 1.0312,
+      "step": 418
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.827298944237158e-05,
+      "loss": 0.8037,
+      "step": 419
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.826174931577876e-05,
+      "loss": 0.874,
+      "step": 420
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.825047337696533e-05,
+      "loss": 0.8457,
+      "step": 421
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.823916163429847e-05,
+      "loss": 0.9072,
+      "step": 422
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.822781409617199e-05,
+      "loss": 0.667,
+      "step": 423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.821643077100622e-05,
+      "loss": 0.9316,
+      "step": 424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.820501166724806e-05,
+      "loss": 0.7432,
+      "step": 425
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.819355679337097e-05,
+      "loss": 0.748,
+      "step": 426
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.818206615787492e-05,
+      "loss": 0.7891,
+      "step": 427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.817053976928645e-05,
+      "loss": 0.8428,
+      "step": 428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.815897763615864e-05,
+      "loss": 0.7373,
+      "step": 429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.814737976707103e-05,
+      "loss": 0.793,
+      "step": 430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.813574617062975e-05,
+      "loss": 0.7822,
+      "step": 431
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.812407685546742e-05,
+      "loss": 0.7031,
+      "step": 432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.811237183024312e-05,
+      "loss": 0.8467,
+      "step": 433
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.810063110364249e-05,
+      "loss": 0.7627,
+      "step": 434
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.808885468437763e-05,
+      "loss": 0.644,
+      "step": 435
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.807704258118716e-05,
+      "loss": 0.9697,
+      "step": 436
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.80651948028361e-05,
+      "loss": 0.7842,
+      "step": 437
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.8053311358116e-05,
+      "loss": 0.8555,
+      "step": 438
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.80413922558449e-05,
+      "loss": 0.8799,
+      "step": 439
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.802943750486726e-05,
+      "loss": 0.752,
+      "step": 440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.801744711405399e-05,
+      "loss": 0.7686,
+      "step": 441
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.800542109230246e-05,
+      "loss": 0.8037,
+      "step": 442
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.799335944853648e-05,
+      "loss": 0.7324,
+      "step": 443
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.798126219170629e-05,
+      "loss": 0.9902,
+      "step": 444
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.796912933078856e-05,
+      "loss": 0.7031,
+      "step": 445
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.795696087478639e-05,
+      "loss": 0.7881,
+      "step": 446
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.794475683272923e-05,
+      "loss": 0.709,
+      "step": 447
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.793251721367305e-05,
+      "loss": 0.7188,
+      "step": 448
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.79202420267001e-05,
+      "loss": 0.9746,
+      "step": 449
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.79079312809191e-05,
+      "loss": 0.6787,
+      "step": 450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.789558498546515e-05,
+      "loss": 0.6069,
+      "step": 451
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.788320314949969e-05,
+      "loss": 0.875,
+      "step": 452
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.787078578221058e-05,
+      "loss": 0.8887,
+      "step": 453
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.785833289281199e-05,
+      "loss": 1.0088,
+      "step": 454
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.78458444905445e-05,
+      "loss": 0.8916,
+      "step": 455
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.783332058467503e-05,
+      "loss": 0.7295,
+      "step": 456
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.782076118449682e-05,
+      "loss": 0.9678,
+      "step": 457
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.780816629932947e-05,
+      "loss": 0.7744,
+      "step": 458
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.779553593851891e-05,
+      "loss": 0.8926,
+      "step": 459
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.77828701114374e-05,
+      "loss": 0.8604,
+      "step": 460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.777016882748351e-05,
+      "loss": 0.7812,
+      "step": 461
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.77574320960821e-05,
+      "loss": 0.7549,
+      "step": 462
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.774465992668438e-05,
+      "loss": 0.9014,
+      "step": 463
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.77318523287678e-05,
+      "loss": 0.6387,
+      "step": 464
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.771900931183615e-05,
+      "loss": 0.8242,
+      "step": 465
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.770613088541948e-05,
+      "loss": 0.8076,
+      "step": 466
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.76932170590741e-05,
+      "loss": 0.7656,
+      "step": 467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.768026784238263e-05,
+      "loss": 0.6709,
+      "step": 468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.766728324495391e-05,
+      "loss": 0.7959,
+      "step": 469
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.765426327642307e-05,
+      "loss": 0.7676,
+      "step": 470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.764120794645142e-05,
+      "loss": 0.8291,
+      "step": 471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.762811726472661e-05,
+      "loss": 0.7861,
+      "step": 472
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.761499124096244e-05,
+      "loss": 1.0459,
+      "step": 473
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.760182988489896e-05,
+      "loss": 0.8086,
+      "step": 474
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.758863320630247e-05,
+      "loss": 1.1641,
+      "step": 475
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.757540121496544e-05,
+      "loss": 0.7139,
+      "step": 476
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.756213392070653e-05,
+      "loss": 0.8809,
+      "step": 477
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.754883133337066e-05,
+      "loss": 0.9219,
+      "step": 478
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.753549346282889e-05,
+      "loss": 0.7681,
+      "step": 479
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.752212031897847e-05,
+      "loss": 0.8203,
+      "step": 480
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.750871191174283e-05,
+      "loss": 0.7373,
+      "step": 481
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.749526825107154e-05,
+      "loss": 0.8408,
+      "step": 482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.748178934694041e-05,
+      "loss": 0.5942,
+      "step": 483
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.746827520935129e-05,
+      "loss": 1.0381,
+      "step": 484
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.745472584833226e-05,
+      "loss": 0.7271,
+      "step": 485
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.744114127393751e-05,
+      "loss": 0.9756,
+      "step": 486
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.742752149624734e-05,
+      "loss": 0.96,
+      "step": 487
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.741386652536821e-05,
+      "loss": 0.7236,
+      "step": 488
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.740017637143266e-05,
+      "loss": 0.8057,
+      "step": 489
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.738645104459937e-05,
+      "loss": 0.707,
+      "step": 490
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.737269055505307e-05,
+      "loss": 0.8623,
+      "step": 491
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.735889491300465e-05,
+      "loss": 0.9814,
+      "step": 492
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.734506412869102e-05,
+      "loss": 0.8887,
+      "step": 493
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.733119821237522e-05,
+      "loss": 0.6919,
+      "step": 494
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.731729717434632e-05,
+      "loss": 0.8643,
+      "step": 495
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.730336102491947e-05,
+      "loss": 0.8599,
+      "step": 496
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.728938977443587e-05,
+      "loss": 0.7432,
+      "step": 497
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.727538343326278e-05,
+      "loss": 0.8311,
+      "step": 498
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.726134201179347e-05,
+      "loss": 0.9482,
+      "step": 499
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.724726552044725e-05,
+      "loss": 0.8945,
+      "step": 500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.72331539696695e-05,
+      "loss": 0.8682,
+      "step": 501
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.721900736993152e-05,
+      "loss": 0.835,
+      "step": 502
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.720482573173073e-05,
+      "loss": 0.9111,
+      "step": 503
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.719060906559047e-05,
+      "loss": 0.9053,
+      "step": 504
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.717635738206008e-05,
+      "loss": 0.7734,
+      "step": 505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.716207069171492e-05,
+      "loss": 0.8486,
+      "step": 506
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.71477490051563e-05,
+      "loss": 0.7666,
+      "step": 507
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.71333923330115e-05,
+      "loss": 0.999,
+      "step": 508
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.71190006859338e-05,
+      "loss": 0.7744,
+      "step": 509
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.710457407460236e-05,
+      "loss": 0.7031,
+      "step": 510
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.709011250972233e-05,
+      "loss": 0.915,
+      "step": 511
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.707561600202482e-05,
+      "loss": 0.665,
+      "step": 512
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.706108456226679e-05,
+      "loss": 0.8408,
+      "step": 513
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.704651820123123e-05,
+      "loss": 0.8906,
+      "step": 514
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.703191692972694e-05,
+      "loss": 0.8926,
+      "step": 515
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.70172807585887e-05,
+      "loss": 0.7344,
+      "step": 516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.700260969867712e-05,
+      "loss": 0.7607,
+      "step": 517
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.698790376087877e-05,
+      "loss": 0.7607,
+      "step": 518
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.697316295610603e-05,
+      "loss": 0.9727,
+      "step": 519
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.695838729529722e-05,
+      "loss": 0.834,
+      "step": 520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.69435767894165e-05,
+      "loss": 0.7871,
+      "step": 521
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.692873144945385e-05,
+      "loss": 0.7842,
+      "step": 522
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.691385128642517e-05,
+      "loss": 0.7959,
+      "step": 523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.68989363113721e-05,
+      "loss": 0.9199,
+      "step": 524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.688398653536222e-05,
+      "loss": 0.7236,
+      "step": 525
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.686900196948885e-05,
+      "loss": 0.8809,
+      "step": 526
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.685398262487118e-05,
+      "loss": 0.9219,
+      "step": 527
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.683892851265418e-05,
+      "loss": 0.8115,
+      "step": 528
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.682383964400862e-05,
+      "loss": 0.9248,
+      "step": 529
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.680871603013111e-05,
+      "loss": 0.627,
+      "step": 530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.679355768224392e-05,
+      "loss": 0.8184,
+      "step": 531
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.677836461159526e-05,
+      "loss": 0.7461,
+      "step": 532
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.676313682945895e-05,
+      "loss": 0.6445,
+      "step": 533
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.674787434713467e-05,
+      "loss": 0.8398,
+      "step": 534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.673257717594779e-05,
+      "loss": 0.6914,
+      "step": 535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.671724532724948e-05,
+      "loss": 0.7998,
+      "step": 536
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.670187881241659e-05,
+      "loss": 1.043,
+      "step": 537
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.66864776428517e-05,
+      "loss": 0.8105,
+      "step": 538
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.667104182998315e-05,
+      "loss": 0.7539,
+      "step": 539
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.665557138526492e-05,
+      "loss": 0.7871,
+      "step": 540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.664006632017674e-05,
+      "loss": 0.8877,
+      "step": 541
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.6624526646224e-05,
+      "loss": 0.7207,
+      "step": 542
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.660895237493779e-05,
+      "loss": 0.6895,
+      "step": 543
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.659334351787487e-05,
+      "loss": 1.0225,
+      "step": 544
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.657770008661766e-05,
+      "loss": 0.7363,
+      "step": 545
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.656202209277423e-05,
+      "loss": 1.041,
+      "step": 546
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.654630954797828e-05,
+      "loss": 0.9341,
+      "step": 547
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.653056246388922e-05,
+      "loss": 0.8203,
+      "step": 548
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.651478085219199e-05,
+      "loss": 0.8672,
+      "step": 549
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.649896472459724e-05,
+      "loss": 0.8232,
+      "step": 550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.648311409284116e-05,
+      "loss": 0.7822,
+      "step": 551
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.64672289686856e-05,
+      "loss": 0.7529,
+      "step": 552
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.645130936391799e-05,
+      "loss": 0.8496,
+      "step": 553
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.64353552903513e-05,
+      "loss": 0.8896,
+      "step": 554
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.641936675982414e-05,
+      "loss": 0.8418,
+      "step": 555
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.640334378420066e-05,
+      "loss": 0.8838,
+      "step": 556
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.638728637537057e-05,
+      "loss": 0.8486,
+      "step": 557
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.637119454524915e-05,
+      "loss": 0.9473,
+      "step": 558
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.635506830577717e-05,
+      "loss": 0.9463,
+      "step": 559
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.633890766892099e-05,
+      "loss": 1.0225,
+      "step": 560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.632271264667249e-05,
+      "loss": 0.7324,
+      "step": 561
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.630648325104902e-05,
+      "loss": 0.7275,
+      "step": 562
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.629021949409348e-05,
+      "loss": 0.8574,
+      "step": 563
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.627392138787423e-05,
+      "loss": 0.6865,
+      "step": 564
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.625758894448518e-05,
+      "loss": 0.6328,
+      "step": 565
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.624122217604566e-05,
+      "loss": 0.4521,
+      "step": 566
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.622482109470048e-05,
+      "loss": 0.5127,
+      "step": 567
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.620838571261994e-05,
+      "loss": 0.6631,
+      "step": 568
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.619191604199977e-05,
+      "loss": 0.478,
+      "step": 569
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.617541209506115e-05,
+      "loss": 0.4097,
+      "step": 570
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.615887388405072e-05,
+      "loss": 0.4937,
+      "step": 571
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.614230142124045e-05,
+      "loss": 0.4629,
+      "step": 572
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.612569471892786e-05,
+      "loss": 0.6055,
+      "step": 573
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.610905378943578e-05,
+      "loss": 0.4912,
+      "step": 574
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.609237864511248e-05,
+      "loss": 0.4736,
+      "step": 575
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.60756692983316e-05,
+      "loss": 0.3877,
+      "step": 576
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.60589257614922e-05,
+      "loss": 0.5269,
+      "step": 577
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.604214804701864e-05,
+      "loss": 0.4219,
+      "step": 578
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.60253361673607e-05,
+      "loss": 0.5479,
+      "step": 579
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.600849013499348e-05,
+      "loss": 0.3672,
+      "step": 580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.599160996241746e-05,
+      "loss": 0.4556,
+      "step": 581
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.59746956621584e-05,
+      "loss": 0.5088,
+      "step": 582
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.595774724676742e-05,
+      "loss": 0.4526,
+      "step": 583
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.594076472882096e-05,
+      "loss": 0.3977,
+      "step": 584
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.592374812092076e-05,
+      "loss": 0.5234,
+      "step": 585
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.590669743569382e-05,
+      "loss": 0.4141,
+      "step": 586
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.588961268579246e-05,
+      "loss": 0.4902,
+      "step": 587
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.58724938838943e-05,
+      "loss": 0.3667,
+      "step": 588
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.585534104270219e-05,
+      "loss": 0.5317,
+      "step": 589
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.583815417494423e-05,
+      "loss": 0.5034,
+      "step": 590
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.58209332933738e-05,
+      "loss": 0.4961,
+      "step": 591
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.580367841076953e-05,
+      "loss": 0.4556,
+      "step": 592
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.578638953993523e-05,
+      "loss": 0.3262,
+      "step": 593
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.576906669369997e-05,
+      "loss": 0.623,
+      "step": 594
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.575170988491805e-05,
+      "loss": 0.3716,
+      "step": 595
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.573431912646888e-05,
+      "loss": 0.439,
+      "step": 596
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.571689443125717e-05,
+      "loss": 0.4736,
+      "step": 597
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.569943581221274e-05,
+      "loss": 0.3237,
+      "step": 598
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.568194328229065e-05,
+      "loss": 0.416,
+      "step": 599
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.566441685447105e-05,
+      "loss": 0.4258,
+      "step": 600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.564685654175929e-05,
+      "loss": 0.4248,
+      "step": 601
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.562926235718584e-05,
+      "loss": 0.5312,
+      "step": 602
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.56116343138063e-05,
+      "loss": 0.498,
+      "step": 603
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.559397242470148e-05,
+      "loss": 0.585,
+      "step": 604
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.557627670297715e-05,
+      "loss": 0.5342,
+      "step": 605
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.555854716176431e-05,
+      "loss": 0.4497,
+      "step": 606
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.5540783814219e-05,
+      "loss": 0.4502,
+      "step": 607
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.55229866735224e-05,
+      "loss": 0.5605,
+      "step": 608
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.550515575288069e-05,
+      "loss": 0.5811,
+      "step": 609
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.548729106552515e-05,
+      "loss": 0.4102,
+      "step": 610
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.546939262471212e-05,
+      "loss": 0.4565,
+      "step": 611
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.545146044372302e-05,
+      "loss": 0.5181,
+      "step": 612
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.543349453586424e-05,
+      "loss": 0.5459,
+      "step": 613
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.541549491446724e-05,
+      "loss": 0.5303,
+      "step": 614
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.539746159288847e-05,
+      "loss": 0.5693,
+      "step": 615
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.537939458450943e-05,
+      "loss": 0.5049,
+      "step": 616
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.536129390273658e-05,
+      "loss": 0.4849,
+      "step": 617
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.534315956100136e-05,
+      "loss": 0.5278,
+      "step": 618
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.532499157276023e-05,
+      "loss": 0.7324,
+      "step": 619
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.530678995149457e-05,
+      "loss": 0.4995,
+      "step": 620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.528855471071074e-05,
+      "loss": 0.5703,
+      "step": 621
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.527028586394006e-05,
+      "loss": 0.5635,
+      "step": 622
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.525198342473878e-05,
+      "loss": 0.6294,
+      "step": 623
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.523364740668805e-05,
+      "loss": 0.4067,
+      "step": 624
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.521527782339395e-05,
+      "loss": 0.397,
+      "step": 625
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.519687468848748e-05,
+      "loss": 0.4712,
+      "step": 626
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.517843801562453e-05,
+      "loss": 0.5659,
+      "step": 627
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.51599678184859e-05,
+      "loss": 0.5342,
+      "step": 628
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.514146411077721e-05,
+      "loss": 0.5068,
+      "step": 629
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.512292690622899e-05,
+      "loss": 0.5493,
+      "step": 630
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.510435621859663e-05,
+      "loss": 0.6309,
+      "step": 631
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.508575206166033e-05,
+      "loss": 0.4512,
+      "step": 632
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.506711444922514e-05,
+      "loss": 0.543,
+      "step": 633
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.504844339512095e-05,
+      "loss": 0.7051,
+      "step": 634
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.502973891320248e-05,
+      "loss": 0.439,
+      "step": 635
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.501100101734921e-05,
+      "loss": 0.4985,
+      "step": 636
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.499222972146544e-05,
+      "loss": 0.624,
+      "step": 637
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.497342503948026e-05,
+      "loss": 0.6016,
+      "step": 638
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.495458698534751e-05,
+      "loss": 0.5352,
+      "step": 639
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.493571557304579e-05,
+      "loss": 0.3809,
+      "step": 640
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.491681081657852e-05,
+      "loss": 0.7158,
+      "step": 641
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.489787272997377e-05,
+      "loss": 0.5806,
+      "step": 642
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.487890132728444e-05,
+      "loss": 0.6211,
+      "step": 643
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.485989662258802e-05,
+      "loss": 0.5132,
+      "step": 644
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.484085862998683e-05,
+      "loss": 0.5781,
+      "step": 645
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.482178736360787e-05,
+      "loss": 0.4438,
+      "step": 646
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.480268283760279e-05,
+      "loss": 0.4971,
+      "step": 647
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.478354506614796e-05,
+      "loss": 0.5684,
+      "step": 648
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.476437406344437e-05,
+      "loss": 0.6162,
+      "step": 649
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.47451698437177e-05,
+      "loss": 0.5635,
+      "step": 650
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.47259324212183e-05,
+      "loss": 0.5815,
+      "step": 651
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.470666181022112e-05,
+      "loss": 0.5479,
+      "step": 652
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.468735802502577e-05,
+      "loss": 0.4858,
+      "step": 653
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.466802107995647e-05,
+      "loss": 0.5859,
+      "step": 654
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.464865098936199e-05,
+      "loss": 0.5225,
+      "step": 655
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.462924776761577e-05,
+      "loss": 0.6777,
+      "step": 656
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.460981142911581e-05,
+      "loss": 0.582,
+      "step": 657
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.459034198828467e-05,
+      "loss": 0.4888,
+      "step": 658
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.457083945956948e-05,
+      "loss": 0.5146,
+      "step": 659
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.455130385744193e-05,
+      "loss": 0.6279,
+      "step": 660
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.453173519639825e-05,
+      "loss": 0.6621,
+      "step": 661
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.45121334909592e-05,
+      "loss": 0.6074,
+      "step": 662
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 9.449249875567005e-05,
+      "loss": 0.6211,
+      "step": 663
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 9.447283100510059e-05,
+      "loss": 0.6069,
+      "step": 664
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.445313025384513e-05,
+      "loss": 0.4678,
+      "step": 665
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.443339651652241e-05,
+      "loss": 0.4961,
+      "step": 666
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 9.44136298077757e-05,
+      "loss": 0.668,
+      "step": 667
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 9.439383014227272e-05,
+      "loss": 0.7783,
+      "step": 668
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.437399753470562e-05,
+      "loss": 0.6465,
+      "step": 669
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.435413199979106e-05,
+      "loss": 0.5786,
+      "step": 670
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.433423355227005e-05,
+      "loss": 0.5371,
+      "step": 671
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.431430220690809e-05,
+      "loss": 0.5469,
+      "step": 672
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.429433797849507e-05,
+      "loss": 0.5527,
+      "step": 673
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.427434088184525e-05,
+      "loss": 0.5142,
+      "step": 674
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.425431093179731e-05,
+      "loss": 0.6963,
+      "step": 675
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.423424814321433e-05,
+      "loss": 0.4756,
+      "step": 676
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.421415253098371e-05,
+      "loss": 0.624,
+      "step": 677
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.419402411001722e-05,
+      "loss": 0.5605,
+      "step": 678
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.4173862895251e-05,
+      "loss": 0.4678,
+      "step": 679
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.415366890164552e-05,
+      "loss": 0.5415,
+      "step": 680
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.413344214418554e-05,
+      "loss": 0.3723,
+      "step": 681
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.411318263788015e-05,
+      "loss": 0.6582,
+      "step": 682
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.409289039776274e-05,
+      "loss": 0.6216,
+      "step": 683
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.407256543889104e-05,
+      "loss": 0.5083,
+      "step": 684
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.405220777634696e-05,
+      "loss": 0.5635,
+      "step": 685
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.403181742523675e-05,
+      "loss": 0.6914,
+      "step": 686
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.40113944006909e-05,
+      "loss": 0.7549,
+      "step": 687
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.399093871786413e-05,
+      "loss": 0.6777,
+      "step": 688
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.397045039193544e-05,
+      "loss": 0.5732,
+      "step": 689
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.394992943810797e-05,
+      "loss": 0.7314,
+      "step": 690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.392937587160915e-05,
+      "loss": 0.6816,
+      "step": 691
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.390878970769055e-05,
+      "loss": 0.791,
+      "step": 692
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.388817096162801e-05,
+      "loss": 0.5649,
+      "step": 693
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.386751964872145e-05,
+      "loss": 0.5771,
+      "step": 694
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.384683578429501e-05,
+      "loss": 0.6562,
+      "step": 695
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.382611938369697e-05,
+      "loss": 0.6235,
+      "step": 696
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.38053704622998e-05,
+      "loss": 0.52,
+      "step": 697
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.37845890355e-05,
+      "loss": 0.6929,
+      "step": 698
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.37637751187183e-05,
+      "loss": 0.6143,
+      "step": 699
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.374292872739945e-05,
+      "loss": 0.5791,
+      "step": 700
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.372204987701239e-05,
+      "loss": 0.4751,
+      "step": 701
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.370113858305003e-05,
+      "loss": 0.4951,
+      "step": 702
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.368019486102946e-05,
+      "loss": 0.54,
+      "step": 703
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.36592187264918e-05,
+      "loss": 0.6655,
+      "step": 704
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.363821019500218e-05,
+      "loss": 0.6245,
+      "step": 705
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.36171692821498e-05,
+      "loss": 0.6426,
+      "step": 706
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.359609600354793e-05,
+      "loss": 0.5527,
+      "step": 707
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.357499037483377e-05,
+      "loss": 0.5269,
+      "step": 708
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.355385241166861e-05,
+      "loss": 0.6768,
+      "step": 709
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.353268212973768e-05,
+      "loss": 0.4424,
+      "step": 710
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.351147954475019e-05,
+      "loss": 0.8438,
+      "step": 711
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.349024467243936e-05,
+      "loss": 0.5376,
+      "step": 712
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.346897752856235e-05,
+      "loss": 0.5815,
+      "step": 713
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.344767812890026e-05,
+      "loss": 0.6392,
+      "step": 714
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.342634648925813e-05,
+      "loss": 0.5811,
+      "step": 715
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.340498262546488e-05,
+      "loss": 0.6133,
+      "step": 716
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.338358655337344e-05,
+      "loss": 0.6519,
+      "step": 717
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.336215828886058e-05,
+      "loss": 0.5752,
+      "step": 718
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.334069784782693e-05,
+      "loss": 0.5903,
+      "step": 719
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.331920524619704e-05,
+      "loss": 0.5215,
+      "step": 720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.329768049991931e-05,
+      "loss": 0.4868,
+      "step": 721
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.327612362496601e-05,
+      "loss": 0.627,
+      "step": 722
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 9.325453463733321e-05,
+      "loss": 0.3403,
+      "step": 723
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 9.323291355304086e-05,
+      "loss": 0.5571,
+      "step": 724
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.321126038813269e-05,
+      "loss": 0.6504,
+      "step": 725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.318957515867625e-05,
+      "loss": 0.6992,
+      "step": 726
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 9.316785788076287e-05,
+      "loss": 0.6006,
+      "step": 727
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 9.314610857050768e-05,
+      "loss": 0.5332,
+      "step": 728
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.312432724404956e-05,
+      "loss": 0.5298,
+      "step": 729
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.310251391755119e-05,
+      "loss": 0.7031,
+      "step": 730
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.308066860719893e-05,
+      "loss": 0.707,
+      "step": 731
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.305879132920292e-05,
+      "loss": 0.5024,
+      "step": 732
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 9.303688209979702e-05,
+      "loss": 0.4321,
+      "step": 733
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 9.301494093523874e-05,
+      "loss": 0.627,
+      "step": 734
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.299296785180939e-05,
+      "loss": 0.5698,
+      "step": 735
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.29709628658139e-05,
+      "loss": 0.5967,
+      "step": 736
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.294892599358085e-05,
+      "loss": 0.5029,
+      "step": 737
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.292685725146254e-05,
+      "loss": 0.606,
+      "step": 738
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.290475665583489e-05,
+      "loss": 0.5718,
+      "step": 739
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.288262422309744e-05,
+      "loss": 0.6338,
+      "step": 740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.286045996967341e-05,
+      "loss": 0.5127,
+      "step": 741
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.283826391200956e-05,
+      "loss": 0.4873,
+      "step": 742
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.281603606657632e-05,
+      "loss": 0.7031,
+      "step": 743
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.279377644986763e-05,
+      "loss": 0.5664,
+      "step": 744
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.277148507840108e-05,
+      "loss": 0.6553,
+      "step": 745
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.274916196871779e-05,
+      "loss": 0.6387,
+      "step": 746
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.272680713738245e-05,
+      "loss": 0.5923,
+      "step": 747
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.270442060098323e-05,
+      "loss": 0.6504,
+      "step": 748
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.268200237613189e-05,
+      "loss": 0.6123,
+      "step": 749
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.265955247946369e-05,
+      "loss": 0.4492,
+      "step": 750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.263707092763737e-05,
+      "loss": 0.4502,
+      "step": 751
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.261455773733519e-05,
+      "loss": 0.3247,
+      "step": 752
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.259201292526287e-05,
+      "loss": 0.304,
+      "step": 753
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.256943650814957e-05,
+      "loss": 0.3396,
+      "step": 754
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.254682850274798e-05,
+      "loss": 0.3174,
+      "step": 755
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.252418892583415e-05,
+      "loss": 0.4292,
+      "step": 756
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.250151779420755e-05,
+      "loss": 0.4229,
+      "step": 757
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.247881512469115e-05,
+      "loss": 0.2568,
+      "step": 758
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.245608093413126e-05,
+      "loss": 0.3157,
+      "step": 759
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.243331523939757e-05,
+      "loss": 0.2727,
+      "step": 760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.241051805738321e-05,
+      "loss": 0.3096,
+      "step": 761
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.238768940500462e-05,
+      "loss": 0.3167,
+      "step": 762
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.236482929920159e-05,
+      "loss": 0.4053,
+      "step": 763
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.234193775693726e-05,
+      "loss": 0.4048,
+      "step": 764
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.231901479519816e-05,
+      "loss": 0.3164,
+      "step": 765
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.229606043099403e-05,
+      "loss": 0.2524,
+      "step": 766
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.227307468135796e-05,
+      "loss": 0.3403,
+      "step": 767
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.225005756334635e-05,
+      "loss": 0.2991,
+      "step": 768
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.222700909403885e-05,
+      "loss": 0.3599,
+      "step": 769
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.220392929053837e-05,
+      "loss": 0.2822,
+      "step": 770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.218081816997109e-05,
+      "loss": 0.2083,
+      "step": 771
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.215767574948644e-05,
+      "loss": 0.2498,
+      "step": 772
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.213450204625703e-05,
+      "loss": 0.2834,
+      "step": 773
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.211129707747872e-05,
+      "loss": 0.208,
+      "step": 774
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.208806086037055e-05,
+      "loss": 0.252,
+      "step": 775
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.206479341217478e-05,
+      "loss": 0.2725,
+      "step": 776
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.204149475015679e-05,
+      "loss": 0.2517,
+      "step": 777
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.201816489160517e-05,
+      "loss": 0.2295,
+      "step": 778
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.199480385383166e-05,
+      "loss": 0.3501,
+      "step": 779
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.19714116541711e-05,
+      "loss": 0.3386,
+      "step": 780
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.194798830998145e-05,
+      "loss": 0.344,
+      "step": 781
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.192453383864386e-05,
+      "loss": 0.2617,
+      "step": 782
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.190104825756248e-05,
+      "loss": 0.3428,
+      "step": 783
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.18775315841646e-05,
+      "loss": 0.2773,
+      "step": 784
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.185398383590055e-05,
+      "loss": 0.3105,
+      "step": 785
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.183040503024377e-05,
+      "loss": 0.2725,
+      "step": 786
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.180679518469069e-05,
+      "loss": 0.2737,
+      "step": 787
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.178315431676078e-05,
+      "loss": 0.1917,
+      "step": 788
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.175948244399656e-05,
+      "loss": 0.168,
+      "step": 789
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.173577958396357e-05,
+      "loss": 0.3213,
+      "step": 790
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.171204575425027e-05,
+      "loss": 0.312,
+      "step": 791
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.168828097246817e-05,
+      "loss": 0.2461,
+      "step": 792
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 9.166448525625173e-05,
+      "loss": 0.3442,
+      "step": 793
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 9.164065862325835e-05,
+      "loss": 0.4321,
+      "step": 794
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.161680109116837e-05,
+      "loss": 0.2869,
+      "step": 795
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.159291267768508e-05,
+      "loss": 0.2109,
+      "step": 796
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.156899340053469e-05,
+      "loss": 0.3276,
+      "step": 797
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.154504327746625e-05,
+      "loss": 0.2119,
+      "step": 798
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.152106232625179e-05,
+      "loss": 0.5078,
+      "step": 799
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.149705056468615e-05,
+      "loss": 0.321,
+      "step": 800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.147300801058706e-05,
+      "loss": 0.2957,
+      "step": 801
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.144893468179509e-05,
+      "loss": 0.4468,
+      "step": 802
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.142483059617367e-05,
+      "loss": 0.3433,
+      "step": 803
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.140069577160899e-05,
+      "loss": 0.2407,
+      "step": 804
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.137653022601011e-05,
+      "loss": 0.29,
+      "step": 805
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.135233397730888e-05,
+      "loss": 0.4092,
+      "step": 806
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.132810704345992e-05,
+      "loss": 0.3159,
+      "step": 807
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.13038494424406e-05,
+      "loss": 0.3911,
+      "step": 808
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.127956119225107e-05,
+      "loss": 0.4092,
+      "step": 809
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.125524231091423e-05,
+      "loss": 0.3682,
+      "step": 810
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.12308928164757e-05,
+      "loss": 0.4229,
+      "step": 811
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.120651272700382e-05,
+      "loss": 0.4141,
+      "step": 812
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.118210206058962e-05,
+      "loss": 0.261,
+      "step": 813
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.115766083534682e-05,
+      "loss": 0.3433,
+      "step": 814
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.113318906941183e-05,
+      "loss": 0.457,
+      "step": 815
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.110868678094372e-05,
+      "loss": 0.3354,
+      "step": 816
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.10841539881242e-05,
+      "loss": 0.4116,
+      "step": 817
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.105959070915761e-05,
+      "loss": 0.3716,
+      "step": 818
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.103499696227095e-05,
+      "loss": 0.4268,
+      "step": 819
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.101037276571377e-05,
+      "loss": 0.3755,
+      "step": 820
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.098571813775828e-05,
+      "loss": 0.4546,
+      "step": 821
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.096103309669921e-05,
+      "loss": 0.333,
+      "step": 822
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.093631766085389e-05,
+      "loss": 0.302,
+      "step": 823
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.09115718485622e-05,
+      "loss": 0.3574,
+      "step": 824
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 9.088679567818656e-05,
+      "loss": 0.4561,
+      "step": 825
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 9.08619891681119e-05,
+      "loss": 0.2754,
+      "step": 826
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.083715233674573e-05,
+      "loss": 0.3267,
+      "step": 827
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.081228520251794e-05,
+      "loss": 0.5059,
+      "step": 828
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 9.078738778388104e-05,
+      "loss": 0.2896,
+      "step": 829
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 9.076246009930991e-05,
+      "loss": 0.3877,
+      "step": 830
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.073750216730193e-05,
+      "loss": 0.2915,
+      "step": 831
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.071251400637695e-05,
+      "loss": 0.3267,
+      "step": 832
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.068749563507716e-05,
+      "loss": 0.4043,
+      "step": 833
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.06624470719673e-05,
+      "loss": 0.3691,
+      "step": 834
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.06373683356344e-05,
+      "loss": 0.6001,
+      "step": 835
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.061225944468795e-05,
+      "loss": 0.3582,
+      "step": 836
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.058712041775975e-05,
+      "loss": 0.3215,
+      "step": 837
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 9.056195127350404e-05,
+      "loss": 0.3801,
+      "step": 838
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 9.053675203059734e-05,
+      "loss": 0.3779,
+      "step": 839
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.051152270773857e-05,
+      "loss": 0.3352,
+      "step": 840
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.048626332364891e-05,
+      "loss": 0.313,
+      "step": 841
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 9.046097389707187e-05,
+      "loss": 0.3499,
+      "step": 842
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 9.043565444677326e-05,
+      "loss": 0.4187,
+      "step": 843
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.041030499154115e-05,
+      "loss": 0.3643,
+      "step": 844
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.038492555018591e-05,
+      "loss": 0.3984,
+      "step": 845
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.035951614154012e-05,
+      "loss": 0.4395,
+      "step": 846
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.033407678445863e-05,
+      "loss": 0.3289,
+      "step": 847
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.030860749781848e-05,
+      "loss": 0.3726,
+      "step": 848
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.028310830051896e-05,
+      "loss": 0.3496,
+      "step": 849
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.02575792114815e-05,
+      "loss": 0.4314,
+      "step": 850
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.023202024964979e-05,
+      "loss": 0.4355,
+      "step": 851
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.02064314339896e-05,
+      "loss": 0.3359,
+      "step": 852
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.018081278348892e-05,
+      "loss": 0.4102,
+      "step": 853
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.015516431715783e-05,
+      "loss": 0.4043,
+      "step": 854
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.012948605402857e-05,
+      "loss": 0.3484,
+      "step": 855
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.010377801315547e-05,
+      "loss": 0.5225,
+      "step": 856
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.007804021361496e-05,
+      "loss": 0.2924,
+      "step": 857
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.005227267450557e-05,
+      "loss": 0.3442,
+      "step": 858
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.002647541494788e-05,
+      "loss": 0.3115,
+      "step": 859
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.000064845408452e-05,
+      "loss": 0.4766,
+      "step": 860
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.997479181108017e-05,
+      "loss": 0.3804,
+      "step": 861
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.994890550512151e-05,
+      "loss": 0.3813,
+      "step": 862
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.992298955541729e-05,
+      "loss": 0.4028,
+      "step": 863
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.989704398119821e-05,
+      "loss": 0.2715,
+      "step": 864
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.987106880171694e-05,
+      "loss": 0.4297,
+      "step": 865
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.984506403624815e-05,
+      "loss": 0.416,
+      "step": 866
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.981902970408849e-05,
+      "loss": 0.4019,
+      "step": 867
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.979296582455648e-05,
+      "loss": 0.4375,
+      "step": 868
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 8.976687241699258e-05,
+      "loss": 0.3042,
+      "step": 869
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.974074950075923e-05,
+      "loss": 0.3735,
+      "step": 870
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 8.97145970952407e-05,
+      "loss": 0.2791,
+      "step": 871
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.968841521984315e-05,
+      "loss": 0.3391,
+      "step": 872
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 8.966220389399462e-05,
+      "loss": 0.3672,
+      "step": 873
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.963596313714501e-05,
+      "loss": 0.3438,
+      "step": 874
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 8.960969296876604e-05,
+      "loss": 0.3787,
+      "step": 875
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.958339340835128e-05,
+      "loss": 0.4424,
+      "step": 876
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 8.955706447541608e-05,
+      "loss": 0.2856,
+      "step": 877
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 8.953070618949762e-05,
+      "loss": 0.3364,
+      "step": 878
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.95043185701548e-05,
+      "loss": 0.3271,
+      "step": 879
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 8.947790163696839e-05,
+      "loss": 0.3916,
+      "step": 880
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.94514554095408e-05,
+      "loss": 0.4004,
+      "step": 881
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 8.942497990749624e-05,
+      "loss": 0.3477,
+      "step": 882
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.939847515048065e-05,
+      "loss": 0.3645,
+      "step": 883
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 8.937194115816165e-05,
+      "loss": 0.4014,
+      "step": 884
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.934537795022855e-05,
+      "loss": 0.3147,
+      "step": 885
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.931878554639237e-05,
+      "loss": 0.4746,
+      "step": 886
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.929216396638576e-05,
+      "loss": 0.3921,
+      "step": 887
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.926551322996305e-05,
+      "loss": 0.4146,
+      "step": 888
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.92388333569002e-05,
+      "loss": 0.5962,
+      "step": 889
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.921212436699476e-05,
+      "loss": 0.3701,
+      "step": 890
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.918538628006592e-05,
+      "loss": 0.4663,
+      "step": 891
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.915861911595446e-05,
+      "loss": 0.5571,
+      "step": 892
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.913182289452272e-05,
+      "loss": 0.5,
+      "step": 893
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.910499763565458e-05,
+      "loss": 0.3765,
+      "step": 894
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.907814335925555e-05,
+      "loss": 0.3751,
+      "step": 895
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.905126008525257e-05,
+      "loss": 0.3911,
+      "step": 896
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.902434783359417e-05,
+      "loss": 0.4292,
+      "step": 897
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.899740662425035e-05,
+      "loss": 0.4006,
+      "step": 898
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.897043647721259e-05,
+      "loss": 0.3135,
+      "step": 899
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.89434374124939e-05,
+      "loss": 0.5308,
+      "step": 900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.891640945012867e-05,
+      "loss": 0.5894,
+      "step": 901
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.888935261017279e-05,
+      "loss": 0.4702,
+      "step": 902
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.886226691270352e-05,
+      "loss": 0.5454,
+      "step": 903
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.883515237781963e-05,
+      "loss": 0.4536,
+      "step": 904
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.880800902564118e-05,
+      "loss": 0.4463,
+      "step": 905
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.878083687630969e-05,
+      "loss": 0.4243,
+      "step": 906
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.8753635949988e-05,
+      "loss": 0.3804,
+      "step": 907
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.872640626686037e-05,
+      "loss": 0.2542,
+      "step": 908
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.869914784713231e-05,
+      "loss": 0.52,
+      "step": 909
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.86718607110307e-05,
+      "loss": 0.4048,
+      "step": 910
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.864454487880377e-05,
+      "loss": 0.4873,
+      "step": 911
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.861720037072096e-05,
+      "loss": 0.3914,
+      "step": 912
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.858982720707305e-05,
+      "loss": 0.377,
+      "step": 913
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.856242540817204e-05,
+      "loss": 0.3511,
+      "step": 914
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.853499499435122e-05,
+      "loss": 0.4717,
+      "step": 915
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.850753598596511e-05,
+      "loss": 0.5068,
+      "step": 916
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.84800484033894e-05,
+      "loss": 0.541,
+      "step": 917
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.845253226702104e-05,
+      "loss": 0.3101,
+      "step": 918
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.842498759727815e-05,
+      "loss": 0.5708,
+      "step": 919
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.83974144146e-05,
+      "loss": 0.3091,
+      "step": 920
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.836981273944703e-05,
+      "loss": 0.4434,
+      "step": 921
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.834218259230083e-05,
+      "loss": 0.4878,
+      "step": 922
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.831452399366415e-05,
+      "loss": 0.4214,
+      "step": 923
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.828683696406079e-05,
+      "loss": 0.4519,
+      "step": 924
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.825912152403568e-05,
+      "loss": 0.3916,
+      "step": 925
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.823137769415482e-05,
+      "loss": 0.3787,
+      "step": 926
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.820360549500529e-05,
+      "loss": 0.4326,
+      "step": 927
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.81758049471952e-05,
+      "loss": 0.3252,
+      "step": 928
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.814797607135374e-05,
+      "loss": 0.4534,
+      "step": 929
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.812011888813106e-05,
+      "loss": 0.4497,
+      "step": 930
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.809223341819837e-05,
+      "loss": 0.3206,
+      "step": 931
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.806431968224784e-05,
+      "loss": 0.4697,
+      "step": 932
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.80363777009926e-05,
+      "loss": 0.2905,
+      "step": 933
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.800840749516679e-05,
+      "loss": 0.4287,
+      "step": 934
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.798040908552545e-05,
+      "loss": 0.3589,
+      "step": 935
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.795238249284454e-05,
+      "loss": 0.4038,
+      "step": 936
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.792432773792098e-05,
+      "loss": 0.416,
+      "step": 937
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.789624484157257e-05,
+      "loss": 0.4219,
+      "step": 938
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.786813382463796e-05,
+      "loss": 0.3467,
+      "step": 939
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.78399947079767e-05,
+      "loss": 0.304,
+      "step": 940
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.78118275124692e-05,
+      "loss": 0.241,
+      "step": 941
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.778363225901666e-05,
+      "loss": 0.2544,
+      "step": 942
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 8.775540896854115e-05,
+      "loss": 0.2598,
+      "step": 943
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 8.77271576619855e-05,
+      "loss": 0.1963,
+      "step": 944
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.769887836031337e-05,
+      "loss": 0.249,
+      "step": 945
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.767057108450918e-05,
+      "loss": 0.3047,
+      "step": 946
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 8.764223585557808e-05,
+      "loss": 0.3555,
+      "step": 947
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 8.761387269454601e-05,
+      "loss": 0.1692,
+      "step": 948
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 8.758548162245959e-05,
+      "loss": 0.3101,
+      "step": 949
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 8.755706266038619e-05,
+      "loss": 0.3826,
+      "step": 950
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 8.752861582941386e-05,
+      "loss": 0.1978,
+      "step": 951
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 8.750014115065133e-05,
+      "loss": 0.2471,
+      "step": 952
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 8.747163864522796e-05,
+      "loss": 0.2058,
+      "step": 953
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 8.744310833429384e-05,
+      "loss": 0.1614,
+      "step": 954
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 8.74145502390196e-05,
+      "loss": 0.2604,
+      "step": 955
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 8.738596438059658e-05,
+      "loss": 0.2144,
+      "step": 956
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 8.735735078023666e-05,
+      "loss": 0.2241,
+      "step": 957
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 8.732870945917228e-05,
+      "loss": 0.3318,
+      "step": 958
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 8.730004043865651e-05,
+      "loss": 0.269,
+      "step": 959
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 8.727134373996298e-05,
+      "loss": 0.2378,
+      "step": 960
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 8.724261938438581e-05,
+      "loss": 0.1733,
+      "step": 961
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 8.721386739323965e-05,
+      "loss": 0.1744,
+      "step": 962
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 8.718508778785968e-05,
+      "loss": 0.2808,
+      "step": 963
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 8.715628058960157e-05,
+      "loss": 0.2268,
+      "step": 964
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 8.712744581984145e-05,
+      "loss": 0.2456,
+      "step": 965
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 8.709858349997592e-05,
+      "loss": 0.2278,
+      "step": 966
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 8.706969365142202e-05,
+      "loss": 0.2788,
+      "step": 967
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 8.70407762956172e-05,
+      "loss": 0.3596,
+      "step": 968
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 8.701183145401936e-05,
+      "loss": 0.2327,
+      "step": 969
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 8.698285914810677e-05,
+      "loss": 0.2256,
+      "step": 970
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 8.695385939937806e-05,
+      "loss": 0.215,
+      "step": 971
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 8.692483222935228e-05,
+      "loss": 0.2168,
+      "step": 972
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 8.689577765956877e-05,
+      "loss": 0.2839,
+      "step": 973
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 8.686669571158724e-05,
+      "loss": 0.2454,
+      "step": 974
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 8.683758640698769e-05,
+      "loss": 0.3428,
+      "step": 975
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 8.680844976737044e-05,
+      "loss": 0.2524,
+      "step": 976
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 8.677928581435609e-05,
+      "loss": 0.2123,
+      "step": 977
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 8.675009456958549e-05,
+      "loss": 0.2314,
+      "step": 978
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.672087605471978e-05,
+      "loss": 0.2601,
+      "step": 979
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.669163029144028e-05,
+      "loss": 0.251,
+      "step": 980
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.666235730144858e-05,
+      "loss": 0.2517,
+      "step": 981
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.663305710646646e-05,
+      "loss": 0.2314,
+      "step": 982
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.660372972823583e-05,
+      "loss": 0.2328,
+      "step": 983
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.657437518851889e-05,
+      "loss": 0.2583,
+      "step": 984
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.654499350909788e-05,
+      "loss": 0.1689,
+      "step": 985
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.651558471177525e-05,
+      "loss": 0.2639,
+      "step": 986
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.648614881837353e-05,
+      "loss": 0.2606,
+      "step": 987
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 8.645668585073539e-05,
+      "loss": 0.3015,
+      "step": 988
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 8.642719583072356e-05,
+      "loss": 0.2812,
+      "step": 989
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 8.639767878022087e-05,
+      "loss": 0.2249,
+      "step": 990
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 8.63681347211302e-05,
+      "loss": 0.2422,
+      "step": 991
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 8.633856367537446e-05,
+      "loss": 0.2898,
+      "step": 992
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 8.630896566489659e-05,
+      "loss": 0.2424,
+      "step": 993
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 8.627934071165953e-05,
+      "loss": 0.152,
+      "step": 994
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 8.624968883764626e-05,
+      "loss": 0.2615,
+      "step": 995
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 8.622001006485967e-05,
+      "loss": 0.1906,
+      "step": 996
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 8.619030441532265e-05,
+      "loss": 0.2632,
+      "step": 997
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 8.616057191107804e-05,
+      "loss": 0.3289,
+      "step": 998
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 8.613081257418856e-05,
+      "loss": 0.1729,
+      "step": 999
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 8.610102642673689e-05,
+      "loss": 0.1376,
+      "step": 1000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 8.607121349082561e-05,
+      "loss": 0.2383,
+      "step": 1001
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 8.604137378857713e-05,
+      "loss": 0.2852,
+      "step": 1002
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 8.601150734213375e-05,
+      "loss": 0.3921,
+      "step": 1003
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 8.59816141736576e-05,
+      "loss": 0.3359,
+      "step": 1004
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.595169430533067e-05,
+      "loss": 0.2366,
+      "step": 1005
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 8.592174775935473e-05,
+      "loss": 0.1802,
+      "step": 1006
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.589177455795138e-05,
+      "loss": 0.2615,
+      "step": 1007
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 8.586177472336197e-05,
+      "loss": 0.3726,
+      "step": 1008
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 8.583174827784763e-05,
+      "loss": 0.3008,
+      "step": 1009
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 8.580169524368921e-05,
+      "loss": 0.2009,
+      "step": 1010
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.577161564318731e-05,
+      "loss": 0.1746,
+      "step": 1011
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 8.574150949866225e-05,
+      "loss": 0.2991,
+      "step": 1012
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 8.571137683245404e-05,
+      "loss": 0.3569,
+      "step": 1013
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 8.568121766692236e-05,
+      "loss": 0.1549,
+      "step": 1014
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.565103202444658e-05,
+      "loss": 0.2302,
+      "step": 1015
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 8.562081992742567e-05,
+      "loss": 0.2137,
+      "step": 1016
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 8.559058139827829e-05,
+      "loss": 0.3262,
+      "step": 1017
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 8.556031645944268e-05,
+      "loss": 0.2534,
+      "step": 1018
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 8.553002513337667e-05,
+      "loss": 0.2036,
+      "step": 1019
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 8.549970744255766e-05,
+      "loss": 0.1895,
+      "step": 1020
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 8.546936340948267e-05,
+      "loss": 0.2703,
+      "step": 1021
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.543899305666823e-05,
+      "loss": 0.2478,
+      "step": 1022
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 8.540859640665036e-05,
+      "loss": 0.1788,
+      "step": 1023
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.537817348198469e-05,
+      "loss": 0.2505,
+      "step": 1024
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 8.534772430524625e-05,
+      "loss": 0.188,
+      "step": 1025
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.531724889902964e-05,
+      "loss": 0.1472,
+      "step": 1026
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.528674728594882e-05,
+      "loss": 0.2844,
+      "step": 1027
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.525621948863729e-05,
+      "loss": 0.2627,
+      "step": 1028
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.522566552974793e-05,
+      "loss": 0.1666,
+      "step": 1029
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.519508543195304e-05,
+      "loss": 0.2112,
+      "step": 1030
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.516447921794435e-05,
+      "loss": 0.23,
+      "step": 1031
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.51338469104329e-05,
+      "loss": 0.163,
+      "step": 1032
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.510318853214916e-05,
+      "loss": 0.1912,
+      "step": 1033
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.507250410584293e-05,
+      "loss": 0.198,
+      "step": 1034
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.504179365428329e-05,
+      "loss": 0.1697,
+      "step": 1035
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.501105720025868e-05,
+      "loss": 0.3662,
+      "step": 1036
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.498029476657685e-05,
+      "loss": 0.2952,
+      "step": 1037
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.494950637606477e-05,
+      "loss": 0.2471,
+      "step": 1038
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.491869205156872e-05,
+      "loss": 0.2173,
+      "step": 1039
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.488785181595423e-05,
+      "loss": 0.2627,
+      "step": 1040
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.485698569210598e-05,
+      "loss": 0.2484,
+      "step": 1041
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.482609370292796e-05,
+      "loss": 0.2651,
+      "step": 1042
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.479517587134329e-05,
+      "loss": 0.3364,
+      "step": 1043
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.47642322202943e-05,
+      "loss": 0.217,
+      "step": 1044
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.473326277274243e-05,
+      "loss": 0.3027,
+      "step": 1045
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.470226755166832e-05,
+      "loss": 0.3474,
+      "step": 1046
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.467124658007171e-05,
+      "loss": 0.2404,
+      "step": 1047
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.464019988097145e-05,
+      "loss": 0.334,
+      "step": 1048
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.460912747740547e-05,
+      "loss": 0.3113,
+      "step": 1049
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.457802939243078e-05,
+      "loss": 0.3093,
+      "step": 1050
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.454690564912347e-05,
+      "loss": 0.2595,
+      "step": 1051
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.45157562705786e-05,
+      "loss": 0.3169,
+      "step": 1052
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.448458127991037e-05,
+      "loss": 0.2734,
+      "step": 1053
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.445338070025187e-05,
+      "loss": 0.3872,
+      "step": 1054
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.442215455475524e-05,
+      "loss": 0.3066,
+      "step": 1055
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.439090286659156e-05,
+      "loss": 0.3491,
+      "step": 1056
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.43596256589509e-05,
+      "loss": 0.2996,
+      "step": 1057
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.432832295504224e-05,
+      "loss": 0.2441,
+      "step": 1058
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.429699477809347e-05,
+      "loss": 0.2606,
+      "step": 1059
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.426564115135143e-05,
+      "loss": 0.2776,
+      "step": 1060
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.423426209808178e-05,
+      "loss": 0.3857,
+      "step": 1061
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.42028576415691e-05,
+      "loss": 0.269,
+      "step": 1062
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.41714278051168e-05,
+      "loss": 0.3926,
+      "step": 1063
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.41399726120471e-05,
+      "loss": 0.3413,
+      "step": 1064
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.410849208570109e-05,
+      "loss": 0.3394,
+      "step": 1065
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.407698624943859e-05,
+      "loss": 0.2727,
+      "step": 1066
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.404545512663827e-05,
+      "loss": 0.2163,
+      "step": 1067
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.401389874069749e-05,
+      "loss": 0.2244,
+      "step": 1068
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.398231711503246e-05,
+      "loss": 0.3457,
+      "step": 1069
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.395071027307803e-05,
+      "loss": 0.292,
+      "step": 1070
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.391907823828776e-05,
+      "loss": 0.386,
+      "step": 1071
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.388742103413397e-05,
+      "loss": 0.2397,
+      "step": 1072
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.38557386841076e-05,
+      "loss": 0.2417,
+      "step": 1073
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.382403121171827e-05,
+      "loss": 0.2932,
+      "step": 1074
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.379229864049424e-05,
+      "loss": 0.3296,
+      "step": 1075
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.376054099398241e-05,
+      "loss": 0.3545,
+      "step": 1076
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.372875829574825e-05,
+      "loss": 0.2449,
+      "step": 1077
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.369695056937584e-05,
+      "loss": 0.3206,
+      "step": 1078
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.366511783846784e-05,
+      "loss": 0.3765,
+      "step": 1079
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.363326012664549e-05,
+      "loss": 0.2927,
+      "step": 1080
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.360137745754848e-05,
+      "loss": 0.2495,
+      "step": 1081
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.356946985483509e-05,
+      "loss": 0.2439,
+      "step": 1082
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.35375373421821e-05,
+      "loss": 0.2815,
+      "step": 1083
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.350557994328477e-05,
+      "loss": 0.1901,
+      "step": 1084
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.347359768185677e-05,
+      "loss": 0.384,
+      "step": 1085
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 8.344159058163031e-05,
+      "loss": 0.2421,
+      "step": 1086
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 8.340955866635597e-05,
+      "loss": 0.2029,
+      "step": 1087
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 8.337750195980276e-05,
+      "loss": 0.3123,
+      "step": 1088
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 8.334542048575808e-05,
+      "loss": 0.3394,
+      "step": 1089
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 8.331331426802773e-05,
+      "loss": 0.2961,
+      "step": 1090
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 8.328118333043584e-05,
+      "loss": 0.2579,
+      "step": 1091
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 8.32490276968249e-05,
+      "loss": 0.3569,
+      "step": 1092
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 8.321684739105573e-05,
+      "loss": 0.4189,
+      "step": 1093
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 8.318464243700748e-05,
+      "loss": 0.2676,
+      "step": 1094
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 8.31524128585775e-05,
+      "loss": 0.3037,
+      "step": 1095
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 8.312015867968153e-05,
+      "loss": 0.2126,
+      "step": 1096
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.308787992425348e-05,
+      "loss": 0.2711,
+      "step": 1097
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 8.305557661624554e-05,
+      "loss": 0.3833,
+      "step": 1098
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 8.302324877962809e-05,
+      "loss": 0.2727,
+      "step": 1099
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 8.299089643838975e-05,
+      "loss": 0.2227,
+      "step": 1100
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 8.295851961653727e-05,
+      "loss": 0.276,
+      "step": 1101
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 8.292611833809563e-05,
+      "loss": 0.2354,
+      "step": 1102
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 8.289369262710788e-05,
+      "loss": 0.188,
+      "step": 1103
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 8.286124250763529e-05,
+      "loss": 0.4031,
+      "step": 1104
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 8.282876800375716e-05,
+      "loss": 0.3308,
+      "step": 1105
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 8.279626913957095e-05,
+      "loss": 0.3948,
+      "step": 1106
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 8.276374593919212e-05,
+      "loss": 0.1921,
+      "step": 1107
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 8.273119842675428e-05,
+      "loss": 0.269,
+      "step": 1108
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.2698626626409e-05,
+      "loss": 0.2229,
+      "step": 1109
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.266603056232594e-05,
+      "loss": 0.1918,
+      "step": 1110
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 8.26334102586927e-05,
+      "loss": 0.197,
+      "step": 1111
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 8.260076573971494e-05,
+      "loss": 0.2837,
+      "step": 1112
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.25680970296162e-05,
+      "loss": 0.2227,
+      "step": 1113
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 8.253540415263805e-05,
+      "loss": 0.4258,
+      "step": 1114
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 8.250268713303995e-05,
+      "loss": 0.3289,
+      "step": 1115
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 8.246994599509928e-05,
+      "loss": 0.3115,
+      "step": 1116
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 8.243718076311134e-05,
+      "loss": 0.3455,
+      "step": 1117
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 8.240439146138928e-05,
+      "loss": 0.3306,
+      "step": 1118
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 8.237157811426413e-05,
+      "loss": 0.2668,
+      "step": 1119
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 8.233874074608475e-05,
+      "loss": 0.3213,
+      "step": 1120
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 8.230587938121783e-05,
+      "loss": 0.3286,
+      "step": 1121
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 8.227299404404787e-05,
+      "loss": 0.2998,
+      "step": 1122
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 8.224008475897715e-05,
+      "loss": 0.2344,
+      "step": 1123
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.220715155042573e-05,
+      "loss": 0.3549,
+      "step": 1124
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.217419444283141e-05,
+      "loss": 0.3167,
+      "step": 1125
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 8.214121346064974e-05,
+      "loss": 0.2588,
+      "step": 1126
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 8.210820862835395e-05,
+      "loss": 0.3242,
+      "step": 1127
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 8.207517997043504e-05,
+      "loss": 0.2178,
+      "step": 1128
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 8.204212751140162e-05,
+      "loss": 0.2178,
+      "step": 1129
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 8.200905127578e-05,
+      "loss": 0.1823,
+      "step": 1130
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 8.197595128811408e-05,
+      "loss": 0.1255,
+      "step": 1131
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 8.194282757296547e-05,
+      "loss": 0.1913,
+      "step": 1132
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 8.190968015491334e-05,
+      "loss": 0.2417,
+      "step": 1133
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 8.18765090585544e-05,
+      "loss": 0.2185,
+      "step": 1134
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 8.184331430850302e-05,
+      "loss": 0.2041,
+      "step": 1135
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 8.181009592939109e-05,
+      "loss": 0.135,
+      "step": 1136
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 8.177685394586801e-05,
+      "loss": 0.1934,
+      "step": 1137
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 8.17435883826007e-05,
+      "loss": 0.2094,
+      "step": 1138
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 8.17102992642736e-05,
+      "loss": 0.13,
+      "step": 1139
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 8.167698661558865e-05,
+      "loss": 0.1488,
+      "step": 1140
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 8.164365046126514e-05,
+      "loss": 0.1991,
+      "step": 1141
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 8.161029082603995e-05,
+      "loss": 0.217,
+      "step": 1142
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 8.157690773466728e-05,
+      "loss": 0.1293,
+      "step": 1143
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 8.154350121191879e-05,
+      "loss": 0.1671,
+      "step": 1144
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 8.151007128258348e-05,
+      "loss": 0.1206,
+      "step": 1145
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 8.147661797146774e-05,
+      "loss": 0.1776,
+      "step": 1146
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 8.144314130339536e-05,
+      "loss": 0.1185,
+      "step": 1147
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 8.14096413032074e-05,
+      "loss": 0.2057,
+      "step": 1148
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 8.137611799576222e-05,
+      "loss": 0.2107,
+      "step": 1149
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 8.134257140593553e-05,
+      "loss": 0.2246,
+      "step": 1150
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 8.130900155862029e-05,
+      "loss": 0.1777,
+      "step": 1151
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 8.127540847872672e-05,
+      "loss": 0.1514,
+      "step": 1152
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 8.124179219118226e-05,
+      "loss": 0.2458,
+      "step": 1153
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 8.120815272093163e-05,
+      "loss": 0.1938,
+      "step": 1154
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 8.117449009293668e-05,
+      "loss": 0.2686,
+      "step": 1155
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 8.114080433217649e-05,
+      "loss": 0.152,
+      "step": 1156
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 8.110709546364728e-05,
+      "loss": 0.1659,
+      "step": 1157
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 8.107336351236244e-05,
+      "loss": 0.1826,
+      "step": 1158
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 8.103960850335248e-05,
+      "loss": 0.2598,
+      "step": 1159
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 8.100583046166501e-05,
+      "loss": 0.2288,
+      "step": 1160
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 8.097202941236473e-05,
+      "loss": 0.2307,
+      "step": 1161
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 8.093820538053345e-05,
+      "loss": 0.1607,
+      "step": 1162
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 8.090435839126997e-05,
+      "loss": 0.1426,
+      "step": 1163
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 8.087048846969018e-05,
+      "loss": 0.1498,
+      "step": 1164
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 8.0836595640927e-05,
+      "loss": 0.1091,
+      "step": 1165
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 8.080267993013024e-05,
+      "loss": 0.1486,
+      "step": 1166
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 8.076874136246681e-05,
+      "loss": 0.1194,
+      "step": 1167
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 8.073477996312054e-05,
+      "loss": 0.25,
+      "step": 1168
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 8.07007957572922e-05,
+      "loss": 0.156,
+      "step": 1169
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 8.066678877019945e-05,
+      "loss": 0.111,
+      "step": 1170
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 8.063275902707692e-05,
+      "loss": 0.2234,
+      "step": 1171
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 8.059870655317606e-05,
+      "loss": 0.1616,
+      "step": 1172
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 8.056463137376524e-05,
+      "loss": 0.1689,
+      "step": 1173
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 8.053053351412967e-05,
+      "loss": 0.2344,
+      "step": 1174
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 8.049641299957133e-05,
+      "loss": 0.1525,
+      "step": 1175
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 8.046226985540911e-05,
+      "loss": 0.1467,
+      "step": 1176
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 8.042810410697861e-05,
+      "loss": 0.2178,
+      "step": 1177
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 8.039391577963222e-05,
+      "loss": 0.1278,
+      "step": 1178
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 8.035970489873915e-05,
+      "loss": 0.1493,
+      "step": 1179
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 8.032547148968524e-05,
+      "loss": 0.1242,
+      "step": 1180
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 8.029121557787312e-05,
+      "loss": 0.2301,
+      "step": 1181
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 8.02569371887221e-05,
+      "loss": 0.1311,
+      "step": 1182
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 8.022263634766818e-05,
+      "loss": 0.2452,
+      "step": 1183
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 8.018831308016397e-05,
+      "loss": 0.1984,
+      "step": 1184
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 8.015396741167878e-05,
+      "loss": 0.1534,
+      "step": 1185
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 8.011959936769852e-05,
+      "loss": 0.1184,
+      "step": 1186
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 8.008520897372574e-05,
+      "loss": 0.1418,
+      "step": 1187
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 8.005079625527949e-05,
+      "loss": 0.1348,
+      "step": 1188
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 8.001636123789545e-05,
+      "loss": 0.186,
+      "step": 1189
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 7.998190394712583e-05,
+      "loss": 0.1766,
+      "step": 1190
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.994742440853937e-05,
+      "loss": 0.201,
+      "step": 1191
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 7.991292264772132e-05,
+      "loss": 0.2444,
+      "step": 1192
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.987839869027343e-05,
+      "loss": 0.1631,
+      "step": 1193
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 7.98438525618139e-05,
+      "loss": 0.1912,
+      "step": 1194
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.980928428797738e-05,
+      "loss": 0.1278,
+      "step": 1195
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 7.977469389441498e-05,
+      "loss": 0.2114,
+      "step": 1196
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.97400814067942e-05,
+      "loss": 0.1484,
+      "step": 1197
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 7.970544685079895e-05,
+      "loss": 0.2224,
+      "step": 1198
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.967079025212948e-05,
+      "loss": 0.1442,
+      "step": 1199
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 7.963611163650246e-05,
+      "loss": 0.1851,
+      "step": 1200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.960141102965084e-05,
+      "loss": 0.1817,
+      "step": 1201
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 7.956668845732391e-05,
+      "loss": 0.1091,
+      "step": 1202
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.953194394528729e-05,
+      "loss": 0.1283,
+      "step": 1203
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 7.94971775193228e-05,
+      "loss": 0.2688,
+      "step": 1204
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.94623892052286e-05,
+      "loss": 0.1412,
+      "step": 1205
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 7.942757902881907e-05,
+      "loss": 0.2866,
+      "step": 1206
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 7.939274701592479e-05,
+      "loss": 0.1799,
+      "step": 1207
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.935789319239257e-05,
+      "loss": 0.2473,
+      "step": 1208
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 7.932301758408538e-05,
+      "loss": 0.2659,
+      "step": 1209
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.928812021688236e-05,
+      "loss": 0.1575,
+      "step": 1210
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 7.925320111667885e-05,
+      "loss": 0.2077,
+      "step": 1211
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.921826030938621e-05,
+      "loss": 0.2486,
+      "step": 1212
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 7.918329782093201e-05,
+      "loss": 0.1836,
+      "step": 1213
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.914831367725988e-05,
+      "loss": 0.2231,
+      "step": 1214
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 7.911330790432946e-05,
+      "loss": 0.2217,
+      "step": 1215
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.907828052811653e-05,
+      "loss": 0.2178,
+      "step": 1216
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 7.904323157461279e-05,
+      "loss": 0.1802,
+      "step": 1217
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.900816106982611e-05,
+      "loss": 0.3208,
+      "step": 1218
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 7.89730690397802e-05,
+      "loss": 0.1206,
+      "step": 1219
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.893795551051482e-05,
+      "loss": 0.2476,
+      "step": 1220
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 7.890282050808564e-05,
+      "loss": 0.1985,
+      "step": 1221
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 7.886766405856434e-05,
+      "loss": 0.2808,
+      "step": 1222
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.883248618803842e-05,
+      "loss": 0.2109,
+      "step": 1223
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 7.879728692261135e-05,
+      "loss": 0.2239,
+      "step": 1224
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.876206628840241e-05,
+      "loss": 0.2039,
+      "step": 1225
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 7.872682431154681e-05,
+      "loss": 0.2368,
+      "step": 1226
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.869156101819555e-05,
+      "loss": 0.2144,
+      "step": 1227
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 7.865627643451543e-05,
+      "loss": 0.2118,
+      "step": 1228
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.86209705866891e-05,
+      "loss": 0.2417,
+      "step": 1229
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 7.858564350091497e-05,
+      "loss": 0.1893,
+      "step": 1230
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.855029520340718e-05,
+      "loss": 0.3621,
+      "step": 1231
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 7.851492572039564e-05,
+      "loss": 0.1755,
+      "step": 1232
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.8479535078126e-05,
+      "loss": 0.238,
+      "step": 1233
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 7.844412330285955e-05,
+      "loss": 0.1746,
+      "step": 1234
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.84086904208733e-05,
+      "loss": 0.1516,
+      "step": 1235
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 7.837323645845995e-05,
+      "loss": 0.2043,
+      "step": 1236
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 7.83377614419278e-05,
+      "loss": 0.1798,
+      "step": 1237
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.830226539760077e-05,
+      "loss": 0.2605,
+      "step": 1238
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 7.826674835181839e-05,
+      "loss": 0.2954,
+      "step": 1239
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.823121033093582e-05,
+      "loss": 0.1781,
+      "step": 1240
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 7.81956513613237e-05,
+      "loss": 0.1727,
+      "step": 1241
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.816007146936828e-05,
+      "loss": 0.1531,
+      "step": 1242
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 7.812447068147132e-05,
+      "loss": 0.1594,
+      "step": 1243
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.808884902405011e-05,
+      "loss": 0.2605,
+      "step": 1244
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 7.805320652353733e-05,
+      "loss": 0.1973,
+      "step": 1245
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.801754320638127e-05,
+      "loss": 0.197,
+      "step": 1246
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 7.798185909904552e-05,
+      "loss": 0.1763,
+      "step": 1247
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.794615422800923e-05,
+      "loss": 0.1672,
+      "step": 1248
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 7.791042861976685e-05,
+      "loss": 0.2122,
+      "step": 1249
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.78746823008283e-05,
+      "loss": 0.2007,
+      "step": 1250
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 7.783891529771881e-05,
+      "loss": 0.2667,
+      "step": 1251
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.780312763697896e-05,
+      "loss": 0.231,
+      "step": 1252
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 7.776731934516472e-05,
+      "loss": 0.2631,
+      "step": 1253
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 7.773149044884732e-05,
+      "loss": 0.1506,
+      "step": 1254
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.769564097461328e-05,
+      "loss": 0.2793,
+      "step": 1255
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 7.765977094906439e-05,
+      "loss": 0.203,
+      "step": 1256
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.762388039881772e-05,
+      "loss": 0.2,
+      "step": 1257
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 7.758796935050552e-05,
+      "loss": 0.1436,
+      "step": 1258
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.755203783077529e-05,
+      "loss": 0.1917,
+      "step": 1259
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 7.751608586628971e-05,
+      "loss": 0.3213,
+      "step": 1260
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.748011348372663e-05,
+      "loss": 0.175,
+      "step": 1261
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 7.744412070977907e-05,
+      "loss": 0.1252,
+      "step": 1262
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.740810757115511e-05,
+      "loss": 0.2068,
+      "step": 1263
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 7.737207409457802e-05,
+      "loss": 0.1567,
+      "step": 1264
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.733602030678617e-05,
+      "loss": 0.1669,
+      "step": 1265
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 7.729994623453293e-05,
+      "loss": 0.228,
+      "step": 1266
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.726385190458675e-05,
+      "loss": 0.1222,
+      "step": 1267
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 7.722773734373114e-05,
+      "loss": 0.171,
+      "step": 1268
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 7.719160257876461e-05,
+      "loss": 0.2678,
+      "step": 1269
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.715544763650062e-05,
+      "loss": 0.2739,
+      "step": 1270
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 7.711927254376767e-05,
+      "loss": 0.1805,
+      "step": 1271
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.708307732740919e-05,
+      "loss": 0.2081,
+      "step": 1272
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 7.704686201428349e-05,
+      "loss": 0.2166,
+      "step": 1273
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.701062663126386e-05,
+      "loss": 0.241,
+      "step": 1274
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 7.697437120523845e-05,
+      "loss": 0.2068,
+      "step": 1275
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.69380957631103e-05,
+      "loss": 0.1724,
+      "step": 1276
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 7.69018003317973e-05,
+      "loss": 0.2634,
+      "step": 1277
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.686548493823215e-05,
+      "loss": 0.196,
+      "step": 1278
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 7.682914960936238e-05,
+      "loss": 0.1324,
+      "step": 1279
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 7.679279437215033e-05,
+      "loss": 0.1912,
+      "step": 1280
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 7.675641925357307e-05,
+      "loss": 0.1888,
+      "step": 1281
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 7.672002428062245e-05,
+      "loss": 0.2358,
+      "step": 1282
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 7.668360948030508e-05,
+      "loss": 0.1836,
+      "step": 1283
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 7.664717487964222e-05,
+      "loss": 0.3035,
+      "step": 1284
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 7.661072050566986e-05,
+      "loss": 0.1017,
+      "step": 1285
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 7.657424638543867e-05,
+      "loss": 0.2213,
+      "step": 1286
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 7.653775254601395e-05,
+      "loss": 0.2288,
+      "step": 1287
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 7.650123901447566e-05,
+      "loss": 0.1931,
+      "step": 1288
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 7.646470581791833e-05,
+      "loss": 0.1497,
+      "step": 1289
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 7.642815298345114e-05,
+      "loss": 0.1954,
+      "step": 1290
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 7.639158053819778e-05,
+      "loss": 0.1799,
+      "step": 1291
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 7.635498850929653e-05,
+      "loss": 0.1729,
+      "step": 1292
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 7.631837692390021e-05,
+      "loss": 0.2642,
+      "step": 1293
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 7.628174580917612e-05,
+      "loss": 0.1731,
+      "step": 1294
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 7.624509519230608e-05,
+      "loss": 0.3219,
+      "step": 1295
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 7.620842510048633e-05,
+      "loss": 0.2783,
+      "step": 1296
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 7.617173556092765e-05,
+      "loss": 0.2688,
+      "step": 1297
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 7.613502660085517e-05,
+      "loss": 0.2375,
+      "step": 1298
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 7.609829824750849e-05,
+      "loss": 0.1462,
+      "step": 1299
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 7.606155052814152e-05,
+      "loss": 0.2097,
+      "step": 1300
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 7.602478347002267e-05,
+      "loss": 0.2773,
+      "step": 1301
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 7.598799710043454e-05,
+      "loss": 0.2817,
+      "step": 1302
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 7.59511914466742e-05,
+      "loss": 0.2085,
+      "step": 1303
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 7.591436653605297e-05,
+      "loss": 0.2421,
+      "step": 1304
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 7.587752239589643e-05,
+      "loss": 0.3281,
+      "step": 1305
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 7.584065905354448e-05,
+      "loss": 0.2303,
+      "step": 1306
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 7.580377653635122e-05,
+      "loss": 0.1575,
+      "step": 1307
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 7.576687487168507e-05,
+      "loss": 0.1974,
+      "step": 1308
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 7.572995408692853e-05,
+      "loss": 0.1836,
+      "step": 1309
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 7.569301420947841e-05,
+      "loss": 0.192,
+      "step": 1310
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 7.56560552667456e-05,
+      "loss": 0.2324,
+      "step": 1311
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 7.561907728615519e-05,
+      "loss": 0.2122,
+      "step": 1312
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 7.558208029514634e-05,
+      "loss": 0.1934,
+      "step": 1313
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 7.554506432117238e-05,
+      "loss": 0.1989,
+      "step": 1314
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 7.55080293917007e-05,
+      "loss": 0.3574,
+      "step": 1315
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 7.547097553421274e-05,
+      "loss": 0.1255,
+      "step": 1316
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 7.5433902776204e-05,
+      "loss": 0.1611,
+      "step": 1317
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 7.539681114518401e-05,
+      "loss": 0.1565,
+      "step": 1318
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 7.535970066867635e-05,
+      "loss": 0.1897,
+      "step": 1319
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 7.532257137421844e-05,
+      "loss": 0.1236,
+      "step": 1320
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 7.528542328936181e-05,
+      "loss": 0.1692,
+      "step": 1321
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 7.524825644167189e-05,
+      "loss": 0.1279,
+      "step": 1322
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 7.521107085872798e-05,
+      "loss": 0.1685,
+      "step": 1323
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 7.517386656812338e-05,
+      "loss": 0.0621,
+      "step": 1324
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 7.513664359746518e-05,
+      "loss": 0.1495,
+      "step": 1325
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 7.509940197437437e-05,
+      "loss": 0.1169,
+      "step": 1326
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 7.506214172648578e-05,
+      "loss": 0.1582,
+      "step": 1327
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 7.502486288144808e-05,
+      "loss": 0.1078,
+      "step": 1328
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 7.498756546692371e-05,
+      "loss": 0.0916,
+      "step": 1329
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 7.49502495105889e-05,
+      "loss": 0.1409,
+      "step": 1330
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 7.491291504013362e-05,
+      "loss": 0.098,
+      "step": 1331
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 7.487556208326163e-05,
+      "loss": 0.1808,
+      "step": 1332
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 7.483819066769034e-05,
+      "loss": 0.1104,
+      "step": 1333
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 7.48008008211509e-05,
+      "loss": 0.1452,
+      "step": 1334
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 7.476339257138816e-05,
+      "loss": 0.1443,
+      "step": 1335
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 7.472596594616054e-05,
+      "loss": 0.1813,
+      "step": 1336
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 7.468852097324019e-05,
+      "loss": 0.1302,
+      "step": 1337
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 7.465105768041283e-05,
+      "loss": 0.1097,
+      "step": 1338
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 7.461357609547774e-05,
+      "loss": 0.0999,
+      "step": 1339
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 7.457607624624787e-05,
+      "loss": 0.1431,
+      "step": 1340
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 7.453855816054962e-05,
+      "loss": 0.1345,
+      "step": 1341
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 7.450102186622298e-05,
+      "loss": 0.1023,
+      "step": 1342
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 7.446346739112143e-05,
+      "loss": 0.2119,
+      "step": 1343
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 7.442589476311198e-05,
+      "loss": 0.098,
+      "step": 1344
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 7.438830401007504e-05,
+      "loss": 0.1516,
+      "step": 1345
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 7.435069515990451e-05,
+      "loss": 0.2012,
+      "step": 1346
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 7.431306824050777e-05,
+      "loss": 0.0645,
+      "step": 1347
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 7.42754232798055e-05,
+      "loss": 0.1169,
+      "step": 1348
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 7.423776030573181e-05,
+      "loss": 0.1464,
+      "step": 1349
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 7.420007934623426e-05,
+      "loss": 0.1437,
+      "step": 1350
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 7.416238042927364e-05,
+      "loss": 0.1101,
+      "step": 1351
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 7.412466358282409e-05,
+      "loss": 0.1127,
+      "step": 1352
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.408692883487312e-05,
+      "loss": 0.1794,
+      "step": 1353
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 7.404917621342147e-05,
+      "loss": 0.2041,
+      "step": 1354
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.401140574648312e-05,
+      "loss": 0.175,
+      "step": 1355
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 7.397361746208536e-05,
+      "loss": 0.1199,
+      "step": 1356
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.393581138826864e-05,
+      "loss": 0.1225,
+      "step": 1357
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 7.389798755308667e-05,
+      "loss": 0.1514,
+      "step": 1358
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.386014598460625e-05,
+      "loss": 0.1183,
+      "step": 1359
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 7.382228671090744e-05,
+      "loss": 0.1191,
+      "step": 1360
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.378440976008338e-05,
+      "loss": 0.1312,
+      "step": 1361
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 7.374651516024033e-05,
+      "loss": 0.1176,
+      "step": 1362
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 7.370860293949768e-05,
+      "loss": 0.248,
+      "step": 1363
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.367067312598783e-05,
+      "loss": 0.1315,
+      "step": 1364
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 7.363272574785632e-05,
+      "loss": 0.1289,
+      "step": 1365
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.359476083326163e-05,
+      "loss": 0.1405,
+      "step": 1366
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 7.355677841037535e-05,
+      "loss": 0.1605,
+      "step": 1367
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.351877850738198e-05,
+      "loss": 0.1245,
+      "step": 1368
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 7.348076115247907e-05,
+      "loss": 0.1384,
+      "step": 1369
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.344272637387701e-05,
+      "loss": 0.1832,
+      "step": 1370
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 7.340467419979922e-05,
+      "loss": 0.1827,
+      "step": 1371
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.336660465848202e-05,
+      "loss": 0.2107,
+      "step": 1372
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 7.332851777817453e-05,
+      "loss": 0.1139,
+      "step": 1373
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 7.329041358713885e-05,
+      "loss": 0.1323,
+      "step": 1374
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 7.325229211364981e-05,
+      "loss": 0.0983,
+      "step": 1375
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 7.321415338599518e-05,
+      "loss": 0.1418,
+      "step": 1376
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 7.317599743247546e-05,
+      "loss": 0.1289,
+      "step": 1377
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 7.313782428140396e-05,
+      "loss": 0.1447,
+      "step": 1378
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.309963396110671e-05,
+      "loss": 0.1371,
+      "step": 1379
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 7.306142649992256e-05,
+      "loss": 0.1157,
+      "step": 1380
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.3023201926203e-05,
+      "loss": 0.1062,
+      "step": 1381
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 7.298496026831223e-05,
+      "loss": 0.1713,
+      "step": 1382
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.294670155462719e-05,
+      "loss": 0.1497,
+      "step": 1383
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 7.290842581353741e-05,
+      "loss": 0.1334,
+      "step": 1384
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.28701330734451e-05,
+      "loss": 0.1422,
+      "step": 1385
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.283182336276503e-05,
+      "loss": 0.1953,
+      "step": 1386
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.279349670992463e-05,
+      "loss": 0.0981,
+      "step": 1387
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.275515314336383e-05,
+      "loss": 0.1627,
+      "step": 1388
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.271679269153519e-05,
+      "loss": 0.1273,
+      "step": 1389
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.267841538290372e-05,
+      "loss": 0.1444,
+      "step": 1390
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.264002124594697e-05,
+      "loss": 0.0724,
+      "step": 1391
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.260161030915503e-05,
+      "loss": 0.1676,
+      "step": 1392
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.256318260103034e-05,
+      "loss": 0.0991,
+      "step": 1393
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.252473815008794e-05,
+      "loss": 0.1599,
+      "step": 1394
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.248627698485513e-05,
+      "loss": 0.1585,
+      "step": 1395
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.244779913387174e-05,
+      "loss": 0.1083,
+      "step": 1396
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.240930462568991e-05,
+      "loss": 0.1877,
+      "step": 1397
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.237079348887416e-05,
+      "loss": 0.1442,
+      "step": 1398
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.233226575200138e-05,
+      "loss": 0.1315,
+      "step": 1399
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.229372144366069e-05,
+      "loss": 0.1777,
+      "step": 1400
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 7.225516059245365e-05,
+      "loss": 0.1424,
+      "step": 1401
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.221658322699392e-05,
+      "loss": 0.2437,
+      "step": 1402
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 7.217798937590759e-05,
+      "loss": 0.1431,
+      "step": 1403
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.213937906783284e-05,
+      "loss": 0.1188,
+      "step": 1404
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 7.210075233142013e-05,
+      "loss": 0.1559,
+      "step": 1405
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.206210919533214e-05,
+      "loss": 0.1066,
+      "step": 1406
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 7.202344968824364e-05,
+      "loss": 0.1461,
+      "step": 1407
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.198477383884162e-05,
+      "loss": 0.1765,
+      "step": 1408
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 7.194608167582511e-05,
+      "loss": 0.1038,
+      "step": 1409
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 7.190737322790536e-05,
+      "loss": 0.1541,
+      "step": 1410
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.186864852380562e-05,
+      "loss": 0.1324,
+      "step": 1411
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 7.182990759226123e-05,
+      "loss": 0.1266,
+      "step": 1412
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.179115046201956e-05,
+      "loss": 0.1517,
+      "step": 1413
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 7.175237716184002e-05,
+      "loss": 0.1857,
+      "step": 1414
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.171358772049401e-05,
+      "loss": 0.1481,
+      "step": 1415
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 7.16747821667649e-05,
+      "loss": 0.129,
+      "step": 1416
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.163596052944801e-05,
+      "loss": 0.1499,
+      "step": 1417
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.159712283735061e-05,
+      "loss": 0.1222,
+      "step": 1418
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.155826911929192e-05,
+      "loss": 0.1045,
+      "step": 1419
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.151939940410295e-05,
+      "loss": 0.1785,
+      "step": 1420
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.148051372062668e-05,
+      "loss": 0.0867,
+      "step": 1421
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.144161209771788e-05,
+      "loss": 0.1957,
+      "step": 1422
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.140269456424317e-05,
+      "loss": 0.0767,
+      "step": 1423
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.1363761149081e-05,
+      "loss": 0.1603,
+      "step": 1424
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.132481188112155e-05,
+      "loss": 0.1641,
+      "step": 1425
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.128584678926682e-05,
+      "loss": 0.157,
+      "step": 1426
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.124686590243049e-05,
+      "loss": 0.1078,
+      "step": 1427
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.120786924953803e-05,
+      "loss": 0.2627,
+      "step": 1428
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.116885685952655e-05,
+      "loss": 0.1047,
+      "step": 1429
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.112982876134487e-05,
+      "loss": 0.2621,
+      "step": 1430
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.109078498395348e-05,
+      "loss": 0.1326,
+      "step": 1431
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.105172555632443e-05,
+      "loss": 0.0909,
+      "step": 1432
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.10126505074415e-05,
+      "loss": 0.1256,
+      "step": 1433
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.097355986629995e-05,
+      "loss": 0.2025,
+      "step": 1434
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.093445366190669e-05,
+      "loss": 0.2139,
+      "step": 1435
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.089533192328011e-05,
+      "loss": 0.2699,
+      "step": 1436
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.085619467945021e-05,
+      "loss": 0.1949,
+      "step": 1437
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.08170419594584e-05,
+      "loss": 0.2263,
+      "step": 1438
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.077787379235766e-05,
+      "loss": 0.1958,
+      "step": 1439
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.073869020721239e-05,
+      "loss": 0.1647,
+      "step": 1440
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.069949123309841e-05,
+      "loss": 0.1638,
+      "step": 1441
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.066027689910302e-05,
+      "loss": 0.1515,
+      "step": 1442
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.062104723432485e-05,
+      "loss": 0.2299,
+      "step": 1443
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.058180226787397e-05,
+      "loss": 0.1487,
+      "step": 1444
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.054254202887174e-05,
+      "loss": 0.1216,
+      "step": 1445
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.050326654645093e-05,
+      "loss": 0.2324,
+      "step": 1446
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.046397584975554e-05,
+      "loss": 0.1231,
+      "step": 1447
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.042466996794091e-05,
+      "loss": 0.2776,
+      "step": 1448
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.038534893017364e-05,
+      "loss": 0.1667,
+      "step": 1449
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.034601276563155e-05,
+      "loss": 0.1565,
+      "step": 1450
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.030666150350376e-05,
+      "loss": 0.124,
+      "step": 1451
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.026729517299047e-05,
+      "loss": 0.1453,
+      "step": 1452
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.022791380330319e-05,
+      "loss": 0.1796,
+      "step": 1453
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.018851742366447e-05,
+      "loss": 0.1681,
+      "step": 1454
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.014910606330813e-05,
+      "loss": 0.2474,
+      "step": 1455
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.010967975147897e-05,
+      "loss": 0.1792,
+      "step": 1456
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.007023851743299e-05,
+      "loss": 0.106,
+      "step": 1457
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.003078239043723e-05,
+      "loss": 0.1796,
+      "step": 1458
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 6.999131139976973e-05,
+      "loss": 0.0981,
+      "step": 1459
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.995182557471963e-05,
+      "loss": 0.1234,
+      "step": 1460
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.991232494458703e-05,
+      "loss": 0.0969,
+      "step": 1461
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.987280953868308e-05,
+      "loss": 0.1602,
+      "step": 1462
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.983327938632979e-05,
+      "loss": 0.1829,
+      "step": 1463
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.979373451686024e-05,
+      "loss": 0.0729,
+      "step": 1464
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 6.975417495961831e-05,
+      "loss": 0.2094,
+      "step": 1465
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.971460074395888e-05,
+      "loss": 0.1804,
+      "step": 1466
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.96750118992476e-05,
+      "loss": 0.1877,
+      "step": 1467
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.96354084548611e-05,
+      "loss": 0.2236,
+      "step": 1468
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.959579044018675e-05,
+      "loss": 0.1731,
+      "step": 1469
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.955615788462274e-05,
+      "loss": 0.1411,
+      "step": 1470
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 6.95165108175781e-05,
+      "loss": 0.1305,
+      "step": 1471
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.94768492684726e-05,
+      "loss": 0.2127,
+      "step": 1472
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.943717326673674e-05,
+      "loss": 0.1234,
+      "step": 1473
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.939748284181175e-05,
+      "loss": 0.1626,
+      "step": 1474
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.935777802314959e-05,
+      "loss": 0.1849,
+      "step": 1475
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.93180588402129e-05,
+      "loss": 0.1483,
+      "step": 1476
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.927832532247492e-05,
+      "loss": 0.1311,
+      "step": 1477
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.92385774994196e-05,
+      "loss": 0.2322,
+      "step": 1478
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.919881540054145e-05,
+      "loss": 0.2114,
+      "step": 1479
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.915903905534565e-05,
+      "loss": 0.1641,
+      "step": 1480
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.911924849334782e-05,
+      "loss": 0.1117,
+      "step": 1481
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.907944374407426e-05,
+      "loss": 0.1016,
+      "step": 1482
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.903962483706174e-05,
+      "loss": 0.2517,
+      "step": 1483
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.899979180185755e-05,
+      "loss": 0.1779,
+      "step": 1484
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.895994466801943e-05,
+      "loss": 0.2793,
+      "step": 1485
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.892008346511561e-05,
+      "loss": 0.188,
+      "step": 1486
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.888020822272478e-05,
+      "loss": 0.1424,
+      "step": 1487
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.884031897043599e-05,
+      "loss": 0.1063,
+      "step": 1488
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.880041573784875e-05,
+      "loss": 0.2078,
+      "step": 1489
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.87604985545729e-05,
+      "loss": 0.1558,
+      "step": 1490
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.872056745022866e-05,
+      "loss": 0.1315,
+      "step": 1491
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.868062245444655e-05,
+      "loss": 0.0969,
+      "step": 1492
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.86406635968674e-05,
+      "loss": 0.1914,
+      "step": 1493
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.86006909071424e-05,
+      "loss": 0.1185,
+      "step": 1494
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.856070441493287e-05,
+      "loss": 0.1848,
+      "step": 1495
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.85207041499105e-05,
+      "loss": 0.2256,
+      "step": 1496
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.84806901417571e-05,
+      "loss": 0.1541,
+      "step": 1497
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.84406624201648e-05,
+      "loss": 0.1443,
+      "step": 1498
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.840062101483577e-05,
+      "loss": 0.1954,
+      "step": 1499
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.836056595548239e-05,
+      "loss": 0.1887,
+      "step": 1500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.832049727182721e-05,
+      "loss": 0.1197,
+      "step": 1501
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.828041499360283e-05,
+      "loss": 0.1656,
+      "step": 1502
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.824031915055198e-05,
+      "loss": 0.189,
+      "step": 1503
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.820020977242742e-05,
+      "loss": 0.0792,
+      "step": 1504
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.8160086888992e-05,
+      "loss": 0.1741,
+      "step": 1505
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.811995053001854e-05,
+      "loss": 0.1339,
+      "step": 1506
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.80798007252899e-05,
+      "loss": 0.1514,
+      "step": 1507
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.803963750459889e-05,
+      "loss": 0.1334,
+      "step": 1508
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.799946089774831e-05,
+      "loss": 0.1704,
+      "step": 1509
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.795927093455084e-05,
+      "loss": 0.1575,
+      "step": 1510
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.79190676448291e-05,
+      "loss": 0.1349,
+      "step": 1511
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.787885105841562e-05,
+      "loss": 0.0906,
+      "step": 1512
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.783862120515276e-05,
+      "loss": 0.0787,
+      "step": 1513
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.779837811489275e-05,
+      "loss": 0.1165,
+      "step": 1514
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.775812181749763e-05,
+      "loss": 0.1164,
+      "step": 1515
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.771785234283925e-05,
+      "loss": 0.1163,
+      "step": 1516
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.767756972079922e-05,
+      "loss": 0.1314,
+      "step": 1517
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.763727398126891e-05,
+      "loss": 0.1117,
+      "step": 1518
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.759696515414947e-05,
+      "loss": 0.1587,
+      "step": 1519
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.755664326935166e-05,
+      "loss": 0.1169,
+      "step": 1520
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.751630835679604e-05,
+      "loss": 0.1287,
+      "step": 1521
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.747596044641276e-05,
+      "loss": 0.1105,
+      "step": 1522
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.74355995681417e-05,
+      "loss": 0.0996,
+      "step": 1523
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.739522575193222e-05,
+      "loss": 0.0914,
+      "step": 1524
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.735483902774343e-05,
+      "loss": 0.1171,
+      "step": 1525
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.731443942554397e-05,
+      "loss": 0.1226,
+      "step": 1526
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.727402697531194e-05,
+      "loss": 0.1271,
+      "step": 1527
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.723360170703512e-05,
+      "loss": 0.1805,
+      "step": 1528
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.71931636507107e-05,
+      "loss": 0.0721,
+      "step": 1529
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.715271283634545e-05,
+      "loss": 0.1025,
+      "step": 1530
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.711224929395549e-05,
+      "loss": 0.0682,
+      "step": 1531
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.70717730535665e-05,
+      "loss": 0.1029,
+      "step": 1532
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.703128414521349e-05,
+      "loss": 0.0757,
+      "step": 1533
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.699078259894094e-05,
+      "loss": 0.1528,
+      "step": 1534
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.695026844480266e-05,
+      "loss": 0.0955,
+      "step": 1535
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.690974171286182e-05,
+      "loss": 0.1147,
+      "step": 1536
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.686920243319098e-05,
+      "loss": 0.0826,
+      "step": 1537
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.682865063587191e-05,
+      "loss": 0.0932,
+      "step": 1538
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.67880863509958e-05,
+      "loss": 0.0946,
+      "step": 1539
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.674750960866296e-05,
+      "loss": 0.1397,
+      "step": 1540
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.670692043898306e-05,
+      "loss": 0.0755,
+      "step": 1541
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.666631887207493e-05,
+      "loss": 0.1028,
+      "step": 1542
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.662570493806661e-05,
+      "loss": 0.067,
+      "step": 1543
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 6.658507866709537e-05,
+      "loss": 0.0685,
+      "step": 1544
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.654444008930753e-05,
+      "loss": 0.0967,
+      "step": 1545
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 6.650378923485864e-05,
+      "loss": 0.1437,
+      "step": 1546
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.64631261339133e-05,
+      "loss": 0.1005,
+      "step": 1547
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 6.642245081664523e-05,
+      "loss": 0.0709,
+      "step": 1548
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.638176331323719e-05,
+      "loss": 0.1249,
+      "step": 1549
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 6.634106365388102e-05,
+      "loss": 0.1072,
+      "step": 1550
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 6.630035186877754e-05,
+      "loss": 0.0839,
+      "step": 1551
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.625962798813658e-05,
+      "loss": 0.1198,
+      "step": 1552
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 6.621889204217695e-05,
+      "loss": 0.1112,
+      "step": 1553
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.61781440611264e-05,
+      "loss": 0.1857,
+      "step": 1554
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 6.613738407522168e-05,
+      "loss": 0.0936,
+      "step": 1555
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.609661211470834e-05,
+      "loss": 0.091,
+      "step": 1556
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 6.605582820984087e-05,
+      "loss": 0.0941,
+      "step": 1557
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.601503239088263e-05,
+      "loss": 0.0924,
+      "step": 1558
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 6.59742246881058e-05,
+      "loss": 0.0721,
+      "step": 1559
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.59334051317914e-05,
+      "loss": 0.1606,
+      "step": 1560
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 6.589257375222921e-05,
+      "loss": 0.0986,
+      "step": 1561
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.585173057971787e-05,
+      "loss": 0.1031,
+      "step": 1562
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 6.581087564456464e-05,
+      "loss": 0.1436,
+      "step": 1563
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.577000897708561e-05,
+      "loss": 0.1001,
+      "step": 1564
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 6.572913060760555e-05,
+      "loss": 0.0925,
+      "step": 1565
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 6.568824056645788e-05,
+      "loss": 0.138,
+      "step": 1566
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.564733888398472e-05,
+      "loss": 0.0719,
+      "step": 1567
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 6.560642559053682e-05,
+      "loss": 0.0672,
+      "step": 1568
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.556550071647355e-05,
+      "loss": 0.0962,
+      "step": 1569
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 6.552456429216283e-05,
+      "loss": 0.1174,
+      "step": 1570
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.548361634798123e-05,
+      "loss": 0.0429,
+      "step": 1571
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 6.544265691431379e-05,
+      "loss": 0.1166,
+      "step": 1572
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.540168602155413e-05,
+      "loss": 0.1383,
+      "step": 1573
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 6.536070370010434e-05,
+      "loss": 0.0501,
+      "step": 1574
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.5319709980375e-05,
+      "loss": 0.0815,
+      "step": 1575
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 6.527870489278518e-05,
+      "loss": 0.1214,
+      "step": 1576
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.523768846776233e-05,
+      "loss": 0.0913,
+      "step": 1577
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 6.519666073574233e-05,
+      "loss": 0.2252,
+      "step": 1578
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.515562172716947e-05,
+      "loss": 0.0734,
+      "step": 1579
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 6.511457147249643e-05,
+      "loss": 0.0975,
+      "step": 1580
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.507351000218414e-05,
+      "loss": 0.0916,
+      "step": 1581
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 6.503243734670197e-05,
+      "loss": 0.0987,
+      "step": 1582
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 6.49913535365275e-05,
+      "loss": 0.1075,
+      "step": 1583
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.495025860214665e-05,
+      "loss": 0.0883,
+      "step": 1584
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 6.490915257405354e-05,
+      "loss": 0.0921,
+      "step": 1585
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.486803548275056e-05,
+      "loss": 0.1108,
+      "step": 1586
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 6.482690735874832e-05,
+      "loss": 0.0975,
+      "step": 1587
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.478576823256555e-05,
+      "loss": 0.0531,
+      "step": 1588
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 6.474461813472921e-05,
+      "loss": 0.0864,
+      "step": 1589
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.470345709577439e-05,
+      "loss": 0.0925,
+      "step": 1590
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 6.466228514624429e-05,
+      "loss": 0.1111,
+      "step": 1591
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.462110231669016e-05,
+      "loss": 0.0779,
+      "step": 1592
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 6.457990863767144e-05,
+      "loss": 0.041,
+      "step": 1593
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.45387041397555e-05,
+      "loss": 0.0837,
+      "step": 1594
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 6.449748885351781e-05,
+      "loss": 0.238,
+      "step": 1595
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.44562628095418e-05,
+      "loss": 0.1342,
+      "step": 1596
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 6.441502603841891e-05,
+      "loss": 0.1013,
+      "step": 1597
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 6.437377857074858e-05,
+      "loss": 0.0869,
+      "step": 1598
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.433252043713808e-05,
+      "loss": 0.0794,
+      "step": 1599
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 6.429125166820267e-05,
+      "loss": 0.173,
+      "step": 1600
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.424997229456553e-05,
+      "loss": 0.0668,
+      "step": 1601
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 6.420868234685764e-05,
+      "loss": 0.1014,
+      "step": 1602
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.416738185571785e-05,
+      "loss": 0.1544,
+      "step": 1603
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 6.412607085179282e-05,
+      "loss": 0.2125,
+      "step": 1604
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.408474936573708e-05,
+      "loss": 0.0704,
+      "step": 1605
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 6.404341742821286e-05,
+      "loss": 0.0954,
+      "step": 1606
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.400207506989015e-05,
+      "loss": 0.113,
+      "step": 1607
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 6.396072232144673e-05,
+      "loss": 0.1351,
+      "step": 1608
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.391935921356804e-05,
+      "loss": 0.0691,
+      "step": 1609
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 6.387798577694724e-05,
+      "loss": 0.0526,
+      "step": 1610
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.38366020422851e-05,
+      "loss": 0.0686,
+      "step": 1611
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 6.379520804029011e-05,
+      "loss": 0.1102,
+      "step": 1612
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 6.37538038016783e-05,
+      "loss": 0.0882,
+      "step": 1613
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.371238935717336e-05,
+      "loss": 0.1675,
+      "step": 1614
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 6.367096473750651e-05,
+      "loss": 0.0839,
+      "step": 1615
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.362952997341656e-05,
+      "loss": 0.1609,
+      "step": 1616
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 6.358808509564981e-05,
+      "loss": 0.1199,
+      "step": 1617
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.354663013496006e-05,
+      "loss": 0.1843,
+      "step": 1618
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.350516512210865e-05,
+      "loss": 0.1185,
+      "step": 1619
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.34636900878643e-05,
+      "loss": 0.1414,
+      "step": 1620
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.342220506300323e-05,
+      "loss": 0.1183,
+      "step": 1621
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.338071007830904e-05,
+      "loss": 0.0578,
+      "step": 1622
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.333920516457275e-05,
+      "loss": 0.0999,
+      "step": 1623
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.329769035259269e-05,
+      "loss": 0.1086,
+      "step": 1624
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.325616567317461e-05,
+      "loss": 0.1537,
+      "step": 1625
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.321463115713153e-05,
+      "loss": 0.1527,
+      "step": 1626
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.31730868352838e-05,
+      "loss": 0.1185,
+      "step": 1627
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.3131532738459e-05,
+      "loss": 0.0355,
+      "step": 1628
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.308996889749201e-05,
+      "loss": 0.1286,
+      "step": 1629
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.304839534322494e-05,
+      "loss": 0.1091,
+      "step": 1630
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.300681210650707e-05,
+      "loss": 0.0795,
+      "step": 1631
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.29652192181949e-05,
+      "loss": 0.1671,
+      "step": 1632
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.292361670915206e-05,
+      "loss": 0.0807,
+      "step": 1633
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.288200461024937e-05,
+      "loss": 0.2134,
+      "step": 1634
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.284038295236471e-05,
+      "loss": 0.1451,
+      "step": 1635
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.279875176638306e-05,
+      "loss": 0.2125,
+      "step": 1636
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.275711108319655e-05,
+      "loss": 0.1118,
+      "step": 1637
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.271546093370424e-05,
+      "loss": 0.0884,
+      "step": 1638
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.267380134881229e-05,
+      "loss": 0.1238,
+      "step": 1639
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.263213235943382e-05,
+      "loss": 0.1141,
+      "step": 1640
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.259045399648898e-05,
+      "loss": 0.1125,
+      "step": 1641
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.254876629090482e-05,
+      "loss": 0.1343,
+      "step": 1642
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.250706927361537e-05,
+      "loss": 0.1481,
+      "step": 1643
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.246536297556152e-05,
+      "loss": 0.0939,
+      "step": 1644
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.24236474276911e-05,
+      "loss": 0.1105,
+      "step": 1645
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.238192266095877e-05,
+      "loss": 0.1364,
+      "step": 1646
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.234018870632601e-05,
+      "loss": 0.0629,
+      "step": 1647
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.22984455947612e-05,
+      "loss": 0.0884,
+      "step": 1648
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.225669335723942e-05,
+      "loss": 0.1003,
+      "step": 1649
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.221493202474256e-05,
+      "loss": 0.1093,
+      "step": 1650
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.21731616282593e-05,
+      "loss": 0.1804,
+      "step": 1651
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.213138219878496e-05,
+      "loss": 0.1067,
+      "step": 1652
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.208959376732164e-05,
+      "loss": 0.0937,
+      "step": 1653
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.204779636487809e-05,
+      "loss": 0.1749,
+      "step": 1654
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.200599002246971e-05,
+      "loss": 0.0837,
+      "step": 1655
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.196417477111854e-05,
+      "loss": 0.1713,
+      "step": 1656
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 6.192235064185322e-05,
+      "loss": 0.1184,
+      "step": 1657
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.188051766570902e-05,
+      "loss": 0.1238,
+      "step": 1658
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.183867587372771e-05,
+      "loss": 0.1331,
+      "step": 1659
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 6.179682529695766e-05,
+      "loss": 0.1163,
+      "step": 1660
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.175496596645372e-05,
+      "loss": 0.1573,
+      "step": 1661
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.171309791327728e-05,
+      "loss": 0.1729,
+      "step": 1662
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.167122116849613e-05,
+      "loss": 0.1434,
+      "step": 1663
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.162933576318456e-05,
+      "loss": 0.1187,
+      "step": 1664
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.158744172842331e-05,
+      "loss": 0.0822,
+      "step": 1665
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.154553909529948e-05,
+      "loss": 0.1042,
+      "step": 1666
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.150362789490653e-05,
+      "loss": 0.1312,
+      "step": 1667
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.146170815834436e-05,
+      "loss": 0.0596,
+      "step": 1668
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.141977991671913e-05,
+      "loss": 0.127,
+      "step": 1669
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.137784320114333e-05,
+      "loss": 0.1416,
+      "step": 1670
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.133589804273574e-05,
+      "loss": 0.0825,
+      "step": 1671
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.12939444726214e-05,
+      "loss": 0.1126,
+      "step": 1672
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.125198252193165e-05,
+      "loss": 0.1035,
+      "step": 1673
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.121001222180394e-05,
+      "loss": 0.074,
+      "step": 1674
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.116803360338201e-05,
+      "loss": 0.1796,
+      "step": 1675
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.112604669781572e-05,
+      "loss": 0.1981,
+      "step": 1676
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 6.108405153626111e-05,
+      "loss": 0.0895,
+      "step": 1677
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 6.104204814988033e-05,
+      "loss": 0.1072,
+      "step": 1678
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 6.10000365698416e-05,
+      "loss": 0.1115,
+      "step": 1679
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 6.095801682731932e-05,
+      "loss": 0.1251,
+      "step": 1680
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 6.091598895349382e-05,
+      "loss": 0.2005,
+      "step": 1681
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 6.0873952979551566e-05,
+      "loss": 0.0687,
+      "step": 1682
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 6.083190893668498e-05,
+      "loss": 0.1065,
+      "step": 1683
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 6.07898568560925e-05,
+      "loss": 0.0923,
+      "step": 1684
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 6.0747796768978485e-05,
+      "loss": 0.1051,
+      "step": 1685
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 6.07057287065533e-05,
+      "loss": 0.0825,
+      "step": 1686
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 6.066365270003318e-05,
+      "loss": 0.0923,
+      "step": 1687
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 6.062156878064026e-05,
+      "loss": 0.1298,
+      "step": 1688
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 6.0579476979602554e-05,
+      "loss": 0.0426,
+      "step": 1689
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 6.0537377328153924e-05,
+      "loss": 0.1603,
+      "step": 1690
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 6.049526985753409e-05,
+      "loss": 0.2227,
+      "step": 1691
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 6.045315459898849e-05,
+      "loss": 0.0963,
+      "step": 1692
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 6.041103158376844e-05,
+      "loss": 0.0633,
+      "step": 1693
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 6.0368900843130935e-05,
+      "loss": 0.0869,
+      "step": 1694
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 6.032676240833876e-05,
+      "loss": 0.0632,
+      "step": 1695
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 6.0284616310660344e-05,
+      "loss": 0.1694,
+      "step": 1696
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 6.024246258136987e-05,
+      "loss": 0.0713,
+      "step": 1697
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 6.020030125174716e-05,
+      "loss": 0.0805,
+      "step": 1698
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 6.015813235307764e-05,
+      "loss": 0.1217,
+      "step": 1699
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 6.01159559166524e-05,
+      "loss": 0.0825,
+      "step": 1700
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 6.007377197376811e-05,
+      "loss": 0.0864,
+      "step": 1701
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 6.003158055572701e-05,
+      "loss": 0.1668,
+      "step": 1702
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.998938169383687e-05,
+      "loss": 0.0931,
+      "step": 1703
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.9947175419410985e-05,
+      "loss": 0.0864,
+      "step": 1704
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.9904961763768196e-05,
+      "loss": 0.1122,
+      "step": 1705
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.986274075823276e-05,
+      "loss": 0.0515,
+      "step": 1706
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.9820512434134425e-05,
+      "loss": 0.1429,
+      "step": 1707
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.9778276822808365e-05,
+      "loss": 0.0577,
+      "step": 1708
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.9736033955595136e-05,
+      "loss": 0.0858,
+      "step": 1709
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.969378386384071e-05,
+      "loss": 0.0984,
+      "step": 1710
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 5.9651526578896435e-05,
+      "loss": 0.0715,
+      "step": 1711
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.9609262132118927e-05,
+      "loss": 0.1026,
+      "step": 1712
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.956699055487018e-05,
+      "loss": 0.1553,
+      "step": 1713
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.9524711878517466e-05,
+      "loss": 0.0939,
+      "step": 1714
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.9482426134433286e-05,
+      "loss": 0.0251,
+      "step": 1715
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.944013335399546e-05,
+      "loss": 0.1276,
+      "step": 1716
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 5.939783356858696e-05,
+      "loss": 0.0574,
+      "step": 1717
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.935552680959598e-05,
+      "loss": 0.1302,
+      "step": 1718
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.9313213108415896e-05,
+      "loss": 0.0762,
+      "step": 1719
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.927089249644524e-05,
+      "loss": 0.0588,
+      "step": 1720
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 5.922856500508764e-05,
+      "loss": 0.0916,
+      "step": 1721
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.918623066575187e-05,
+      "loss": 0.0463,
+      "step": 1722
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.914388950985178e-05,
+      "loss": 0.08,
+      "step": 1723
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.910154156880622e-05,
+      "loss": 0.097,
+      "step": 1724
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.905918687403915e-05,
+      "loss": 0.144,
+      "step": 1725
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 5.901682545697951e-05,
+      "loss": 0.0467,
+      "step": 1726
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.897445734906122e-05,
+      "loss": 0.1138,
+      "step": 1727
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.893208258172319e-05,
+      "loss": 0.0574,
+      "step": 1728
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.8889701186409227e-05,
+      "loss": 0.0726,
+      "step": 1729
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.884731319456812e-05,
+      "loss": 0.0472,
+      "step": 1730
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.880491863765348e-05,
+      "loss": 0.1211,
+      "step": 1731
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 5.8762517547123875e-05,
+      "loss": 0.0902,
+      "step": 1732
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.872010995444264e-05,
+      "loss": 0.0522,
+      "step": 1733
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.8677695891077965e-05,
+      "loss": 0.0804,
+      "step": 1734
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.8635275388502845e-05,
+      "loss": 0.1315,
+      "step": 1735
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.8592848478195083e-05,
+      "loss": 0.1296,
+      "step": 1736
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.8550415191637175e-05,
+      "loss": 0.073,
+      "step": 1737
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 5.850797556031641e-05,
+      "loss": 0.0998,
+      "step": 1738
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.8465529615724726e-05,
+      "loss": 0.0732,
+      "step": 1739
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.842307738935878e-05,
+      "loss": 0.1642,
+      "step": 1740
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 5.838061891271991e-05,
+      "loss": 0.0622,
+      "step": 1741
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.833815421731403e-05,
+      "loss": 0.0724,
+      "step": 1742
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.829568333465175e-05,
+      "loss": 0.0696,
+      "step": 1743
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.82532062962482e-05,
+      "loss": 0.095,
+      "step": 1744
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.821072313362309e-05,
+      "loss": 0.0716,
+      "step": 1745
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.816823387830073e-05,
+      "loss": 0.0786,
+      "step": 1746
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 5.8125738561809865e-05,
+      "loss": 0.0848,
+      "step": 1747
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.8083237215683836e-05,
+      "loss": 0.0764,
+      "step": 1748
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.8040729871460355e-05,
+      "loss": 0.0522,
+      "step": 1749
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.799821656068166e-05,
+      "loss": 0.0329,
+      "step": 1750
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.7955697314894395e-05,
+      "loss": 0.1014,
+      "step": 1751
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.79131721656496e-05,
+      "loss": 0.1129,
+      "step": 1752
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 5.787064114450268e-05,
+      "loss": 0.0786,
+      "step": 1753
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.7828104283013454e-05,
+      "loss": 0.075,
+      "step": 1754
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.778556161274603e-05,
+      "loss": 0.0934,
+      "step": 1755
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.77430131652688e-05,
+      "loss": 0.127,
+      "step": 1756
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.770045897215451e-05,
+      "loss": 0.096,
+      "step": 1757
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 5.7657899064980146e-05,
+      "loss": 0.1191,
+      "step": 1758
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.761533347532689e-05,
+      "loss": 0.0876,
+      "step": 1759
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.75727622347802e-05,
+      "loss": 0.0444,
+      "step": 1760
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.75301853749297e-05,
+      "loss": 0.1289,
+      "step": 1761
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 5.748760292736919e-05,
+      "loss": 0.0609,
+      "step": 1762
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.7445014923696606e-05,
+      "loss": 0.0796,
+      "step": 1763
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.740242139551401e-05,
+      "loss": 0.0569,
+      "step": 1764
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.7359822374427566e-05,
+      "loss": 0.066,
+      "step": 1765
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.731721789204754e-05,
+      "loss": 0.1001,
+      "step": 1766
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.727460797998818e-05,
+      "loss": 0.0727,
+      "step": 1767
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 5.723199266986786e-05,
+      "loss": 0.0528,
+      "step": 1768
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.718937199330888e-05,
+      "loss": 0.0316,
+      "step": 1769
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.714674598193756e-05,
+      "loss": 0.0946,
+      "step": 1770
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.710411466738417e-05,
+      "loss": 0.0914,
+      "step": 1771
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.706147808128288e-05,
+      "loss": 0.0588,
+      "step": 1772
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 5.701883625527188e-05,
+      "loss": 0.0533,
+      "step": 1773
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.697618922099309e-05,
+      "loss": 0.1184,
+      "step": 1774
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.693353701009244e-05,
+      "loss": 0.0717,
+      "step": 1775
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.689087965421962e-05,
+      "loss": 0.0729,
+      "step": 1776
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.684821718502815e-05,
+      "loss": 0.1157,
+      "step": 1777
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.680554963417535e-05,
+      "loss": 0.088,
+      "step": 1778
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 5.676287703332234e-05,
+      "loss": 0.0714,
+      "step": 1779
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.6720199414133934e-05,
+      "loss": 0.1658,
+      "step": 1780
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.667751680827871e-05,
+      "loss": 0.1393,
+      "step": 1781
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.663482924742893e-05,
+      "loss": 0.0824,
+      "step": 1782
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.659213676326054e-05,
+      "loss": 0.1023,
+      "step": 1783
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.6549439387453115e-05,
+      "loss": 0.044,
+      "step": 1784
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 5.6506737151689906e-05,
+      "loss": 0.0952,
+      "step": 1785
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.6464030087657714e-05,
+      "loss": 0.0976,
+      "step": 1786
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.6421318227046996e-05,
+      "loss": 0.0999,
+      "step": 1787
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 5.6378601601551685e-05,
+      "loss": 0.0864,
+      "step": 1788
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.63358802428693e-05,
+      "loss": 0.0928,
+      "step": 1789
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.629315418270086e-05,
+      "loss": 0.2285,
+      "step": 1790
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.62504234527509e-05,
+      "loss": 0.0455,
+      "step": 1791
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.620768808472736e-05,
+      "loss": 0.1088,
+      "step": 1792
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.6164948110341674e-05,
+      "loss": 0.0536,
+      "step": 1793
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 5.6122203561308674e-05,
+      "loss": 0.1024,
+      "step": 1794
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.607945446934659e-05,
+      "loss": 0.0749,
+      "step": 1795
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.603670086617703e-05,
+      "loss": 0.1158,
+      "step": 1796
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.599394278352491e-05,
+      "loss": 0.0716,
+      "step": 1797
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.595118025311855e-05,
+      "loss": 0.1395,
+      "step": 1798
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.590841330668948e-05,
+      "loss": 0.0786,
+      "step": 1799
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 5.586564197597256e-05,
+      "loss": 0.1323,
+      "step": 1800
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.5822866292705914e-05,
+      "loss": 0.0839,
+      "step": 1801
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.578008628863085e-05,
+      "loss": 0.0963,
+      "step": 1802
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.57373019954919e-05,
+      "loss": 0.1201,
+      "step": 1803
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.569451344503681e-05,
+      "loss": 0.0823,
+      "step": 1804
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 5.565172066901646e-05,
+      "loss": 0.0398,
+      "step": 1805
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.560892369918484e-05,
+      "loss": 0.0774,
+      "step": 1806
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.556612256729909e-05,
+      "loss": 0.0618,
+      "step": 1807
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.552331730511944e-05,
+      "loss": 0.0851,
+      "step": 1808
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.5480507944409154e-05,
+      "loss": 0.0806,
+      "step": 1809
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.5437694516934534e-05,
+      "loss": 0.1256,
+      "step": 1810
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 5.539487705446497e-05,
+      "loss": 0.084,
+      "step": 1811
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.535205558877277e-05,
+      "loss": 0.0478,
+      "step": 1812
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.530923015163323e-05,
+      "loss": 0.0949,
+      "step": 1813
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.526640077482461e-05,
+      "loss": 0.0822,
+      "step": 1814
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 5.522356749012807e-05,
+      "loss": 0.1755,
+      "step": 1815
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.5180730329327725e-05,
+      "loss": 0.0796,
+      "step": 1816
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.5137889324210476e-05,
+      "loss": 0.0331,
+      "step": 1817
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.5095044506566164e-05,
+      "loss": 0.1199,
+      "step": 1818
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.505219590818741e-05,
+      "loss": 0.1099,
+      "step": 1819
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.500934356086964e-05,
+      "loss": 0.0766,
+      "step": 1820
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.496648749641109e-05,
+      "loss": 0.0794,
+      "step": 1821
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.492362774661273e-05,
+      "loss": 0.1034,
+      "step": 1822
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.488076434327829e-05,
+      "loss": 0.0736,
+      "step": 1823
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.483789731821415e-05,
+      "loss": 0.1208,
+      "step": 1824
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.479502670322947e-05,
+      "loss": 0.0983,
+      "step": 1825
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 5.475215253013599e-05,
+      "loss": 0.0742,
+      "step": 1826
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.470927483074816e-05,
+      "loss": 0.1245,
+      "step": 1827
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.4666393636882965e-05,
+      "loss": 0.0591,
+      "step": 1828
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.4623508980360085e-05,
+      "loss": 0.0786,
+      "step": 1829
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.4580620893001675e-05,
+      "loss": 0.1197,
+      "step": 1830
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.453772940663249e-05,
+      "loss": 0.0773,
+      "step": 1831
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 5.449483455307979e-05,
+      "loss": 0.0997,
+      "step": 1832
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.4451936364173337e-05,
+      "loss": 0.1237,
+      "step": 1833
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.440903487174538e-05,
+      "loss": 0.0615,
+      "step": 1834
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 5.436613010763056e-05,
+      "loss": 0.0596,
+      "step": 1835
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.432322210366606e-05,
+      "loss": 0.1071,
+      "step": 1836
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.428031089169135e-05,
+      "loss": 0.0656,
+      "step": 1837
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.4237396503548354e-05,
+      "loss": 0.0854,
+      "step": 1838
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.419447897108132e-05,
+      "loss": 0.0576,
+      "step": 1839
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.415155832613683e-05,
+      "loss": 0.076,
+      "step": 1840
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 5.4108634600563814e-05,
+      "loss": 0.0862,
+      "step": 1841
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.4065707826213405e-05,
+      "loss": 0.112,
+      "step": 1842
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.402277803493909e-05,
+      "loss": 0.1256,
+      "step": 1843
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.397984525859656e-05,
+      "loss": 0.1159,
+      "step": 1844
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.393690952904369e-05,
+      "loss": 0.0631,
+      "step": 1845
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.3893970878140585e-05,
+      "loss": 0.0746,
+      "step": 1846
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.385102933774951e-05,
+      "loss": 0.0453,
+      "step": 1847
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.380808493973487e-05,
+      "loss": 0.1265,
+      "step": 1848
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.376513771596319e-05,
+      "loss": 0.1156,
+      "step": 1849
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.3722187698303086e-05,
+      "loss": 0.0748,
+      "step": 1850
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.3679234918625255e-05,
+      "loss": 0.0389,
+      "step": 1851
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.363627940880245e-05,
+      "loss": 0.0554,
+      "step": 1852
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.359332120070942e-05,
+      "loss": 0.077,
+      "step": 1853
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.355036032622297e-05,
+      "loss": 0.0916,
+      "step": 1854
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.3507396817221824e-05,
+      "loss": 0.0842,
+      "step": 1855
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 5.346443070558671e-05,
+      "loss": 0.085,
+      "step": 1856
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.342146202320024e-05,
+      "loss": 0.1217,
+      "step": 1857
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.337849080194699e-05,
+      "loss": 0.1191,
+      "step": 1858
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.333551707371336e-05,
+      "loss": 0.0782,
+      "step": 1859
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.329254087038765e-05,
+      "loss": 0.0433,
+      "step": 1860
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.324956222386001e-05,
+      "loss": 0.0734,
+      "step": 1861
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.320658116602234e-05,
+      "loss": 0.1308,
+      "step": 1862
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.316359772876839e-05,
+      "loss": 0.1459,
+      "step": 1863
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.312061194399366e-05,
+      "loss": 0.1291,
+      "step": 1864
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.307762384359536e-05,
+      "loss": 0.0533,
+      "step": 1865
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.303463345947248e-05,
+      "loss": 0.0677,
+      "step": 1866
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 5.2991640823525634e-05,
+      "loss": 0.0533,
+      "step": 1867
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.294864596765716e-05,
+      "loss": 0.05,
+      "step": 1868
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.290564892377104e-05,
+      "loss": 0.1478,
+      "step": 1869
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.2862649723772826e-05,
+      "loss": 0.1203,
+      "step": 1870
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.281964839956973e-05,
+      "loss": 0.118,
+      "step": 1871
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.277664498307053e-05,
+      "loss": 0.0779,
+      "step": 1872
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 5.273363950618555e-05,
+      "loss": 0.0612,
+      "step": 1873
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.26906320008266e-05,
+      "loss": 0.0771,
+      "step": 1874
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.264762249890707e-05,
+      "loss": 0.1268,
+      "step": 1875
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.2604611032341776e-05,
+      "loss": 0.1036,
+      "step": 1876
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.256159763304702e-05,
+      "loss": 0.0273,
+      "step": 1877
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.2518582332940516e-05,
+      "loss": 0.1128,
+      "step": 1878
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.247556516394142e-05,
+      "loss": 0.0979,
+      "step": 1879
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.243254615797025e-05,
+      "loss": 0.0801,
+      "step": 1880
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.238952534694889e-05,
+      "loss": 0.0678,
+      "step": 1881
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 5.2346502762800565e-05,
+      "loss": 0.0741,
+      "step": 1882
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2303478437449825e-05,
+      "loss": 0.0919,
+      "step": 1883
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.2260452402822504e-05,
+      "loss": 0.0787,
+      "step": 1884
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.221742469084569e-05,
+      "loss": 0.046,
+      "step": 1885
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.2174395333447746e-05,
+      "loss": 0.1003,
+      "step": 1886
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.2131364362558245e-05,
+      "loss": 0.1131,
+      "step": 1887
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 5.208833181010794e-05,
+      "loss": 0.0195,
+      "step": 1888
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.2045297708028776e-05,
+      "loss": 0.0398,
+      "step": 1889
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.200226208825384e-05,
+      "loss": 0.0385,
+      "step": 1890
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.195922498271737e-05,
+      "loss": 0.0706,
+      "step": 1891
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 5.191618642335463e-05,
+      "loss": 0.0665,
+      "step": 1892
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.187314644210207e-05,
+      "loss": 0.1017,
+      "step": 1893
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 5.183010507089713e-05,
+      "loss": 0.0523,
+      "step": 1894
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 5.1787062341678285e-05,
+      "loss": 0.08,
+      "step": 1895
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.174401828638501e-05,
+      "loss": 0.0494,
+      "step": 1896
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 5.1700972936957816e-05,
+      "loss": 0.0471,
+      "step": 1897
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.1657926325338115e-05,
+      "loss": 0.0503,
+      "step": 1898
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 5.161487848346828e-05,
+      "loss": 0.0493,
+      "step": 1899
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.157182944329161e-05,
+      "loss": 0.0604,
+      "step": 1900
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 5.152877923675226e-05,
+      "loss": 0.0785,
+      "step": 1901
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.148572789579529e-05,
+      "loss": 0.0575,
+      "step": 1902
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 5.144267545236655e-05,
+      "loss": 0.0802,
+      "step": 1903
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.1399621938412765e-05,
+      "loss": 0.0483,
+      "step": 1904
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 5.1356567385881436e-05,
+      "loss": 0.0415,
+      "step": 1905
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.131351182672082e-05,
+      "loss": 0.0664,
+      "step": 1906
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 5.127045529287992e-05,
+      "loss": 0.0287,
+      "step": 1907
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.122739781630846e-05,
+      "loss": 0.0418,
+      "step": 1908
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 5.118433942895691e-05,
+      "loss": 0.0751,
+      "step": 1909
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.114128016277634e-05,
+      "loss": 0.1086,
+      "step": 1910
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 5.1098220049718536e-05,
+      "loss": 0.0222,
+      "step": 1911
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 5.1055159121735894e-05,
+      "loss": 0.1281,
+      "step": 1912
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.10120974107814e-05,
+      "loss": 0.0814,
+      "step": 1913
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 5.096903494880862e-05,
+      "loss": 0.0422,
+      "step": 1914
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.09259717677717e-05,
+      "loss": 0.0445,
+      "step": 1915
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 5.08829078996253e-05,
+      "loss": 0.0725,
+      "step": 1916
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.0839843376324614e-05,
+      "loss": 0.0464,
+      "step": 1917
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 5.079677822982528e-05,
+      "loss": 0.063,
+      "step": 1918
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.0753712492083446e-05,
+      "loss": 0.0322,
+      "step": 1919
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 5.0710646195055677e-05,
+      "loss": 0.1091,
+      "step": 1920
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.066757937069891e-05,
+      "loss": 0.112,
+      "step": 1921
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 5.062451205097058e-05,
+      "loss": 0.037,
+      "step": 1922
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.058144426782838e-05,
+      "loss": 0.0168,
+      "step": 1923
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 5.053837605323042e-05,
+      "loss": 0.0531,
+      "step": 1924
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.049530743913507e-05,
+      "loss": 0.0726,
+      "step": 1925
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 5.0452238457501064e-05,
+      "loss": 0.0272,
+      "step": 1926
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 5.040916914028735e-05,
+      "loss": 0.0742,
+      "step": 1927
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.036609951945315e-05,
+      "loss": 0.0433,
+      "step": 1928
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 5.0323029626957916e-05,
+      "loss": 0.0478,
+      "step": 1929
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.02799594947613e-05,
+      "loss": 0.069,
+      "step": 1930
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 5.023688915482312e-05,
+      "loss": 0.041,
+      "step": 1931
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.0193818639103363e-05,
+      "loss": 0.0524,
+      "step": 1932
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 5.015074797956212e-05,
+      "loss": 0.0693,
+      "step": 1933
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.010767720815963e-05,
+      "loss": 0.0458,
+      "step": 1934
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 5.006460635685617e-05,
+      "loss": 0.1063,
+      "step": 1935
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 5.0021535457612136e-05,
+      "loss": 0.0643,
+      "step": 1936
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.997846454238787e-05,
+      "loss": 0.0408,
+      "step": 1937
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.9935393643143833e-05,
+      "loss": 0.0694,
+      "step": 1938
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.989232279184039e-05,
+      "loss": 0.0918,
+      "step": 1939
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.984925202043789e-05,
+      "loss": 0.0387,
+      "step": 1940
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.9806181360896655e-05,
+      "loss": 0.076,
+      "step": 1941
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.97631108451769e-05,
+      "loss": 0.0544,
+      "step": 1942
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9720040505238715e-05,
+      "loss": 0.0305,
+      "step": 1943
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.9676970373042096e-05,
+      "loss": 0.0515,
+      "step": 1944
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.963390048054685e-05,
+      "loss": 0.0649,
+      "step": 1945
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.959083085971266e-05,
+      "loss": 0.0558,
+      "step": 1946
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.9547761542498955e-05,
+      "loss": 0.0385,
+      "step": 1947
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.950469256086493e-05,
+      "loss": 0.0959,
+      "step": 1948
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.946162394676959e-05,
+      "loss": 0.0987,
+      "step": 1949
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.9418555732171626e-05,
+      "loss": 0.0795,
+      "step": 1950
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.937548794902942e-05,
+      "loss": 0.0351,
+      "step": 1951
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.933242062930109e-05,
+      "loss": 0.054,
+      "step": 1952
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.928935380494435e-05,
+      "loss": 0.056,
+      "step": 1953
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.924628750791656e-05,
+      "loss": 0.0672,
+      "step": 1954
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.9203221770174724e-05,
+      "loss": 0.089,
+      "step": 1955
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.916015662367539e-05,
+      "loss": 0.0615,
+      "step": 1956
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.911709210037471e-05,
+      "loss": 0.1145,
+      "step": 1957
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.907402823222832e-05,
+      "loss": 0.0457,
+      "step": 1958
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.903096505119139e-05,
+      "loss": 0.0764,
+      "step": 1959
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.898790258921861e-05,
+      "loss": 0.0899,
+      "step": 1960
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.8944840878264125e-05,
+      "loss": 0.1115,
+      "step": 1961
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.8901779950281455e-05,
+      "loss": 0.119,
+      "step": 1962
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.885871983722367e-05,
+      "loss": 0.1045,
+      "step": 1963
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8815660571043116e-05,
+      "loss": 0.0599,
+      "step": 1964
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.877260218369154e-05,
+      "loss": 0.0552,
+      "step": 1965
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8729544707120094e-05,
+      "loss": 0.0646,
+      "step": 1966
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.8686488173279206e-05,
+      "loss": 0.0872,
+      "step": 1967
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.864343261411857e-05,
+      "loss": 0.0624,
+      "step": 1968
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.860037806158724e-05,
+      "loss": 0.0648,
+      "step": 1969
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8557324547633454e-05,
+      "loss": 0.0532,
+      "step": 1970
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.8514272104204725e-05,
+      "loss": 0.1436,
+      "step": 1971
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.847122076324776e-05,
+      "loss": 0.0553,
+      "step": 1972
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.842817055670841e-05,
+      "loss": 0.1221,
+      "step": 1973
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.838512151653173e-05,
+      "loss": 0.0749,
+      "step": 1974
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.8342073674661904e-05,
+      "loss": 0.0649,
+      "step": 1975
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.829902706304219e-05,
+      "loss": 0.0503,
+      "step": 1976
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.825598171361501e-05,
+      "loss": 0.0912,
+      "step": 1977
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.821293765832175e-05,
+      "loss": 0.13,
+      "step": 1978
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.816989492910288e-05,
+      "loss": 0.0612,
+      "step": 1979
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.812685355789793e-05,
+      "loss": 0.0748,
+      "step": 1980
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8083813576645364e-05,
+      "loss": 0.104,
+      "step": 1981
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.8040775017282644e-05,
+      "loss": 0.1118,
+      "step": 1982
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.7997737911746174e-05,
+      "loss": 0.0468,
+      "step": 1983
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.795470229197123e-05,
+      "loss": 0.0895,
+      "step": 1984
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.7911668189892064e-05,
+      "loss": 0.1653,
+      "step": 1985
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.786863563744178e-05,
+      "loss": 0.0917,
+      "step": 1986
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.782560466655225e-05,
+      "loss": 0.0643,
+      "step": 1987
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.7782575309154324e-05,
+      "loss": 0.0476,
+      "step": 1988
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.773954759717752e-05,
+      "loss": 0.0213,
+      "step": 1989
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.769652156255019e-05,
+      "loss": 0.0854,
+      "step": 1990
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.7653497237199446e-05,
+      "loss": 0.0308,
+      "step": 1991
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.761047465305114e-05,
+      "loss": 0.0573,
+      "step": 1992
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.756745384202976e-05,
+      "loss": 0.0695,
+      "step": 1993
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.752443483605859e-05,
+      "loss": 0.0855,
+      "step": 1994
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.748141766705949e-05,
+      "loss": 0.0371,
+      "step": 1995
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.743840236695299e-05,
+      "loss": 0.0696,
+      "step": 1996
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.739538896765824e-05,
+      "loss": 0.0634,
+      "step": 1997
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.735237750109294e-05,
+      "loss": 0.1157,
+      "step": 1998
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.73093679991734e-05,
+      "loss": 0.0528,
+      "step": 1999
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.7266360493814465e-05,
+      "loss": 0.0416,
+      "step": 2000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.722335501692946e-05,
+      "loss": 0.1459,
+      "step": 2001
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.7180351600430274e-05,
+      "loss": 0.0355,
+      "step": 2002
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.7137350276227185e-05,
+      "loss": 0.0635,
+      "step": 2003
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.7094351076228974e-05,
+      "loss": 0.0364,
+      "step": 2004
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.7051354032342844e-05,
+      "loss": 0.0576,
+      "step": 2005
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.7008359176474364e-05,
+      "loss": 0.0499,
+      "step": 2006
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.6965366540527526e-05,
+      "loss": 0.0871,
+      "step": 2007
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.692237615640465e-05,
+      "loss": 0.0457,
+      "step": 2008
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.687938805600636e-05,
+      "loss": 0.0361,
+      "step": 2009
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.6836402271231615e-05,
+      "loss": 0.1117,
+      "step": 2010
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.679341883397768e-05,
+      "loss": 0.0468,
+      "step": 2011
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.6750437776139996e-05,
+      "loss": 0.044,
+      "step": 2012
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.670745912961236e-05,
+      "loss": 0.0745,
+      "step": 2013
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.666448292628666e-05,
+      "loss": 0.0587,
+      "step": 2014
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.6621509198053026e-05,
+      "loss": 0.0335,
+      "step": 2015
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.6578537976799766e-05,
+      "loss": 0.1359,
+      "step": 2016
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.6535569294413315e-05,
+      "loss": 0.0815,
+      "step": 2017
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.649260318277819e-05,
+      "loss": 0.1193,
+      "step": 2018
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.6449639673777044e-05,
+      "loss": 0.0694,
+      "step": 2019
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.640667879929058e-05,
+      "loss": 0.0435,
+      "step": 2020
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.6363720591197555e-05,
+      "loss": 0.0312,
+      "step": 2021
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.632076508137476e-05,
+      "loss": 0.0823,
+      "step": 2022
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.627781230169691e-05,
+      "loss": 0.0463,
+      "step": 2023
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.6234862284036815e-05,
+      "loss": 0.0742,
+      "step": 2024
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.619191506026514e-05,
+      "loss": 0.0685,
+      "step": 2025
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.614897066225049e-05,
+      "loss": 0.0919,
+      "step": 2026
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.6106029121859426e-05,
+      "loss": 0.0481,
+      "step": 2027
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.6063090470956325e-05,
+      "loss": 0.0532,
+      "step": 2028
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.602015474140345e-05,
+      "loss": 0.048,
+      "step": 2029
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.597722196506092e-05,
+      "loss": 0.0446,
+      "step": 2030
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.59342921737866e-05,
+      "loss": 0.0864,
+      "step": 2031
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.5891365399436204e-05,
+      "loss": 0.0366,
+      "step": 2032
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.5848441673863185e-05,
+      "loss": 0.0809,
+      "step": 2033
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.5805521028918685e-05,
+      "loss": 0.0925,
+      "step": 2034
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.576260349645165e-05,
+      "loss": 0.0593,
+      "step": 2035
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.571968910830866e-05,
+      "loss": 0.0533,
+      "step": 2036
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.567677789633394e-05,
+      "loss": 0.071,
+      "step": 2037
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.563386989236944e-05,
+      "loss": 0.0828,
+      "step": 2038
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.559096512825465e-05,
+      "loss": 0.0754,
+      "step": 2039
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.554806363582667e-05,
+      "loss": 0.0768,
+      "step": 2040
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.5505165446920214e-05,
+      "loss": 0.0618,
+      "step": 2041
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.5462270593367537e-05,
+      "loss": 0.0573,
+      "step": 2042
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.541937910699834e-05,
+      "loss": 0.0537,
+      "step": 2043
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.537649101963993e-05,
+      "loss": 0.0753,
+      "step": 2044
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.533360636311703e-05,
+      "loss": 0.0688,
+      "step": 2045
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.529072516925186e-05,
+      "loss": 0.0918,
+      "step": 2046
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.524784746986403e-05,
+      "loss": 0.0474,
+      "step": 2047
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.5204973296770534e-05,
+      "loss": 0.0669,
+      "step": 2048
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.5162102681785863e-05,
+      "loss": 0.0652,
+      "step": 2049
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.5119235656721736e-05,
+      "loss": 0.0755,
+      "step": 2050
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.5076372253387275e-05,
+      "loss": 0.062,
+      "step": 2051
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.5033512503588925e-05,
+      "loss": 0.0625,
+      "step": 2052
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4990656439130376e-05,
+      "loss": 0.0505,
+      "step": 2053
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.49478040918126e-05,
+      "loss": 0.0413,
+      "step": 2054
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.490495549343384e-05,
+      "loss": 0.0413,
+      "step": 2055
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.486211067578952e-05,
+      "loss": 0.1002,
+      "step": 2056
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.481926967067228e-05,
+      "loss": 0.081,
+      "step": 2057
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.477643250987194e-05,
+      "loss": 0.0803,
+      "step": 2058
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.473359922517541e-05,
+      "loss": 0.0385,
+      "step": 2059
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4690769848366785e-05,
+      "loss": 0.0768,
+      "step": 2060
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.464794441122725e-05,
+      "loss": 0.0986,
+      "step": 2061
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.460512294553503e-05,
+      "loss": 0.08,
+      "step": 2062
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.456230548306547e-05,
+      "loss": 0.08,
+      "step": 2063
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451949205559087e-05,
+      "loss": 0.0578,
+      "step": 2064
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.447668269488057e-05,
+      "loss": 0.0269,
+      "step": 2065
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.443387743270092e-05,
+      "loss": 0.0617,
+      "step": 2066
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.439107630081518e-05,
+      "loss": 0.0409,
+      "step": 2067
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.434827933098356e-05,
+      "loss": 0.0341,
+      "step": 2068
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.430548655496319e-05,
+      "loss": 0.0882,
+      "step": 2069
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.42626980045081e-05,
+      "loss": 0.0516,
+      "step": 2070
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.421991371136916e-05,
+      "loss": 0.0665,
+      "step": 2071
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.417713370729411e-05,
+      "loss": 0.0835,
+      "step": 2072
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.413435802402743e-05,
+      "loss": 0.0469,
+      "step": 2073
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.409158669331054e-05,
+      "loss": 0.0525,
+      "step": 2074
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.404881974688147e-05,
+      "loss": 0.021,
+      "step": 2075
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.4006057216475087e-05,
+      "loss": 0.0545,
+      "step": 2076
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.396329913382298e-05,
+      "loss": 0.0721,
+      "step": 2077
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.3920545530653426e-05,
+      "loss": 0.0937,
+      "step": 2078
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.387779643869133e-05,
+      "loss": 0.1028,
+      "step": 2079
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.383505188965834e-05,
+      "loss": 0.08,
+      "step": 2080
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.3792311915272646e-05,
+      "loss": 0.0509,
+      "step": 2081
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.374957654724911e-05,
+      "loss": 0.1174,
+      "step": 2082
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.370684581729915e-05,
+      "loss": 0.0891,
+      "step": 2083
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.366411975713071e-05,
+      "loss": 0.0379,
+      "step": 2084
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.3621398398448334e-05,
+      "loss": 0.0848,
+      "step": 2085
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.357868177295302e-05,
+      "loss": 0.0269,
+      "step": 2086
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.353596991234228e-05,
+      "loss": 0.035,
+      "step": 2087
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.349326284831011e-05,
+      "loss": 0.0477,
+      "step": 2088
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.3450560612546896e-05,
+      "loss": 0.0507,
+      "step": 2089
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.340786323673947e-05,
+      "loss": 0.0472,
+      "step": 2090
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.3365170752571075e-05,
+      "loss": 0.0495,
+      "step": 2091
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.332248319172131e-05,
+      "loss": 0.0112,
+      "step": 2092
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.327980058586608e-05,
+      "loss": 0.0893,
+      "step": 2093
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.323712296667768e-05,
+      "loss": 0.043,
+      "step": 2094
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.319445036582465e-05,
+      "loss": 0.0696,
+      "step": 2095
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.315178281497186e-05,
+      "loss": 0.0645,
+      "step": 2096
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.310912034578041e-05,
+      "loss": 0.0839,
+      "step": 2097
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.3066462989907566e-05,
+      "loss": 0.0988,
+      "step": 2098
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.302381077900692e-05,
+      "loss": 0.0585,
+      "step": 2099
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.2981163744728145e-05,
+      "loss": 0.0209,
+      "step": 2100
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.2938521918717115e-05,
+      "loss": 0.082,
+      "step": 2101
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.289588533261585e-05,
+      "loss": 0.0525,
+      "step": 2102
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.285325401806246e-05,
+      "loss": 0.0668,
+      "step": 2103
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.281062800669112e-05,
+      "loss": 0.054,
+      "step": 2104
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.276800733013215e-05,
+      "loss": 0.037,
+      "step": 2105
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.272539202001181e-05,
+      "loss": 0.0439,
+      "step": 2106
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.2682782107952474e-05,
+      "loss": 0.0587,
+      "step": 2107
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.2640177625572445e-05,
+      "loss": 0.0879,
+      "step": 2108
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.2597578604486005e-05,
+      "loss": 0.034,
+      "step": 2109
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.255498507630341e-05,
+      "loss": 0.1043,
+      "step": 2110
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.251239707263083e-05,
+      "loss": 0.0701,
+      "step": 2111
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.24698146250703e-05,
+      "loss": 0.1079,
+      "step": 2112
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.242723776521981e-05,
+      "loss": 0.0827,
+      "step": 2113
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.238466652467312e-05,
+      "loss": 0.0843,
+      "step": 2114
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.2342100935019866e-05,
+      "loss": 0.0516,
+      "step": 2115
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.229954102784549e-05,
+      "loss": 0.1667,
+      "step": 2116
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.2256986834731224e-05,
+      "loss": 0.0309,
+      "step": 2117
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.221443838725398e-05,
+      "loss": 0.0404,
+      "step": 2118
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.217189571698655e-05,
+      "loss": 0.0626,
+      "step": 2119
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.212935885549732e-05,
+      "loss": 0.0677,
+      "step": 2120
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.208682783435041e-05,
+      "loss": 0.0472,
+      "step": 2121
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.204430268510563e-05,
+      "loss": 0.0475,
+      "step": 2122
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.2001783439318336e-05,
+      "loss": 0.0672,
+      "step": 2123
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.195927012853966e-05,
+      "loss": 0.0699,
+      "step": 2124
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.191676278431619e-05,
+      "loss": 0.0202,
+      "step": 2125
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.1874261438190133e-05,
+      "loss": 0.0539,
+      "step": 2126
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.1831766121699283e-05,
+      "loss": 0.0589,
+      "step": 2127
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.178927686637693e-05,
+      "loss": 0.0508,
+      "step": 2128
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.174679370375182e-05,
+      "loss": 0.052,
+      "step": 2129
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.1704316665348267e-05,
+      "loss": 0.0468,
+      "step": 2130
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.1661845782685966e-05,
+      "loss": 0.0872,
+      "step": 2131
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.1619381087280096e-05,
+      "loss": 0.0787,
+      "step": 2132
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.1576922610641236e-05,
+      "loss": 0.0257,
+      "step": 2133
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.1534470384275285e-05,
+      "loss": 0.056,
+      "step": 2134
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.149202443968361e-05,
+      "loss": 0.0683,
+      "step": 2135
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.144958480836283e-05,
+      "loss": 0.0566,
+      "step": 2136
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.140715152180492e-05,
+      "loss": 0.1168,
+      "step": 2137
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.136472461149716e-05,
+      "loss": 0.0388,
+      "step": 2138
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.132230410892206e-05,
+      "loss": 0.0797,
+      "step": 2139
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.127989004555738e-05,
+      "loss": 0.0647,
+      "step": 2140
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.1237482452876144e-05,
+      "loss": 0.0546,
+      "step": 2141
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.1195081362346535e-05,
+      "loss": 0.0861,
+      "step": 2142
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.115268680543189e-05,
+      "loss": 0.0511,
+      "step": 2143
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.1110298813590785e-05,
+      "loss": 0.0302,
+      "step": 2144
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.106791741827682e-05,
+      "loss": 0.0405,
+      "step": 2145
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.1025542650938786e-05,
+      "loss": 0.0681,
+      "step": 2146
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.0983174543020517e-05,
+      "loss": 0.0483,
+      "step": 2147
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.0940813125960855e-05,
+      "loss": 0.035,
+      "step": 2148
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.08984584311938e-05,
+      "loss": 0.0342,
+      "step": 2149
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.085611049014825e-05,
+      "loss": 0.0561,
+      "step": 2150
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.081376933424813e-05,
+      "loss": 0.052,
+      "step": 2151
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.0771434994912364e-05,
+      "loss": 0.023,
+      "step": 2152
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.072910750355478e-05,
+      "loss": 0.0531,
+      "step": 2153
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.068678689158411e-05,
+      "loss": 0.0341,
+      "step": 2154
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.064447319040403e-05,
+      "loss": 0.0263,
+      "step": 2155
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.0602166431413046e-05,
+      "loss": 0.0211,
+      "step": 2156
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.055986664600454e-05,
+      "loss": 0.0348,
+      "step": 2157
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.051757386556672e-05,
+      "loss": 0.0269,
+      "step": 2158
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.0475288121482545e-05,
+      "loss": 0.0439,
+      "step": 2159
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.043300944512983e-05,
+      "loss": 0.0243,
+      "step": 2160
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.0390737867881085e-05,
+      "loss": 0.0323,
+      "step": 2161
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.034847342110357e-05,
+      "loss": 0.0258,
+      "step": 2162
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.0306216136159294e-05,
+      "loss": 0.0615,
+      "step": 2163
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.026396604440488e-05,
+      "loss": 0.042,
+      "step": 2164
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.0221723177191653e-05,
+      "loss": 0.0453,
+      "step": 2165
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.0179487565865586e-05,
+      "loss": 0.0238,
+      "step": 2166
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.0137259241767266e-05,
+      "loss": 0.0563,
+      "step": 2167
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.0095038236231816e-05,
+      "loss": 0.0507,
+      "step": 2168
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.0052824580589026e-05,
+      "loss": 0.0409,
+      "step": 2169
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.0010618306163144e-05,
+      "loss": 0.0745,
+      "step": 2170
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.9968419444273e-05,
+      "loss": 0.0436,
+      "step": 2171
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.992622802623189e-05,
+      "loss": 0.0695,
+      "step": 2172
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.9884044083347595e-05,
+      "loss": 0.0447,
+      "step": 2173
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.984186764692237e-05,
+      "loss": 0.0528,
+      "step": 2174
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.979969874825286e-05,
+      "loss": 0.0618,
+      "step": 2175
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.975753741863013e-05,
+      "loss": 0.0379,
+      "step": 2176
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.971538368933966e-05,
+      "loss": 0.0367,
+      "step": 2177
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.967323759166127e-05,
+      "loss": 0.0485,
+      "step": 2178
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.963109915686907e-05,
+      "loss": 0.0359,
+      "step": 2179
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.958896841623157e-05,
+      "loss": 0.0375,
+      "step": 2180
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.954684540101151e-05,
+      "loss": 0.046,
+      "step": 2181
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.9504730142465924e-05,
+      "loss": 0.0698,
+      "step": 2182
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.946262267184608e-05,
+      "loss": 0.0479,
+      "step": 2183
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.942052302039745e-05,
+      "loss": 0.0588,
+      "step": 2184
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.9378431219359755e-05,
+      "loss": 0.0427,
+      "step": 2185
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.933634729996684e-05,
+      "loss": 0.0725,
+      "step": 2186
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.92942712934467e-05,
+      "loss": 0.095,
+      "step": 2187
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.925220323102152e-05,
+      "loss": 0.0339,
+      "step": 2188
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.921014314390753e-05,
+      "loss": 0.0229,
+      "step": 2189
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.9168091063315026e-05,
+      "loss": 0.0654,
+      "step": 2190
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.912604702044844e-05,
+      "loss": 0.057,
+      "step": 2191
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.9084011046506206e-05,
+      "loss": 0.0734,
+      "step": 2192
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.9041983172680694e-05,
+      "loss": 0.0283,
+      "step": 2193
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.899996343015841e-05,
+      "loss": 0.112,
+      "step": 2194
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.895795185011969e-05,
+      "loss": 0.0452,
+      "step": 2195
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.89159484637389e-05,
+      "loss": 0.0924,
+      "step": 2196
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.887395330218429e-05,
+      "loss": 0.0433,
+      "step": 2197
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.883196639661799e-05,
+      "loss": 0.0893,
+      "step": 2198
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.878998777819607e-05,
+      "loss": 0.0278,
+      "step": 2199
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.8748017478068366e-05,
+      "loss": 0.0687,
+      "step": 2200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.870605552737859e-05,
+      "loss": 0.0719,
+      "step": 2201
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.8664101957264274e-05,
+      "loss": 0.0109,
+      "step": 2202
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.86221567988567e-05,
+      "loss": 0.0189,
+      "step": 2203
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.858022008328089e-05,
+      "loss": 0.0449,
+      "step": 2204
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.853829184165565e-05,
+      "loss": 0.0924,
+      "step": 2205
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.849637210509346e-05,
+      "loss": 0.0278,
+      "step": 2206
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.845446090470053e-05,
+      "loss": 0.0462,
+      "step": 2207
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.8412558271576704e-05,
+      "loss": 0.0407,
+      "step": 2208
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.8370664236815426e-05,
+      "loss": 0.031,
+      "step": 2209
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.8328778831503883e-05,
+      "loss": 0.037,
+      "step": 2210
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.828690208672274e-05,
+      "loss": 0.0482,
+      "step": 2211
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.8245034033546276e-05,
+      "loss": 0.024,
+      "step": 2212
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.820317470304234e-05,
+      "loss": 0.0842,
+      "step": 2213
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.8161324126272306e-05,
+      "loss": 0.0377,
+      "step": 2214
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.8119482334291e-05,
+      "loss": 0.0677,
+      "step": 2215
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.807764935814679e-05,
+      "loss": 0.0649,
+      "step": 2216
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.80358252288815e-05,
+      "loss": 0.0725,
+      "step": 2217
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.79940099775303e-05,
+      "loss": 0.0235,
+      "step": 2218
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.795220363512193e-05,
+      "loss": 0.0577,
+      "step": 2219
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.7910406232678363e-05,
+      "loss": 0.06,
+      "step": 2220
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.786861780121504e-05,
+      "loss": 0.0604,
+      "step": 2221
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.7826838371740716e-05,
+      "loss": 0.0378,
+      "step": 2222
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.778506797525744e-05,
+      "loss": 0.0679,
+      "step": 2223
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.77433066427606e-05,
+      "loss": 0.0067,
+      "step": 2224
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.7701554405238816e-05,
+      "loss": 0.0517,
+      "step": 2225
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.7659811293673985e-05,
+      "loss": 0.0176,
+      "step": 2226
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.761807733904124e-05,
+      "loss": 0.0955,
+      "step": 2227
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.7576352572308914e-05,
+      "loss": 0.1299,
+      "step": 2228
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.7534637024438485e-05,
+      "loss": 0.0385,
+      "step": 2229
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.749293072638465e-05,
+      "loss": 0.0255,
+      "step": 2230
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.745123370909518e-05,
+      "loss": 0.0524,
+      "step": 2231
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.740954600351103e-05,
+      "loss": 0.0558,
+      "step": 2232
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.7367867640566195e-05,
+      "loss": 0.0566,
+      "step": 2233
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.732619865118772e-05,
+      "loss": 0.0933,
+      "step": 2234
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.7284539066295777e-05,
+      "loss": 0.0319,
+      "step": 2235
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.724288891680347e-05,
+      "loss": 0.0292,
+      "step": 2236
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.720124823361693e-05,
+      "loss": 0.0685,
+      "step": 2237
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.7159617047635306e-05,
+      "loss": 0.0293,
+      "step": 2238
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.7117995389750656e-05,
+      "loss": 0.0504,
+      "step": 2239
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.707638329084795e-05,
+      "loss": 0.0442,
+      "step": 2240
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.703478078180512e-05,
+      "loss": 0.041,
+      "step": 2241
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.6993187893492956e-05,
+      "loss": 0.0666,
+      "step": 2242
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.695160465677507e-05,
+      "loss": 0.0922,
+      "step": 2243
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.6910031102508004e-05,
+      "loss": 0.0488,
+      "step": 2244
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.686846726154101e-05,
+      "loss": 0.0788,
+      "step": 2245
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.6826913164716216e-05,
+      "loss": 0.0487,
+      "step": 2246
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.678536884286848e-05,
+      "loss": 0.0871,
+      "step": 2247
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.674383432682539e-05,
+      "loss": 0.0193,
+      "step": 2248
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.670230964740732e-05,
+      "loss": 0.0751,
+      "step": 2249
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.666079483542728e-05,
+      "loss": 0.0278,
+      "step": 2250
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.661928992169097e-05,
+      "loss": 0.0306,
+      "step": 2251
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.657779493699678e-05,
+      "loss": 0.0309,
+      "step": 2252
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.6536309912135726e-05,
+      "loss": 0.024,
+      "step": 2253
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.649483487789137e-05,
+      "loss": 0.0415,
+      "step": 2254
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.645336986503995e-05,
+      "loss": 0.0606,
+      "step": 2255
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.64119149043502e-05,
+      "loss": 0.0627,
+      "step": 2256
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.6370470026583436e-05,
+      "loss": 0.0791,
+      "step": 2257
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.632903526249349e-05,
+      "loss": 0.0047,
+      "step": 2258
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3.628761064282663e-05,
+      "loss": 0.0225,
+      "step": 2259
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 3.62461961983217e-05,
+      "loss": 0.0476,
+      "step": 2260
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3.62047919597099e-05,
+      "loss": 0.0316,
+      "step": 2261
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3.61633979577149e-05,
+      "loss": 0.0556,
+      "step": 2262
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3.6122014223052766e-05,
+      "loss": 0.0208,
+      "step": 2263
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3.608064078643197e-05,
+      "loss": 0.0417,
+      "step": 2264
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3.603927767855328e-05,
+      "loss": 0.0164,
+      "step": 2265
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3.599792493010986e-05,
+      "loss": 0.0641,
+      "step": 2266
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.595658257178717e-05,
+      "loss": 0.0514,
+      "step": 2267
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.591525063426293e-05,
+      "loss": 0.0311,
+      "step": 2268
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3.5873929148207186e-05,
+      "loss": 0.0161,
+      "step": 2269
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 3.583261814428217e-05,
+      "loss": 0.0311,
+      "step": 2270
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3.579131765314237e-05,
+      "loss": 0.0379,
+      "step": 2271
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.575002770543447e-05,
+      "loss": 0.0291,
+      "step": 2272
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.570874833179732e-05,
+      "loss": 0.0469,
+      "step": 2273
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3.566747956286194e-05,
+      "loss": 0.0201,
+      "step": 2274
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3.562622142925144e-05,
+      "loss": 0.0679,
+      "step": 2275
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.558497396158108e-05,
+      "loss": 0.0438,
+      "step": 2276
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.55437371904582e-05,
+      "loss": 0.0327,
+      "step": 2277
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.550251114648221e-05,
+      "loss": 0.0506,
+      "step": 2278
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.546129586024449e-05,
+      "loss": 0.0194,
+      "step": 2279
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3.542009136232857e-05,
+      "loss": 0.0568,
+      "step": 2280
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3.537889768330983e-05,
+      "loss": 0.0654,
+      "step": 2281
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.533771485375573e-05,
+      "loss": 0.0339,
+      "step": 2282
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.5296542904225626e-05,
+      "loss": 0.0316,
+      "step": 2283
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.525538186527079e-05,
+      "loss": 0.0505,
+      "step": 2284
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.5214231767434467e-05,
+      "loss": 0.0555,
+      "step": 2285
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3.51730926412517e-05,
+      "loss": 0.0778,
+      "step": 2286
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3.513196451724944e-05,
+      "loss": 0.0426,
+      "step": 2287
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 3.5090847425946464e-05,
+      "loss": 0.0464,
+      "step": 2288
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3.504974139785338e-05,
+      "loss": 0.09,
+      "step": 2289
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3.50086464634725e-05,
+      "loss": 0.0435,
+      "step": 2290
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.496756265329805e-05,
+      "loss": 0.0269,
+      "step": 2291
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.492648999781587e-05,
+      "loss": 0.0189,
+      "step": 2292
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3.4885428527503584e-05,
+      "loss": 0.0616,
+      "step": 2293
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3.484437827283054e-05,
+      "loss": 0.0521,
+      "step": 2294
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 3.4803339264257684e-05,
+      "loss": 0.0424,
+      "step": 2295
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 3.4762311532237693e-05,
+      "loss": 0.0183,
+      "step": 2296
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.472129510721484e-05,
+      "loss": 0.0392,
+      "step": 2297
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.4680290019624994e-05,
+      "loss": 0.0361,
+      "step": 2298
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3.4639296299895675e-05,
+      "loss": 0.0292,
+      "step": 2299
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3.459831397844589e-05,
+      "loss": 0.0346,
+      "step": 2300
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 3.455734308568621e-05,
+      "loss": 0.0703,
+      "step": 2301
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 3.451638365201878e-05,
+      "loss": 0.0396,
+      "step": 2302
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 3.447543570783719e-05,
+      "loss": 0.0189,
+      "step": 2303
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 3.443449928352646e-05,
+      "loss": 0.0157,
+      "step": 2304
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 3.439357440946319e-05,
+      "loss": 0.0372,
+      "step": 2305
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3.4352661116015284e-05,
+      "loss": 0.0287,
+      "step": 2306
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3.431175943354213e-05,
+      "loss": 0.0475,
+      "step": 2307
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3.427086939239447e-05,
+      "loss": 0.0252,
+      "step": 2308
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3.422999102291439e-05,
+      "loss": 0.0321,
+      "step": 2309
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 3.418912435543537e-05,
+      "loss": 0.0308,
+      "step": 2310
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 3.414826942028215e-05,
+      "loss": 0.0662,
+      "step": 2311
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3.410742624777078e-05,
+      "loss": 0.0231,
+      "step": 2312
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3.4066594868208615e-05,
+      "loss": 0.0303,
+      "step": 2313
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3.402577531189423e-05,
+      "loss": 0.0207,
+      "step": 2314
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3.398496760911739e-05,
+      "loss": 0.0525,
+      "step": 2315
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3.394417179015915e-05,
+      "loss": 0.01,
+      "step": 2316
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3.390338788529168e-05,
+      "loss": 0.0278,
+      "step": 2317
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 3.3862615924778325e-05,
+      "loss": 0.0443,
+      "step": 2318
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3.38218559388736e-05,
+      "loss": 0.0258,
+      "step": 2319
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3.378110795782305e-05,
+      "loss": 0.026,
+      "step": 2320
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 3.374037201186344e-05,
+      "loss": 0.0083,
+      "step": 2321
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 3.369964813122248e-05,
+      "loss": 0.0817,
+      "step": 2322
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3.365893634611898e-05,
+      "loss": 0.0438,
+      "step": 2323
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3.361823668676282e-05,
+      "loss": 0.0272,
+      "step": 2324
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 3.357754918335479e-05,
+      "loss": 0.0206,
+      "step": 2325
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 3.3536873866086716e-05,
+      "loss": 0.0869,
+      "step": 2326
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3.3496210765141375e-05,
+      "loss": 0.0388,
+      "step": 2327
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3.3455559910692494e-05,
+      "loss": 0.0491,
+      "step": 2328
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3.3414921332904645e-05,
+      "loss": 0.0872,
+      "step": 2329
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3.337429506193339e-05,
+      "loss": 0.0239,
+      "step": 2330
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 3.333368112792508e-05,
+      "loss": 0.0231,
+      "step": 2331
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 3.3293079561016956e-05,
+      "loss": 0.0508,
+      "step": 2332
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3.3252490391337055e-05,
+      "loss": 0.0111,
+      "step": 2333
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3.321191364900421e-05,
+      "loss": 0.0445,
+      "step": 2334
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 3.317134936412809e-05,
+      "loss": 0.0234,
+      "step": 2335
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 3.313079756680904e-05,
+      "loss": 0.0172,
+      "step": 2336
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 3.3090258287138175e-05,
+      "loss": 0.0274,
+      "step": 2337
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3.304973155519735e-05,
+      "loss": 0.0575,
+      "step": 2338
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3.300921740105908e-05,
+      "loss": 0.0442,
+      "step": 2339
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3.296871585478651e-05,
+      "loss": 0.0638,
+      "step": 2340
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3.292822694643351e-05,
+      "loss": 0.1098,
+      "step": 2341
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.2887750706044513e-05,
+      "loss": 0.0647,
+      "step": 2342
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.2847287163654556e-05,
+      "loss": 0.0414,
+      "step": 2343
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.280683634928931e-05,
+      "loss": 0.0327,
+      "step": 2344
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.2766398292964884e-05,
+      "loss": 0.0189,
+      "step": 2345
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.2725973024688074e-05,
+      "loss": 0.065,
+      "step": 2346
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.2685560574456066e-05,
+      "loss": 0.0443,
+      "step": 2347
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.264516097225656e-05,
+      "loss": 0.0255,
+      "step": 2348
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.260477424806779e-05,
+      "loss": 0.0469,
+      "step": 2349
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.256440043185833e-05,
+      "loss": 0.0556,
+      "step": 2350
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.2524039553587235e-05,
+      "loss": 0.0312,
+      "step": 2351
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.248369164320397e-05,
+      "loss": 0.0405,
+      "step": 2352
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.244335673064836e-05,
+      "loss": 0.0861,
+      "step": 2353
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.2403034845850544e-05,
+      "loss": 0.0728,
+      "step": 2354
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.23627260187311e-05,
+      "loss": 0.0461,
+      "step": 2355
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.2322430279200785e-05,
+      "loss": 0.0722,
+      "step": 2356
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.2282147657160763e-05,
+      "loss": 0.0597,
+      "step": 2357
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.224187818250237e-05,
+      "loss": 0.0286,
+      "step": 2358
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.2201621885107245e-05,
+      "loss": 0.0321,
+      "step": 2359
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.2161378794847254e-05,
+      "loss": 0.0376,
+      "step": 2360
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.21211489415844e-05,
+      "loss": 0.0743,
+      "step": 2361
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.2080932355170904e-05,
+      "loss": 0.022,
+      "step": 2362
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.204072906544917e-05,
+      "loss": 0.0389,
+      "step": 2363
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 3.2000539102251716e-05,
+      "loss": 0.0281,
+      "step": 2364
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.1960362495401106e-05,
+      "loss": 0.0608,
+      "step": 2365
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.192019927471011e-05,
+      "loss": 0.0252,
+      "step": 2366
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.188004946998146e-05,
+      "loss": 0.0827,
+      "step": 2367
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.1839913111008e-05,
+      "loss": 0.0297,
+      "step": 2368
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.179979022757259e-05,
+      "loss": 0.0311,
+      "step": 2369
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.175968084944802e-05,
+      "loss": 0.0524,
+      "step": 2370
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.171958500639718e-05,
+      "loss": 0.0342,
+      "step": 2371
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.1679502728172804e-05,
+      "loss": 0.0431,
+      "step": 2372
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.163943404451761e-05,
+      "loss": 0.0312,
+      "step": 2373
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.159937898516424e-05,
+      "loss": 0.0192,
+      "step": 2374
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.1559337579835214e-05,
+      "loss": 0.013,
+      "step": 2375
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.1519309858242886e-05,
+      "loss": 0.0433,
+      "step": 2376
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.147929585008951e-05,
+      "loss": 0.0321,
+      "step": 2377
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.143929558506715e-05,
+      "loss": 0.0509,
+      "step": 2378
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.139930909285762e-05,
+      "loss": 0.0347,
+      "step": 2379
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.1359336403132614e-05,
+      "loss": 0.0347,
+      "step": 2380
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.131937754555347e-05,
+      "loss": 0.0788,
+      "step": 2381
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.127943254977136e-05,
+      "loss": 0.0249,
+      "step": 2382
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.123950144542711e-05,
+      "loss": 0.0427,
+      "step": 2383
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.119958426215126e-05,
+      "loss": 0.0659,
+      "step": 2384
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.1159681029564025e-05,
+      "loss": 0.038,
+      "step": 2385
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.111979177727524e-05,
+      "loss": 0.0348,
+      "step": 2386
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.107991653488439e-05,
+      "loss": 0.0321,
+      "step": 2387
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.1040055331980576e-05,
+      "loss": 0.0387,
+      "step": 2388
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.100020819814247e-05,
+      "loss": 0.0528,
+      "step": 2389
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.096037516293826e-05,
+      "loss": 0.0376,
+      "step": 2390
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.092055625592574e-05,
+      "loss": 0.0613,
+      "step": 2391
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.088075150665219e-05,
+      "loss": 0.0335,
+      "step": 2392
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.084096094465437e-05,
+      "loss": 0.0344,
+      "step": 2393
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.0801184599458555e-05,
+      "loss": 0.0316,
+      "step": 2394
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.07614225005804e-05,
+      "loss": 0.0121,
+      "step": 2395
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.072167467752509e-05,
+      "loss": 0.0304,
+      "step": 2396
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.068194115978712e-05,
+      "loss": 0.0375,
+      "step": 2397
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.064222197685041e-05,
+      "loss": 0.0822,
+      "step": 2398
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 3.060251715818826e-05,
+      "loss": 0.0562,
+      "step": 2399
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.056282673326329e-05,
+      "loss": 0.0333,
+      "step": 2400
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.052315073152742e-05,
+      "loss": 0.0154,
+      "step": 2401
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.048348918242191e-05,
+      "loss": 0.0484,
+      "step": 2402
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.0443842115377275e-05,
+      "loss": 0.0367,
+      "step": 2403
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.0404209559813264e-05,
+      "loss": 0.0514,
+      "step": 2404
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 3.0364591545138908e-05,
+      "loss": 0.0257,
+      "step": 2405
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.0324988100752393e-05,
+      "loss": 0.0305,
+      "step": 2406
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.0285399256041137e-05,
+      "loss": 0.0191,
+      "step": 2407
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.0245825040381692e-05,
+      "loss": 0.1038,
+      "step": 2408
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.0206265483139763e-05,
+      "loss": 0.0354,
+      "step": 2409
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.0166720613670212e-05,
+      "loss": 0.0373,
+      "step": 2410
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 3.0127190461316946e-05,
+      "loss": 0.0324,
+      "step": 2411
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.0087675055412978e-05,
+      "loss": 0.0223,
+      "step": 2412
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.0048174425280396e-05,
+      "loss": 0.0318,
+      "step": 2413
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.00086886002303e-05,
+      "loss": 0.0253,
+      "step": 2414
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.9969217609562782e-05,
+      "loss": 0.0766,
+      "step": 2415
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.9929761482567016e-05,
+      "loss": 0.0145,
+      "step": 2416
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.9890320248521038e-05,
+      "loss": 0.0587,
+      "step": 2417
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.9850893936691883e-05,
+      "loss": 0.0269,
+      "step": 2418
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.9811482576335537e-05,
+      "loss": 0.0291,
+      "step": 2419
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.977208619669682e-05,
+      "loss": 0.0248,
+      "step": 2420
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.9732704827009537e-05,
+      "loss": 0.0364,
+      "step": 2421
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.969333849649626e-05,
+      "loss": 0.0341,
+      "step": 2422
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.9653987234368442e-05,
+      "loss": 0.0692,
+      "step": 2423
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.9614651069826373e-05,
+      "loss": 0.0108,
+      "step": 2424
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.9575330032059113e-05,
+      "loss": 0.0859,
+      "step": 2425
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.9536024150244474e-05,
+      "loss": 0.0762,
+      "step": 2426
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.9496733453549086e-05,
+      "loss": 0.0342,
+      "step": 2427
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.9457457971128268e-05,
+      "loss": 0.0412,
+      "step": 2428
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.941819773212604e-05,
+      "loss": 0.063,
+      "step": 2429
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.937895276567516e-05,
+      "loss": 0.1143,
+      "step": 2430
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.9339723100896992e-05,
+      "loss": 0.0365,
+      "step": 2431
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.9300508766901598e-05,
+      "loss": 0.08,
+      "step": 2432
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.9261309792787628e-05,
+      "loss": 0.0394,
+      "step": 2433
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.9222126207642335e-05,
+      "loss": 0.087,
+      "step": 2434
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.918295804054161e-05,
+      "loss": 0.0497,
+      "step": 2435
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.914380532054981e-05,
+      "loss": 0.0464,
+      "step": 2436
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.9104668076719876e-05,
+      "loss": 0.0297,
+      "step": 2437
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.906554633809333e-05,
+      "loss": 0.0304,
+      "step": 2438
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.9026440133700068e-05,
+      "loss": 0.0594,
+      "step": 2439
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8987349492558502e-05,
+      "loss": 0.0749,
+      "step": 2440
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.894827444367556e-05,
+      "loss": 0.0456,
+      "step": 2441
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8909215016046553e-05,
+      "loss": 0.0495,
+      "step": 2442
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8870171238655132e-05,
+      "loss": 0.0326,
+      "step": 2443
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.883114314047346e-05,
+      "loss": 0.0446,
+      "step": 2444
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8792130750461986e-05,
+      "loss": 0.0274,
+      "step": 2445
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8753134097569522e-05,
+      "loss": 0.0176,
+      "step": 2446
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8714153210733197e-05,
+      "loss": 0.0475,
+      "step": 2447
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8675188118878438e-05,
+      "loss": 0.0482,
+      "step": 2448
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.863623885091902e-05,
+      "loss": 0.0256,
+      "step": 2449
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8597305435756837e-05,
+      "loss": 0.0508,
+      "step": 2450
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8558387902282123e-05,
+      "loss": 0.0148,
+      "step": 2451
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.851948627937335e-05,
+      "loss": 0.0111,
+      "step": 2452
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8480600595897073e-05,
+      "loss": 0.0112,
+      "step": 2453
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.844173088070809e-05,
+      "loss": 0.0522,
+      "step": 2454
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8402877162649376e-05,
+      "loss": 0.1005,
+      "step": 2455
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8364039470551994e-05,
+      "loss": 0.0515,
+      "step": 2456
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8325217833235117e-05,
+      "loss": 0.0153,
+      "step": 2457
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8286412279506002e-05,
+      "loss": 0.0235,
+      "step": 2458
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8247622838159993e-05,
+      "loss": 0.0086,
+      "step": 2459
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820884953798045e-05,
+      "loss": 0.0516,
+      "step": 2460
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8170092407738792e-05,
+      "loss": 0.0285,
+      "step": 2461
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.813135147619438e-05,
+      "loss": 0.0302,
+      "step": 2462
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8092626772094666e-05,
+      "loss": 0.0128,
+      "step": 2463
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.805391832417491e-05,
+      "loss": 0.0538,
+      "step": 2464
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8015226161158403e-05,
+      "loss": 0.038,
+      "step": 2465
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.7976550311756367e-05,
+      "loss": 0.0189,
+      "step": 2466
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.7937890804667888e-05,
+      "loss": 0.0302,
+      "step": 2467
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.789924766857987e-05,
+      "loss": 0.0426,
+      "step": 2468
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.7860620932167175e-05,
+      "loss": 0.0151,
+      "step": 2469
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.7822010624092432e-05,
+      "loss": 0.0485,
+      "step": 2470
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.778341677300609e-05,
+      "loss": 0.0434,
+      "step": 2471
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.7744839407546375e-05,
+      "loss": 0.0557,
+      "step": 2472
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.7706278556339292e-05,
+      "loss": 0.0396,
+      "step": 2473
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.766773424799865e-05,
+      "loss": 0.0181,
+      "step": 2474
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.7629206511125853e-05,
+      "loss": 0.0417,
+      "step": 2475
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.7590695374310093e-05,
+      "loss": 0.0144,
+      "step": 2476
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.7552200866128282e-05,
+      "loss": 0.0362,
+      "step": 2477
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.7513723015144888e-05,
+      "loss": 0.0169,
+      "step": 2478
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.747526184991207e-05,
+      "loss": 0.0131,
+      "step": 2479
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.743681739896965e-05,
+      "loss": 0.0363,
+      "step": 2480
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.739838969084499e-05,
+      "loss": 0.0597,
+      "step": 2481
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.7359978754053035e-05,
+      "loss": 0.0137,
+      "step": 2482
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.73215846170963e-05,
+      "loss": 0.0297,
+      "step": 2483
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.728320730846483e-05,
+      "loss": 0.018,
+      "step": 2484
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7244846856636176e-05,
+      "loss": 0.0258,
+      "step": 2485
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7206503290075387e-05,
+      "loss": 0.0096,
+      "step": 2486
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7168176637234954e-05,
+      "loss": 0.0633,
+      "step": 2487
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7129866926554914e-05,
+      "loss": 0.0511,
+      "step": 2488
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.70915741864626e-05,
+      "loss": 0.0198,
+      "step": 2489
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.705329844537281e-05,
+      "loss": 0.0554,
+      "step": 2490
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7015039731687773e-05,
+      "loss": 0.0232,
+      "step": 2491
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.6976798073797038e-05,
+      "loss": 0.0321,
+      "step": 2492
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.6938573500077457e-05,
+      "loss": 0.024,
+      "step": 2493
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.6900366038893294e-05,
+      "loss": 0.0238,
+      "step": 2494
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.686217571859606e-05,
+      "loss": 0.0271,
+      "step": 2495
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.6824002567524548e-05,
+      "loss": 0.01,
+      "step": 2496
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.6785846614004828e-05,
+      "loss": 0.0443,
+      "step": 2497
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.674770788635018e-05,
+      "loss": 0.0322,
+      "step": 2498
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.6709586412861183e-05,
+      "loss": 0.0626,
+      "step": 2499
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.6671482221825482e-05,
+      "loss": 0.0176,
+      "step": 2500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.6633395341517987e-05,
+      "loss": 0.0315,
+      "step": 2501
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.659532580020077e-05,
+      "loss": 0.0265,
+      "step": 2502
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.6557273626123014e-05,
+      "loss": 0.0189,
+      "step": 2503
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.6519238847520946e-05,
+      "loss": 0.0102,
+      "step": 2504
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.6481221492618015e-05,
+      "loss": 0.015,
+      "step": 2505
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.6443221589624654e-05,
+      "loss": 0.074,
+      "step": 2506
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.6405239166738377e-05,
+      "loss": 0.0269,
+      "step": 2507
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.6367274252143704e-05,
+      "loss": 0.0114,
+      "step": 2508
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.6329326874012183e-05,
+      "loss": 0.0212,
+      "step": 2509
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.6291397060502343e-05,
+      "loss": 0.0395,
+      "step": 2510
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.6253484839759684e-05,
+      "loss": 0.0149,
+      "step": 2511
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.621559023991662e-05,
+      "loss": 0.0225,
+      "step": 2512
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.6177713289092576e-05,
+      "loss": 0.0649,
+      "step": 2513
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.613985401539377e-05,
+      "loss": 0.0463,
+      "step": 2514
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.6102012446913348e-05,
+      "loss": 0.0221,
+      "step": 2515
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.606418861173136e-05,
+      "loss": 0.073,
+      "step": 2516
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.6026382537914666e-05,
+      "loss": 0.0446,
+      "step": 2517
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.5988594253516884e-05,
+      "loss": 0.0262,
+      "step": 2518
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.5950823786578548e-05,
+      "loss": 0.0156,
+      "step": 2519
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.591307116512689e-05,
+      "loss": 0.0231,
+      "step": 2520
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.587533641717592e-05,
+      "loss": 0.0334,
+      "step": 2521
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.583761957072639e-05,
+      "loss": 0.0661,
+      "step": 2522
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.5799920653765743e-05,
+      "loss": 0.0654,
+      "step": 2523
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.57622396942682e-05,
+      "loss": 0.0202,
+      "step": 2524
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.5724576720194536e-05,
+      "loss": 0.014,
+      "step": 2525
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.568693175949224e-05,
+      "loss": 0.0302,
+      "step": 2526
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.564930484009548e-05,
+      "loss": 0.0229,
+      "step": 2527
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.561169598992499e-05,
+      "loss": 0.0112,
+      "step": 2528
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.557410523688803e-05,
+      "loss": 0.0414,
+      "step": 2529
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.553653260887857e-05,
+      "loss": 0.0133,
+      "step": 2530
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.549897813377703e-05,
+      "loss": 0.0156,
+      "step": 2531
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.5461441839450388e-05,
+      "loss": 0.0228,
+      "step": 2532
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.5423923753752145e-05,
+      "loss": 0.0285,
+      "step": 2533
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.538642390452226e-05,
+      "loss": 0.0195,
+      "step": 2534
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.5348942319587187e-05,
+      "loss": 0.0223,
+      "step": 2535
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.531147902675982e-05,
+      "loss": 0.0195,
+      "step": 2536
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.5274034053839456e-05,
+      "loss": 0.0307,
+      "step": 2537
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.523660742861187e-05,
+      "loss": 0.0652,
+      "step": 2538
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.5199199178849114e-05,
+      "loss": 0.0308,
+      "step": 2539
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.5161809332309665e-05,
+      "loss": 0.0632,
+      "step": 2540
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.5124437916738386e-05,
+      "loss": 0.0438,
+      "step": 2541
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.50870849598664e-05,
+      "loss": 0.025,
+      "step": 2542
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.5049750489411116e-05,
+      "loss": 0.013,
+      "step": 2543
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.50124345330763e-05,
+      "loss": 0.0243,
+      "step": 2544
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.497513711855193e-05,
+      "loss": 0.016,
+      "step": 2545
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.4937858273514226e-05,
+      "loss": 0.048,
+      "step": 2546
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.4900598025625648e-05,
+      "loss": 0.0385,
+      "step": 2547
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.4863356402534827e-05,
+      "loss": 0.0225,
+      "step": 2548
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.482613343187664e-05,
+      "loss": 0.0245,
+      "step": 2549
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.4788929141272026e-05,
+      "loss": 0.0224,
+      "step": 2550
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.4751743558328116e-05,
+      "loss": 0.0474,
+      "step": 2551
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.4714576710638183e-05,
+      "loss": 0.0208,
+      "step": 2552
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.467742862578158e-05,
+      "loss": 0.0098,
+      "step": 2553
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.4640299331323662e-05,
+      "loss": 0.0385,
+      "step": 2554
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.4603188854815973e-05,
+      "loss": 0.0413,
+      "step": 2555
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.4566097223795998e-05,
+      "loss": 0.0546,
+      "step": 2556
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.452902446578727e-05,
+      "loss": 0.0219,
+      "step": 2557
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.449197060829932e-05,
+      "loss": 0.0246,
+      "step": 2558
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.4454935678827635e-05,
+      "loss": 0.0683,
+      "step": 2559
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.441791970485368e-05,
+      "loss": 0.0516,
+      "step": 2560
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.438092271384484e-05,
+      "loss": 0.0124,
+      "step": 2561
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.43439447332544e-05,
+      "loss": 0.0666,
+      "step": 2562
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.4306985790521608e-05,
+      "loss": 0.0171,
+      "step": 2563
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.427004591307148e-05,
+      "loss": 0.0626,
+      "step": 2564
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.4233125128314943e-05,
+      "loss": 0.0254,
+      "step": 2565
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.4196223463648776e-05,
+      "loss": 0.0262,
+      "step": 2566
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.415934094645556e-05,
+      "loss": 0.0506,
+      "step": 2567
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.4122477604103588e-05,
+      "loss": 0.0696,
+      "step": 2568
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.408563346394705e-05,
+      "loss": 0.0116,
+      "step": 2569
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.4048808553325802e-05,
+      "loss": 0.0265,
+      "step": 2570
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.401200289956546e-05,
+      "loss": 0.051,
+      "step": 2571
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.3975216529977352e-05,
+      "loss": 0.0378,
+      "step": 2572
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.393844947185846e-05,
+      "loss": 0.0204,
+      "step": 2573
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.3901701752491536e-05,
+      "loss": 0.0138,
+      "step": 2574
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.386497339914484e-05,
+      "loss": 0.0132,
+      "step": 2575
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.382826443907235e-05,
+      "loss": 0.0431,
+      "step": 2576
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.379157489951367e-05,
+      "loss": 0.032,
+      "step": 2577
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.3754904807693962e-05,
+      "loss": 0.0577,
+      "step": 2578
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.3718254190823892e-05,
+      "loss": 0.0329,
+      "step": 2579
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.36816230760998e-05,
+      "loss": 0.0196,
+      "step": 2580
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.3645011490703478e-05,
+      "loss": 0.0185,
+      "step": 2581
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.3608419461802235e-05,
+      "loss": 0.0124,
+      "step": 2582
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.357184701654888e-05,
+      "loss": 0.0294,
+      "step": 2583
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.3535294182081663e-05,
+      "loss": 0.0364,
+      "step": 2584
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.349876098552436e-05,
+      "loss": 0.0615,
+      "step": 2585
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.346224745398607e-05,
+      "loss": 0.01,
+      "step": 2586
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.3425753614561336e-05,
+      "loss": 0.0069,
+      "step": 2587
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.338927949433017e-05,
+      "loss": 0.0173,
+      "step": 2588
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.3352825120357815e-05,
+      "loss": 0.0271,
+      "step": 2589
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.3316390519694932e-05,
+      "loss": 0.0149,
+      "step": 2590
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.327997571937755e-05,
+      "loss": 0.0396,
+      "step": 2591
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.3243580746426962e-05,
+      "loss": 0.0304,
+      "step": 2592
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.320720562784968e-05,
+      "loss": 0.028,
+      "step": 2593
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.3170850390637628e-05,
+      "loss": 0.0149,
+      "step": 2594
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.313451506176786e-05,
+      "loss": 0.0589,
+      "step": 2595
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.3098199668202712e-05,
+      "loss": 0.0173,
+      "step": 2596
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.306190423688971e-05,
+      "loss": 0.036,
+      "step": 2597
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.3025628794761543e-05,
+      "loss": 0.027,
+      "step": 2598
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.2989373368736163e-05,
+      "loss": 0.0082,
+      "step": 2599
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.2953137985716537e-05,
+      "loss": 0.0185,
+      "step": 2600
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.2916922672590822e-05,
+      "loss": 0.0617,
+      "step": 2601
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.2880727456232325e-05,
+      "loss": 0.0208,
+      "step": 2602
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.28445523634994e-05,
+      "loss": 0.0402,
+      "step": 2603
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.2808397421235406e-05,
+      "loss": 0.0129,
+      "step": 2604
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.2772262656268867e-05,
+      "loss": 0.0325,
+      "step": 2605
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.2736148095413256e-05,
+      "loss": 0.032,
+      "step": 2606
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.2700053765467087e-05,
+      "loss": 0.0316,
+      "step": 2607
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.266397969321384e-05,
+      "loss": 0.0121,
+      "step": 2608
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.262792590542196e-05,
+      "loss": 0.0177,
+      "step": 2609
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.2591892428844908e-05,
+      "loss": 0.0316,
+      "step": 2610
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.255587929022096e-05,
+      "loss": 0.0347,
+      "step": 2611
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.2519886516273364e-05,
+      "loss": 0.0172,
+      "step": 2612
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.2483914133710305e-05,
+      "loss": 0.0631,
+      "step": 2613
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.244796216922473e-05,
+      "loss": 0.0193,
+      "step": 2614
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.2412030649494486e-05,
+      "loss": 0.073,
+      "step": 2615
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.2376119601182288e-05,
+      "loss": 0.0348,
+      "step": 2616
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.234022905093563e-05,
+      "loss": 0.0297,
+      "step": 2617
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.230435902538673e-05,
+      "loss": 0.03,
+      "step": 2618
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.2268509551152684e-05,
+      "loss": 0.0152,
+      "step": 2619
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.223268065483528e-05,
+      "loss": 0.0649,
+      "step": 2620
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.219687236302105e-05,
+      "loss": 0.0198,
+      "step": 2621
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.2161084702281217e-05,
+      "loss": 0.0077,
+      "step": 2622
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.2125317699171704e-05,
+      "loss": 0.0049,
+      "step": 2623
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.2089571380233165e-05,
+      "loss": 0.0222,
+      "step": 2624
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.2053845771990794e-05,
+      "loss": 0.0228,
+      "step": 2625
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.2018140900954476e-05,
+      "loss": 0.0836,
+      "step": 2626
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.1982456793618743e-05,
+      "loss": 0.0326,
+      "step": 2627
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.1946793476462686e-05,
+      "loss": 0.0444,
+      "step": 2628
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.1911150975949906e-05,
+      "loss": 0.0447,
+      "step": 2629
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.1875529318528675e-05,
+      "loss": 0.0621,
+      "step": 2630
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.1839928530631725e-05,
+      "loss": 0.0335,
+      "step": 2631
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.1804348638676315e-05,
+      "loss": 0.0174,
+      "step": 2632
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.1768789669064204e-05,
+      "loss": 0.0139,
+      "step": 2633
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.1733251648181602e-05,
+      "loss": 0.0463,
+      "step": 2634
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.169773460239925e-05,
+      "loss": 0.0148,
+      "step": 2635
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.166223855807222e-05,
+      "loss": 0.0206,
+      "step": 2636
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1626763541540042e-05,
+      "loss": 0.0302,
+      "step": 2637
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.159130957912669e-05,
+      "loss": 0.0079,
+      "step": 2638
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.1555876697140475e-05,
+      "loss": 0.0227,
+      "step": 2639
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 2.152046492187401e-05,
+      "loss": 0.017,
+      "step": 2640
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.148507427960436e-05,
+      "loss": 0.0255,
+      "step": 2641
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.144970479659285e-05,
+      "loss": 0.0234,
+      "step": 2642
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.141435649908505e-05,
+      "loss": 0.0141,
+      "step": 2643
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.1379029413310908e-05,
+      "loss": 0.0247,
+      "step": 2644
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.1343723565484586e-05,
+      "loss": 0.0351,
+      "step": 2645
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.1308438981804473e-05,
+      "loss": 0.0341,
+      "step": 2646
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.12731756884532e-05,
+      "loss": 0.0066,
+      "step": 2647
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.123793371159758e-05,
+      "loss": 0.0459,
+      "step": 2648
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.1202713077388676e-05,
+      "loss": 0.0215,
+      "step": 2649
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.1167513811961598e-05,
+      "loss": 0.0603,
+      "step": 2650
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 2.1132335941435667e-05,
+      "loss": 0.0268,
+      "step": 2651
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.1097179491914355e-05,
+      "loss": 0.0458,
+      "step": 2652
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.1062044489485212e-05,
+      "loss": 0.0107,
+      "step": 2653
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.102693096021981e-05,
+      "loss": 0.0047,
+      "step": 2654
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.09918389301739e-05,
+      "loss": 0.0223,
+      "step": 2655
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.095676842538721e-05,
+      "loss": 0.0191,
+      "step": 2656
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 2.09217194718835e-05,
+      "loss": 0.0072,
+      "step": 2657
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.0886692095670558e-05,
+      "loss": 0.0026,
+      "step": 2658
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.0851686322740123e-05,
+      "loss": 0.0517,
+      "step": 2659
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.0816702179068e-05,
+      "loss": 0.0201,
+      "step": 2660
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 2.07817396906138e-05,
+      "loss": 0.0224,
+      "step": 2661
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.074679888332116e-05,
+      "loss": 0.0233,
+      "step": 2662
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.0711879783117637e-05,
+      "loss": 0.0186,
+      "step": 2663
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.0676982415914647e-05,
+      "loss": 0.0561,
+      "step": 2664
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.0642106807607443e-05,
+      "loss": 0.0281,
+      "step": 2665
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 2.0607252984075215e-05,
+      "loss": 0.0437,
+      "step": 2666
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.0572420971180955e-05,
+      "loss": 0.0191,
+      "step": 2667
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.0537610794771406e-05,
+      "loss": 0.0159,
+      "step": 2668
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.0502822480677215e-05,
+      "loss": 0.0019,
+      "step": 2669
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.0468056054712732e-05,
+      "loss": 0.015,
+      "step": 2670
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.04333115426761e-05,
+      "loss": 0.0049,
+      "step": 2671
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 2.0398588970349176e-05,
+      "loss": 0.0571,
+      "step": 2672
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.0363888363497542e-05,
+      "loss": 0.0257,
+      "step": 2673
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.0329209747870536e-05,
+      "loss": 0.0098,
+      "step": 2674
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.0294553149201078e-05,
+      "loss": 0.0217,
+      "step": 2675
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.02599185932058e-05,
+      "loss": 0.0113,
+      "step": 2676
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.0225306105585023e-05,
+      "loss": 0.0374,
+      "step": 2677
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 2.019071571202264e-05,
+      "loss": 0.0597,
+      "step": 2678
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.015614743818611e-05,
+      "loss": 0.0273,
+      "step": 2679
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.0121601309726572e-05,
+      "loss": 0.0206,
+      "step": 2680
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 2.0087077352278682e-05,
+      "loss": 0.0135,
+      "step": 2681
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.0052575591460637e-05,
+      "loss": 0.0166,
+      "step": 2682
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.0018096052874185e-05,
+      "loss": 0.0078,
+      "step": 2683
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.9983638762104552e-05,
+      "loss": 0.0532,
+      "step": 2684
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.994920374472053e-05,
+      "loss": 0.0129,
+      "step": 2685
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.9914791026274277e-05,
+      "loss": 0.0436,
+      "step": 2686
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.988040063230146e-05,
+      "loss": 0.0146,
+      "step": 2687
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.9846032588321218e-05,
+      "loss": 0.0381,
+      "step": 2688
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.9811686919836053e-05,
+      "loss": 0.0252,
+      "step": 2689
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.9777363652331837e-05,
+      "loss": 0.0073,
+      "step": 2690
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.9743062811277908e-05,
+      "loss": 0.0317,
+      "step": 2691
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.9708784422126892e-05,
+      "loss": 0.0083,
+      "step": 2692
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.967452851031477e-05,
+      "loss": 0.0419,
+      "step": 2693
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.964029510126087e-05,
+      "loss": 0.0427,
+      "step": 2694
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.9606084220367783e-05,
+      "loss": 0.0265,
+      "step": 2695
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.957189589302141e-05,
+      "loss": 0.0048,
+      "step": 2696
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.953773014459091e-05,
+      "loss": 0.0166,
+      "step": 2697
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.9503587000428665e-05,
+      "loss": 0.0592,
+      "step": 2698
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.9469466485870358e-05,
+      "loss": 0.0141,
+      "step": 2699
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.9435368626234774e-05,
+      "loss": 0.0412,
+      "step": 2700
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.9401293446823938e-05,
+      "loss": 0.0278,
+      "step": 2701
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.936724097292309e-05,
+      "loss": 0.0235,
+      "step": 2702
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.9333211229800567e-05,
+      "loss": 0.0389,
+      "step": 2703
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.9299204242707812e-05,
+      "loss": 0.0294,
+      "step": 2704
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.926522003687946e-05,
+      "loss": 0.0109,
+      "step": 2705
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.923125863753319e-05,
+      "loss": 0.0498,
+      "step": 2706
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.9197320069869774e-05,
+      "loss": 0.0291,
+      "step": 2707
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.916340435907303e-05,
+      "loss": 0.0187,
+      "step": 2708
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.9129511530309802e-05,
+      "loss": 0.0326,
+      "step": 2709
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.9095641608730042e-05,
+      "loss": 0.0099,
+      "step": 2710
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.9061794619466573e-05,
+      "loss": 0.0129,
+      "step": 2711
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.902797058763527e-05,
+      "loss": 0.0069,
+      "step": 2712
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.8994169538335e-05,
+      "loss": 0.0158,
+      "step": 2713
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.8960391496647544e-05,
+      "loss": 0.0205,
+      "step": 2714
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.8926636487637567e-05,
+      "loss": 0.0329,
+      "step": 2715
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.889290453635273e-05,
+      "loss": 0.0157,
+      "step": 2716
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.885919566782352e-05,
+      "loss": 0.0531,
+      "step": 2717
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.8825509907063327e-05,
+      "loss": 0.0216,
+      "step": 2718
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.8791847279068375e-05,
+      "loss": 0.0121,
+      "step": 2719
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.875820780881774e-05,
+      "loss": 0.0215,
+      "step": 2720
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.8724591521273294e-05,
+      "loss": 0.0186,
+      "step": 2721
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.869099844137972e-05,
+      "loss": 0.0095,
+      "step": 2722
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.8657428594064462e-05,
+      "loss": 0.0326,
+      "step": 2723
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.862388200423779e-05,
+      "loss": 0.057,
+      "step": 2724
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.859035869679262e-05,
+      "loss": 0.0274,
+      "step": 2725
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.855685869660463e-05,
+      "loss": 0.0049,
+      "step": 2726
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.8523382028532248e-05,
+      "loss": 0.0162,
+      "step": 2727
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.8489928717416548e-05,
+      "loss": 0.0514,
+      "step": 2728
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.845649878808123e-05,
+      "loss": 0.0165,
+      "step": 2729
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.842309226533273e-05,
+      "loss": 0.013,
+      "step": 2730
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.838970917396006e-05,
+      "loss": 0.0163,
+      "step": 2731
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.835634953873487e-05,
+      "loss": 0.0545,
+      "step": 2732
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.832301338441138e-05,
+      "loss": 0.0272,
+      "step": 2733
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.8289700735726383e-05,
+      "loss": 0.0233,
+      "step": 2734
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.825641161739931e-05,
+      "loss": 0.0101,
+      "step": 2735
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.8223146054132006e-05,
+      "loss": 0.0216,
+      "step": 2736
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.8189904070608908e-05,
+      "loss": 0.0207,
+      "step": 2737
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.815668569149697e-05,
+      "loss": 0.0242,
+      "step": 2738
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.812349094144562e-05,
+      "loss": 0.0291,
+      "step": 2739
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.809031984508668e-05,
+      "loss": 0.0231,
+      "step": 2740
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.805717242703453e-05,
+      "loss": 0.0346,
+      "step": 2741
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.802404871188592e-05,
+      "loss": 0.0363,
+      "step": 2742
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.799094872422002e-05,
+      "loss": 0.0333,
+      "step": 2743
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.795787248859839e-05,
+      "loss": 0.0471,
+      "step": 2744
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.792482002956495e-05,
+      "loss": 0.0159,
+      "step": 2745
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.7891791371646055e-05,
+      "loss": 0.0335,
+      "step": 2746
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.785878653935028e-05,
+      "loss": 0.0245,
+      "step": 2747
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.782580555716859e-05,
+      "loss": 0.044,
+      "step": 2748
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.779284844957429e-05,
+      "loss": 0.0153,
+      "step": 2749
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.7759915241022866e-05,
+      "loss": 0.0112,
+      "step": 2750
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.772700595595213e-05,
+      "loss": 0.0472,
+      "step": 2751
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.7694120618782168e-05,
+      "loss": 0.0256,
+      "step": 2752
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.7661259253915267e-05,
+      "loss": 0.0265,
+      "step": 2753
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.7628421885735874e-05,
+      "loss": 0.0072,
+      "step": 2754
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.7595608538610724e-05,
+      "loss": 0.0464,
+      "step": 2755
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.7562819236888667e-05,
+      "loss": 0.0443,
+      "step": 2756
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.7530054004900727e-05,
+      "loss": 0.0329,
+      "step": 2757
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.749731286696007e-05,
+      "loss": 0.049,
+      "step": 2758
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.7464595847361954e-05,
+      "loss": 0.0909,
+      "step": 2759
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.7431902970383818e-05,
+      "loss": 0.0221,
+      "step": 2760
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.7399234260285087e-05,
+      "loss": 0.0333,
+      "step": 2761
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.736658974130729e-05,
+      "loss": 0.0033,
+      "step": 2762
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.733396943767407e-05,
+      "loss": 0.0251,
+      "step": 2763
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.7301373373591016e-05,
+      "loss": 0.0293,
+      "step": 2764
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.726880157324573e-05,
+      "loss": 0.0291,
+      "step": 2765
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.723625406080789e-05,
+      "loss": 0.0335,
+      "step": 2766
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.720373086042907e-05,
+      "loss": 0.0255,
+      "step": 2767
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.7171231996242847e-05,
+      "loss": 0.0509,
+      "step": 2768
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.713875749236472e-05,
+      "loss": 0.0123,
+      "step": 2769
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.7106307372892104e-05,
+      "loss": 0.0322,
+      "step": 2770
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.7073881661904383e-05,
+      "loss": 0.0227,
+      "step": 2771
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.7041480383462737e-05,
+      "loss": 0.0271,
+      "step": 2772
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.700910356161025e-05,
+      "loss": 0.0303,
+      "step": 2773
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.6976751220371924e-05,
+      "loss": 0.0089,
+      "step": 2774
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.6944423383754483e-05,
+      "loss": 0.0291,
+      "step": 2775
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.6912120075746523e-05,
+      "loss": 0.0168,
+      "step": 2776
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.6879841320318478e-05,
+      "loss": 0.0607,
+      "step": 2777
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.684758714142252e-05,
+      "loss": 0.0096,
+      "step": 2778
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.681535756299254e-05,
+      "loss": 0.0245,
+      "step": 2779
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.678315260894427e-05,
+      "loss": 0.0388,
+      "step": 2780
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.675097230317511e-05,
+      "loss": 0.038,
+      "step": 2781
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.671881666956418e-05,
+      "loss": 0.0577,
+      "step": 2782
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.668668573197229e-05,
+      "loss": 0.0149,
+      "step": 2783
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.665457951424192e-05,
+      "loss": 0.0642,
+      "step": 2784
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.662249804019726e-05,
+      "loss": 0.0581,
+      "step": 2785
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.6590441333644047e-05,
+      "loss": 0.0746,
+      "step": 2786
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.6558409418369686e-05,
+      "loss": 0.0077,
+      "step": 2787
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.6526402318143224e-05,
+      "loss": 0.0031,
+      "step": 2788
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.6494420056715253e-05,
+      "loss": 0.0136,
+      "step": 2789
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.6462462657817896e-05,
+      "loss": 0.042,
+      "step": 2790
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.6430530145164913e-05,
+      "loss": 0.0462,
+      "step": 2791
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.6398622542451537e-05,
+      "loss": 0.0897,
+      "step": 2792
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.636673987335453e-05,
+      "loss": 0.0096,
+      "step": 2793
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.633488216153216e-05,
+      "loss": 0.0316,
+      "step": 2794
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.6303049430624158e-05,
+      "loss": 0.018,
+      "step": 2795
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.627124170425177e-05,
+      "loss": 0.0171,
+      "step": 2796
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.6239459006017614e-05,
+      "loss": 0.0142,
+      "step": 2797
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.620770135950576e-05,
+      "loss": 0.0067,
+      "step": 2798
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.617596878828173e-05,
+      "loss": 0.0417,
+      "step": 2799
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.6144261315892424e-05,
+      "loss": 0.0381,
+      "step": 2800
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.6112578965866037e-05,
+      "loss": 0.0096,
+      "step": 2801
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.608092176171224e-05,
+      "loss": 0.0162,
+      "step": 2802
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.6049289726921997e-05,
+      "loss": 0.0171,
+      "step": 2803
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.601768288496754e-05,
+      "loss": 0.0033,
+      "step": 2804
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.5986101259302507e-05,
+      "loss": 0.036,
+      "step": 2805
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.5954544873361755e-05,
+      "loss": 0.044,
+      "step": 2806
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.5923013750561426e-05,
+      "loss": 0.0078,
+      "step": 2807
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.5891507914298935e-05,
+      "loss": 0.0178,
+      "step": 2808
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.5860027387952902e-05,
+      "loss": 0.0009,
+      "step": 2809
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.5828572194883223e-05,
+      "loss": 0.0245,
+      "step": 2810
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.5797142358430912e-05,
+      "loss": 0.0075,
+      "step": 2811
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.5765737901918214e-05,
+      "loss": 0.0296,
+      "step": 2812
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.5734358848648573e-05,
+      "loss": 0.0058,
+      "step": 2813
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.570300522190654e-05,
+      "loss": 0.0073,
+      "step": 2814
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.5671677044957773e-05,
+      "loss": 0.0263,
+      "step": 2815
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.5640374341049108e-05,
+      "loss": 0.0244,
+      "step": 2816
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.5609097133408453e-05,
+      "loss": 0.0299,
+      "step": 2817
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.557784544524478e-05,
+      "loss": 0.0087,
+      "step": 2818
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.554661929974815e-05,
+      "loss": 0.0207,
+      "step": 2819
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.5515418720089637e-05,
+      "loss": 0.0263,
+      "step": 2820
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.548424372942141e-05,
+      "loss": 0.006,
+      "step": 2821
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.545309435087656e-05,
+      "loss": 0.0053,
+      "step": 2822
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.5421970607569225e-05,
+      "loss": 0.0136,
+      "step": 2823
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.5390872522594536e-05,
+      "loss": 0.0288,
+      "step": 2824
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.535980011902857e-05,
+      "loss": 0.0035,
+      "step": 2825
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.532875341992829e-05,
+      "loss": 0.0503,
+      "step": 2826
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.529773244833168e-05,
+      "loss": 0.0409,
+      "step": 2827
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.5266737227257576e-05,
+      "loss": 0.0285,
+      "step": 2828
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.5235767779705717e-05,
+      "loss": 0.0277,
+      "step": 2829
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.520482412865672e-05,
+      "loss": 0.1033,
+      "step": 2830
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.517390629707205e-05,
+      "loss": 0.024,
+      "step": 2831
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.5143014307894032e-05,
+      "loss": 0.0508,
+      "step": 2832
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.5112148184045794e-05,
+      "loss": 0.0142,
+      "step": 2833
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.5081307948431273e-05,
+      "loss": 0.0394,
+      "step": 2834
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.5050493623935247e-05,
+      "loss": 0.009,
+      "step": 2835
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.5019705233423176e-05,
+      "loss": 0.0099,
+      "step": 2836
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.4988942799741324e-05,
+      "loss": 0.0262,
+      "step": 2837
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.4958206345716724e-05,
+      "loss": 0.0321,
+      "step": 2838
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.49274958941571e-05,
+      "loss": 0.0238,
+      "step": 2839
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.489681146785084e-05,
+      "loss": 0.0327,
+      "step": 2840
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.4866153089567103e-05,
+      "loss": 0.0057,
+      "step": 2841
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.4835520782055663e-05,
+      "loss": 0.0095,
+      "step": 2842
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.4804914568046957e-05,
+      "loss": 0.0073,
+      "step": 2843
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.4774334470252077e-05,
+      "loss": 0.0218,
+      "step": 2844
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.4743780511362704e-05,
+      "loss": 0.0135,
+      "step": 2845
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.471325271405119e-05,
+      "loss": 0.031,
+      "step": 2846
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.4682751100970382e-05,
+      "loss": 0.0094,
+      "step": 2847
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.4652275694753742e-05,
+      "loss": 0.0054,
+      "step": 2848
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.4621826518015314e-05,
+      "loss": 0.0411,
+      "step": 2849
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.4591403593349651e-05,
+      "loss": 0.024,
+      "step": 2850
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.4561006943331784e-05,
+      "loss": 0.0163,
+      "step": 2851
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.4530636590517333e-05,
+      "loss": 0.0204,
+      "step": 2852
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.4500292557442346e-05,
+      "loss": 0.0336,
+      "step": 2853
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.4469974866623348e-05,
+      "loss": 0.0365,
+      "step": 2854
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.443968354055733e-05,
+      "loss": 0.0168,
+      "step": 2855
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.4409418601721713e-05,
+      "loss": 0.0176,
+      "step": 2856
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.4379180072574333e-05,
+      "loss": 0.029,
+      "step": 2857
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.4348967975553435e-05,
+      "loss": 0.0068,
+      "step": 2858
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.4318782333077634e-05,
+      "loss": 0.043,
+      "step": 2859
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.4288623167545979e-05,
+      "loss": 0.0232,
+      "step": 2860
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.4258490501337768e-05,
+      "loss": 0.0066,
+      "step": 2861
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.4228384356812696e-05,
+      "loss": 0.0305,
+      "step": 2862
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.4198304756310804e-05,
+      "loss": 0.0169,
+      "step": 2863
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.4168251722152399e-05,
+      "loss": 0.01,
+      "step": 2864
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.4138225276638028e-05,
+      "loss": 0.0185,
+      "step": 2865
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.4108225442048618e-05,
+      "loss": 0.0211,
+      "step": 2866
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.4078252240645268e-05,
+      "loss": 0.0355,
+      "step": 2867
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.4048305694669344e-05,
+      "loss": 0.0215,
+      "step": 2868
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.4018385826342418e-05,
+      "loss": 0.0299,
+      "step": 2869
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.398849265786626e-05,
+      "loss": 0.0129,
+      "step": 2870
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.3958626211422888e-05,
+      "loss": 0.0091,
+      "step": 2871
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.3928786509174401e-05,
+      "loss": 0.029,
+      "step": 2872
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.3898973573263096e-05,
+      "loss": 0.0208,
+      "step": 2873
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.3869187425811437e-05,
+      "loss": 0.0223,
+      "step": 2874
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.3839428088921985e-05,
+      "loss": 0.0193,
+      "step": 2875
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.380969558467735e-05,
+      "loss": 0.0213,
+      "step": 2876
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.3779989935140335e-05,
+      "loss": 0.0188,
+      "step": 2877
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.3750311162353752e-05,
+      "loss": 0.0088,
+      "step": 2878
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.3720659288340476e-05,
+      "loss": 0.0249,
+      "step": 2879
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.3691034335103425e-05,
+      "loss": 0.0056,
+      "step": 2880
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.3661436324625543e-05,
+      "loss": 0.0111,
+      "step": 2881
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 1.363186527886981e-05,
+      "loss": 0.0151,
+      "step": 2882
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.3602321219779135e-05,
+      "loss": 0.0375,
+      "step": 2883
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.3572804169276432e-05,
+      "loss": 0.0136,
+      "step": 2884
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.3543314149264625e-05,
+      "loss": 0.0106,
+      "step": 2885
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.3513851181626485e-05,
+      "loss": 0.0083,
+      "step": 2886
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.3484415288224755e-05,
+      "loss": 0.0096,
+      "step": 2887
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.3455006490902123e-05,
+      "loss": 0.0115,
+      "step": 2888
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.3425624811481136e-05,
+      "loss": 0.0091,
+      "step": 2889
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.3396270271764172e-05,
+      "loss": 0.0144,
+      "step": 2890
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.3366942893533562e-05,
+      "loss": 0.0086,
+      "step": 2891
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.3337642698551428e-05,
+      "loss": 0.0363,
+      "step": 2892
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.3308369708559726e-05,
+      "loss": 0.0216,
+      "step": 2893
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.327912394528023e-05,
+      "loss": 0.0051,
+      "step": 2894
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.32499054304145e-05,
+      "loss": 0.0164,
+      "step": 2895
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.3220714185643923e-05,
+      "loss": 0.0283,
+      "step": 2896
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.3191550232629573e-05,
+      "loss": 0.0177,
+      "step": 2897
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.3162413593012313e-05,
+      "loss": 0.0112,
+      "step": 2898
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 1.3133304288412768e-05,
+      "loss": 0.0142,
+      "step": 2899
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.310422234043125e-05,
+      "loss": 0.0231,
+      "step": 2900
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.3075167770647729e-05,
+      "loss": 0.0024,
+      "step": 2901
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.3046140600621949e-05,
+      "loss": 0.0131,
+      "step": 2902
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.3017140851893245e-05,
+      "loss": 0.0206,
+      "step": 2903
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.2988168545980651e-05,
+      "loss": 0.0174,
+      "step": 2904
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.2959223704382811e-05,
+      "loss": 0.0141,
+      "step": 2905
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.293030634857798e-05,
+      "loss": 0.0195,
+      "step": 2906
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.2901416500024095e-05,
+      "loss": 0.0192,
+      "step": 2907
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.2872554180158563e-05,
+      "loss": 0.0161,
+      "step": 2908
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.2843719410398431e-05,
+      "loss": 0.0041,
+      "step": 2909
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.281491221214034e-05,
+      "loss": 0.0082,
+      "step": 2910
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.2786132606760377e-05,
+      "loss": 0.0159,
+      "step": 2911
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.2757380615614206e-05,
+      "loss": 0.0456,
+      "step": 2912
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.2728656260037025e-05,
+      "loss": 0.0608,
+      "step": 2913
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.26999595613435e-05,
+      "loss": 0.032,
+      "step": 2914
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.267129054082773e-05,
+      "loss": 0.0558,
+      "step": 2915
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 1.2642649219763358e-05,
+      "loss": 0.0448,
+      "step": 2916
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.261403561940342e-05,
+      "loss": 0.0111,
+      "step": 2917
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.2585449760980389e-05,
+      "loss": 0.0216,
+      "step": 2918
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.2556891665706167e-05,
+      "loss": 0.0153,
+      "step": 2919
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.2528361354772029e-05,
+      "loss": 0.0037,
+      "step": 2920
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.2499858849348695e-05,
+      "loss": 0.0223,
+      "step": 2921
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.2471384170586154e-05,
+      "loss": 0.0299,
+      "step": 2922
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.2442937339613808e-05,
+      "loss": 0.0891,
+      "step": 2923
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.2414518377540407e-05,
+      "loss": 0.028,
+      "step": 2924
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.2386127305454015e-05,
+      "loss": 0.0159,
+      "step": 2925
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.2357764144421923e-05,
+      "loss": 0.0197,
+      "step": 2926
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.2329428915490832e-05,
+      "loss": 0.0041,
+      "step": 2927
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.2301121639686636e-05,
+      "loss": 0.0289,
+      "step": 2928
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.2272842338014507e-05,
+      "loss": 0.0026,
+      "step": 2929
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.2244591031458868e-05,
+      "loss": 0.0176,
+      "step": 2930
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.2216367740983337e-05,
+      "loss": 0.0066,
+      "step": 2931
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.2188172487530819e-05,
+      "loss": 0.052,
+      "step": 2932
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 1.2160005292023308e-05,
+      "loss": 0.0213,
+      "step": 2933
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.213186617536204e-05,
+      "loss": 0.0138,
+      "step": 2934
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.2103755158427438e-05,
+      "loss": 0.0231,
+      "step": 2935
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.2075672262079035e-05,
+      "loss": 0.0525,
+      "step": 2936
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 1.2047617507155468e-05,
+      "loss": 0.0153,
+      "step": 2937
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.2019590914474566e-05,
+      "loss": 0.0098,
+      "step": 2938
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.1991592504833233e-05,
+      "loss": 0.0093,
+      "step": 2939
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.1963622299007405e-05,
+      "loss": 0.036,
+      "step": 2940
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.1935680317752174e-05,
+      "loss": 0.0107,
+      "step": 2941
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.1907766581801632e-05,
+      "loss": 0.0123,
+      "step": 2942
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.187988111186894e-05,
+      "loss": 0.0383,
+      "step": 2943
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.1852023928646271e-05,
+      "loss": 0.0098,
+      "step": 2944
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.1824195052804794e-05,
+      "loss": 0.0048,
+      "step": 2945
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.1796394504994729e-05,
+      "loss": 0.031,
+      "step": 2946
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.1768622305845201e-05,
+      "loss": 0.0026,
+      "step": 2947
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.1740878475964328e-05,
+      "loss": 0.0067,
+      "step": 2948
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.1713163035939217e-05,
+      "loss": 0.0156,
+      "step": 2949
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 1.1685476006335865e-05,
+      "loss": 0.0123,
+      "step": 2950
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.1657817407699167e-05,
+      "loss": 0.0456,
+      "step": 2951
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.1630187260552983e-05,
+      "loss": 0.0367,
+      "step": 2952
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.1602585585400022e-05,
+      "loss": 0.0031,
+      "step": 2953
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 1.1575012402721869e-05,
+      "loss": 0.0113,
+      "step": 2954
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.1547467732978967e-05,
+      "loss": 0.0052,
+      "step": 2955
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.1519951596610595e-05,
+      "loss": 0.0113,
+      "step": 2956
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.1492464014034904e-05,
+      "loss": 0.0114,
+      "step": 2957
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.1465005005648783e-05,
+      "loss": 0.0337,
+      "step": 2958
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.143757459182796e-05,
+      "loss": 0.0622,
+      "step": 2959
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.1410172792926959e-05,
+      "loss": 0.0197,
+      "step": 2960
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.1382799629279057e-05,
+      "loss": 0.0289,
+      "step": 2961
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.1355455121196234e-05,
+      "loss": 0.0187,
+      "step": 2962
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.1328139288969292e-05,
+      "loss": 0.0303,
+      "step": 2963
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.1300852152867713e-05,
+      "loss": 0.0133,
+      "step": 2964
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.1273593733139643e-05,
+      "loss": 0.0085,
+      "step": 2965
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.1246364050011998e-05,
+      "loss": 0.008,
+      "step": 2966
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 1.1219163123690324e-05,
+      "loss": 0.0177,
+      "step": 2967
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.1191990974358835e-05,
+      "loss": 0.0331,
+      "step": 2968
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.1164847622180391e-05,
+      "loss": 0.0374,
+      "step": 2969
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.113773308729647e-05,
+      "loss": 0.0185,
+      "step": 2970
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 1.1110647389827234e-05,
+      "loss": 0.0194,
+      "step": 2971
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.1083590549871342e-05,
+      "loss": 0.0328,
+      "step": 2972
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.1056562587506102e-05,
+      "loss": 0.0168,
+      "step": 2973
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.10295635227874e-05,
+      "loss": 0.0188,
+      "step": 2974
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 1.100259337574967e-05,
+      "loss": 0.0277,
+      "step": 2975
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.0975652166405837e-05,
+      "loss": 0.0102,
+      "step": 2976
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0948739914747442e-05,
+      "loss": 0.0256,
+      "step": 2977
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.0921856640744466e-05,
+      "loss": 0.0168,
+      "step": 2978
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0895002364345424e-05,
+      "loss": 0.01,
+      "step": 2979
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.0868177105477305e-05,
+      "loss": 0.0158,
+      "step": 2980
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0841380884045544e-05,
+      "loss": 0.006,
+      "step": 2981
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.0814613719934092e-05,
+      "loss": 0.0275,
+      "step": 2982
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0787875633005257e-05,
+      "loss": 0.0206,
+      "step": 2983
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 1.0761166643099807e-05,
+      "loss": 0.0255,
+      "step": 2984
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.073448677003695e-05,
+      "loss": 0.0026,
+      "step": 2985
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.070783603361426e-05,
+      "loss": 0.0037,
+      "step": 2986
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0681214453607641e-05,
+      "loss": 0.0167,
+      "step": 2987
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 1.0654622049771456e-05,
+      "loss": 0.017,
+      "step": 2988
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.0628058841838363e-05,
+      "loss": 0.0098,
+      "step": 2989
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.0601524849519356e-05,
+      "loss": 0.0075,
+      "step": 2990
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0575020092503763e-05,
+      "loss": 0.0331,
+      "step": 2991
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 1.0548544590459215e-05,
+      "loss": 0.0083,
+      "step": 2992
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.052209836303163e-05,
+      "loss": 0.0126,
+      "step": 2993
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0495681429845205e-05,
+      "loss": 0.0196,
+      "step": 2994
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.0469293810502389e-05,
+      "loss": 0.0154,
+      "step": 2995
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0442935524583935e-05,
+      "loss": 0.0057,
+      "step": 2996
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.0416606591648737e-05,
+      "loss": 0.0265,
+      "step": 2997
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0390307031233959e-05,
+      "loss": 0.0101,
+      "step": 2998
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.0364036862854997e-05,
+      "loss": 0.0069,
+      "step": 2999
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0337796106005398e-05,
+      "loss": 0.0022,
+      "step": 3000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 1.0311584780156863e-05,
+      "loss": 0.0034,
+      "step": 3001
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0285402904759312e-05,
+      "loss": 0.0117,
+      "step": 3002
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.0259250499240775e-05,
+      "loss": 0.0183,
+      "step": 3003
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0233127583007424e-05,
+      "loss": 0.0411,
+      "step": 3004
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 1.0207034175443548e-05,
+      "loss": 0.0847,
+      "step": 3005
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0180970295911513e-05,
+      "loss": 0.0365,
+      "step": 3006
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.0154935963751855e-05,
+      "loss": 0.0706,
+      "step": 3007
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 1.0128931198283081e-05,
+      "loss": 0.0242,
+      "step": 3008
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.0102956018801801e-05,
+      "loss": 0.0203,
+      "step": 3009
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.0077010444582713e-05,
+      "loss": 0.0236,
+      "step": 3010
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 1.0051094494878504e-05,
+      "loss": 0.0018,
+      "step": 3011
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 1.0025208188919844e-05,
+      "loss": 0.0197,
+      "step": 3012
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.999351545915486e-06,
+      "loss": 0.0528,
+      "step": 3013
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 9.973524585052124e-06,
+      "loss": 0.0104,
+      "step": 3014
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.947727325494426e-06,
+      "loss": 0.0037,
+      "step": 3015
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 9.921959786385038e-06,
+      "loss": 0.0085,
+      "step": 3016
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.896221986844539e-06,
+      "loss": 0.0483,
+      "step": 3017
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 9.870513945971444e-06,
+      "loss": 0.0132,
+      "step": 3018
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.84483568284218e-06,
+      "loss": 0.0179,
+      "step": 3019
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 9.819187216511083e-06,
+      "loss": 0.0229,
+      "step": 3020
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.793568566010409e-06,
+      "loss": 0.0218,
+      "step": 3021
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 9.767979750350226e-06,
+      "loss": 0.0349,
+      "step": 3022
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 9.742420788518497e-06,
+      "loss": 0.0071,
+      "step": 3023
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.716891699481052e-06,
+      "loss": 0.0097,
+      "step": 3024
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 9.69139250218154e-06,
+      "loss": 0.0309,
+      "step": 3025
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.665923215541384e-06,
+      "loss": 0.0261,
+      "step": 3026
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 9.640483858459891e-06,
+      "loss": 0.0275,
+      "step": 3027
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.615074449814105e-06,
+      "loss": 0.0182,
+      "step": 3028
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 9.58969500845886e-06,
+      "loss": 0.0218,
+      "step": 3029
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.564345553226756e-06,
+      "loss": 0.0053,
+      "step": 3030
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 9.539026102928134e-06,
+      "loss": 0.0238,
+      "step": 3031
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.513736676351104e-06,
+      "loss": 0.0075,
+      "step": 3032
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 9.488477292261438e-06,
+      "loss": 0.0219,
+      "step": 3033
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.463247969402645e-06,
+      "loss": 0.0348,
+      "step": 3034
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 9.438048726495957e-06,
+      "loss": 0.0205,
+      "step": 3035
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.412879582240258e-06,
+      "loss": 0.0506,
+      "step": 3036
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 9.387740555312059e-06,
+      "loss": 0.0107,
+      "step": 3037
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.3626316643656e-06,
+      "loss": 0.0076,
+      "step": 3038
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 9.337552928032706e-06,
+      "loss": 0.0089,
+      "step": 3039
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 9.312504364922836e-06,
+      "loss": 0.0224,
+      "step": 3040
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.287485993623069e-06,
+      "loss": 0.0059,
+      "step": 3041
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 9.262497832698058e-06,
+      "loss": 0.009,
+      "step": 3042
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.2375399006901e-06,
+      "loss": 0.0065,
+      "step": 3043
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 9.21261221611897e-06,
+      "loss": 0.0071,
+      "step": 3044
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.18771479748205e-06,
+      "loss": 0.09,
+      "step": 3045
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 9.162847663254292e-06,
+      "loss": 0.0099,
+      "step": 3046
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.138010831888105e-06,
+      "loss": 0.0379,
+      "step": 3047
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 9.113204321813452e-06,
+      "loss": 0.0042,
+      "step": 3048
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.088428151437811e-06,
+      "loss": 0.0045,
+      "step": 3049
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 9.063682339146134e-06,
+      "loss": 0.0063,
+      "step": 3050
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.0389669033008e-06,
+      "loss": 0.0644,
+      "step": 3051
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 9.014281862241724e-06,
+      "loss": 0.0431,
+      "step": 3052
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 8.989627234286225e-06,
+      "loss": 0.0214,
+      "step": 3053
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 8.965003037729058e-06,
+      "loss": 0.0257,
+      "step": 3054
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 8.940409290842394e-06,
+      "loss": 0.0151,
+      "step": 3055
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 8.9158460118758e-06,
+      "loss": 0.0229,
+      "step": 3056
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 8.891313219056296e-06,
+      "loss": 0.0579,
+      "step": 3057
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 8.866810930588187e-06,
+      "loss": 0.0519,
+      "step": 3058
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 8.842339164653185e-06,
+      "loss": 0.0266,
+      "step": 3059
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 8.817897939410392e-06,
+      "loss": 0.0101,
+      "step": 3060
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 8.793487272996197e-06,
+      "loss": 0.0071,
+      "step": 3061
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 8.7691071835243e-06,
+      "loss": 0.0212,
+      "step": 3062
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 8.744757689085775e-06,
+      "loss": 0.0304,
+      "step": 3063
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 8.720438807748937e-06,
+      "loss": 0.0533,
+      "step": 3064
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 8.696150557559419e-06,
+      "loss": 0.016,
+      "step": 3065
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 8.671892956540101e-06,
+      "loss": 0.0196,
+      "step": 3066
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 8.64766602269112e-06,
+      "loss": 0.0285,
+      "step": 3067
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 8.623469773989901e-06,
+      "loss": 0.0032,
+      "step": 3068
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 8.599304228391025e-06,
+      "loss": 0.0117,
+      "step": 3069
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 8.575169403826338e-06,
+      "loss": 0.014,
+      "step": 3070
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 8.551065318204904e-06,
+      "loss": 0.0247,
+      "step": 3071
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 8.526991989412947e-06,
+      "loss": 0.0272,
+      "step": 3072
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 8.50294943531385e-06,
+      "loss": 0.0408,
+      "step": 3073
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 8.478937673748211e-06,
+      "loss": 0.0279,
+      "step": 3074
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 8.454956722533764e-06,
+      "loss": 0.0429,
+      "step": 3075
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 8.431006599465324e-06,
+      "loss": 0.0084,
+      "step": 3076
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 8.40708732231492e-06,
+      "loss": 0.0186,
+      "step": 3077
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 8.383198908831635e-06,
+      "loss": 0.0399,
+      "step": 3078
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 8.359341376741658e-06,
+      "loss": 0.0105,
+      "step": 3079
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 8.335514743748274e-06,
+      "loss": 0.0263,
+      "step": 3080
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 8.311719027531817e-06,
+      "loss": 0.0118,
+      "step": 3081
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 8.28795424574973e-06,
+      "loss": 0.0235,
+      "step": 3082
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 8.264220416036438e-06,
+      "loss": 0.0065,
+      "step": 3083
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 8.240517556003424e-06,
+      "loss": 0.0971,
+      "step": 3084
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.216845683239222e-06,
+      "loss": 0.0243,
+      "step": 3085
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 8.193204815309335e-06,
+      "loss": 0.0306,
+      "step": 3086
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 8.169594969756233e-06,
+      "loss": 0.0139,
+      "step": 3087
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.146016164099445e-06,
+      "loss": 0.0463,
+      "step": 3088
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 8.122468415835411e-06,
+      "loss": 0.0155,
+      "step": 3089
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.098951742437527e-06,
+      "loss": 0.0032,
+      "step": 3090
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.075466161356149e-06,
+      "loss": 0.0047,
+      "step": 3091
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.05201169001854e-06,
+      "loss": 0.009,
+      "step": 3092
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 8.028588345828924e-06,
+      "loss": 0.0079,
+      "step": 3093
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 8.005196146168359e-06,
+      "loss": 0.0148,
+      "step": 3094
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 7.981835108394824e-06,
+      "loss": 0.0493,
+      "step": 3095
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 7.958505249843217e-06,
+      "loss": 0.0056,
+      "step": 3096
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 7.935206587825244e-06,
+      "loss": 0.0229,
+      "step": 3097
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 7.911939139629454e-06,
+      "loss": 0.0219,
+      "step": 3098
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 7.888702922521291e-06,
+      "loss": 0.0192,
+      "step": 3099
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 7.86549795374299e-06,
+      "loss": 0.0141,
+      "step": 3100
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 7.842324250513567e-06,
+      "loss": 0.0105,
+      "step": 3101
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 7.81918183002891e-06,
+      "loss": 0.004,
+      "step": 3102
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 7.79607070946164e-06,
+      "loss": 0.0118,
+      "step": 3103
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 7.772990905961164e-06,
+      "loss": 0.0336,
+      "step": 3104
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 7.749942436653662e-06,
+      "loss": 0.0387,
+      "step": 3105
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 7.726925318642042e-06,
+      "loss": 0.0309,
+      "step": 3106
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 7.703939569005991e-06,
+      "loss": 0.0291,
+      "step": 3107
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 7.680985204801856e-06,
+      "loss": 0.0289,
+      "step": 3108
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 7.658062243062724e-06,
+      "loss": 0.0052,
+      "step": 3109
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 7.635170700798417e-06,
+      "loss": 0.0201,
+      "step": 3110
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 7.612310594995398e-06,
+      "loss": 0.0046,
+      "step": 3111
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 7.589481942616788e-06,
+      "loss": 0.014,
+      "step": 3112
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 7.566684760602421e-06,
+      "loss": 0.0039,
+      "step": 3113
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 7.543919065868748e-06,
+      "loss": 0.0111,
+      "step": 3114
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 7.5211848753088566e-06,
+      "loss": 0.0043,
+      "step": 3115
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 7.498482205792456e-06,
+      "loss": 0.0041,
+      "step": 3116
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 7.4758110741658615e-06,
+      "loss": 0.0196,
+      "step": 3117
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 7.453171497252032e-06,
+      "loss": 0.0155,
+      "step": 3118
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 7.430563491850428e-06,
+      "loss": 0.0225,
+      "step": 3119
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 7.4079870747371325e-06,
+      "loss": 0.008,
+      "step": 3120
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 7.38544226266481e-06,
+      "loss": 0.0174,
+      "step": 3121
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 7.362929072362645e-06,
+      "loss": 0.0096,
+      "step": 3122
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 7.34044752053632e-06,
+      "loss": 0.0504,
+      "step": 3123
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 7.317997623868123e-06,
+      "loss": 0.0105,
+      "step": 3124
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 7.2955793990167835e-06,
+      "loss": 0.0095,
+      "step": 3125
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 7.273192862617567e-06,
+      "loss": 0.0125,
+      "step": 3126
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 7.250838031282209e-06,
+      "loss": 0.0137,
+      "step": 3127
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 7.228514921598922e-06,
+      "loss": 0.0359,
+      "step": 3128
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 7.206223550132374e-06,
+      "loss": 0.0107,
+      "step": 3129
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 7.183963933423698e-06,
+      "loss": 0.0259,
+      "step": 3130
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 7.161736087990434e-06,
+      "loss": 0.0103,
+      "step": 3131
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 7.139540030326602e-06,
+      "loss": 0.0095,
+      "step": 3132
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 7.117375776902563e-06,
+      "loss": 0.0206,
+      "step": 3133
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 7.095243344165115e-06,
+      "loss": 0.0267,
+      "step": 3134
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 7.073142748537465e-06,
+      "loss": 0.0071,
+      "step": 3135
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 7.051074006419167e-06,
+      "loss": 0.0052,
+      "step": 3136
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 7.029037134186112e-06,
+      "loss": 0.043,
+      "step": 3137
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 7.007032148190606e-06,
+      "loss": 0.0099,
+      "step": 3138
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 6.9850590647612565e-06,
+      "loss": 0.0618,
+      "step": 3139
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 6.963117900202998e-06,
+      "loss": 0.0043,
+      "step": 3140
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 6.9412086707970834e-06,
+      "loss": 0.0043,
+      "step": 3141
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 6.919331392801065e-06,
+      "loss": 0.0212,
+      "step": 3142
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 6.897486082448817e-06,
+      "loss": 0.021,
+      "step": 3143
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 6.875672755950441e-06,
+      "loss": 0.0118,
+      "step": 3144
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 6.853891429492321e-06,
+      "loss": 0.0146,
+      "step": 3145
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 6.832142119237134e-06,
+      "loss": 0.0131,
+      "step": 3146
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 6.8104248413237706e-06,
+      "loss": 0.0154,
+      "step": 3147
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 6.788739611867317e-06,
+      "loss": 0.0349,
+      "step": 3148
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 6.767086446959142e-06,
+      "loss": 0.0032,
+      "step": 3149
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 6.7454653626667915e-06,
+      "loss": 0.0026,
+      "step": 3150
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 6.7238763750339996e-06,
+      "loss": 0.0068,
+      "step": 3151
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 6.7023195000806945e-06,
+      "loss": 0.0148,
+      "step": 3152
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 6.6807947538029715e-06,
+      "loss": 0.0095,
+      "step": 3153
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 6.659302152173086e-06,
+      "loss": 0.0125,
+      "step": 3154
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 6.6378417111394305e-06,
+      "loss": 0.004,
+      "step": 3155
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 6.616413446626546e-06,
+      "loss": 0.0082,
+      "step": 3156
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 6.595017374535123e-06,
+      "loss": 0.0043,
+      "step": 3157
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 6.5736535107419005e-06,
+      "loss": 0.0074,
+      "step": 3158
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 6.552321871099748e-06,
+      "loss": 0.0119,
+      "step": 3159
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 6.531022471437653e-06,
+      "loss": 0.0066,
+      "step": 3160
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 6.509755327560651e-06,
+      "loss": 0.029,
+      "step": 3161
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 6.488520455249819e-06,
+      "loss": 0.0569,
+      "step": 3162
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 6.46731787026234e-06,
+      "loss": 0.0085,
+      "step": 3163
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 6.446147588331408e-06,
+      "loss": 0.0173,
+      "step": 3164
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 6.4250096251662395e-06,
+      "loss": 0.0087,
+      "step": 3165
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 6.403903996452087e-06,
+      "loss": 0.0346,
+      "step": 3166
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 6.382830717850197e-06,
+      "loss": 0.0067,
+      "step": 3167
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 6.361789804997837e-06,
+      "loss": 0.0187,
+      "step": 3168
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 6.340781273508217e-06,
+      "loss": 0.0071,
+      "step": 3169
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 6.31980513897053e-06,
+      "loss": 0.0106,
+      "step": 3170
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 6.298861416949964e-06,
+      "loss": 0.0346,
+      "step": 3171
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 6.27795012298763e-06,
+      "loss": 0.0053,
+      "step": 3172
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 6.257071272600545e-06,
+      "loss": 0.0072,
+      "step": 3173
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 6.23622488128171e-06,
+      "loss": 0.0062,
+      "step": 3174
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 6.215410964500001e-06,
+      "loss": 0.0161,
+      "step": 3175
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 6.194629537700219e-06,
+      "loss": 0.0234,
+      "step": 3176
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 6.17388061630303e-06,
+      "loss": 0.0026,
+      "step": 3177
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 6.153164215705004e-06,
+      "loss": 0.0096,
+      "step": 3178
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 6.1324803512785685e-06,
+      "loss": 0.0045,
+      "step": 3179
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 6.111829038372008e-06,
+      "loss": 0.0045,
+      "step": 3180
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 6.091210292309446e-06,
+      "loss": 0.0248,
+      "step": 3181
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 6.070624128390873e-06,
+      "loss": 0.003,
+      "step": 3182
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 6.050070561892046e-06,
+      "loss": 0.0151,
+      "step": 3183
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 6.029549608064572e-06,
+      "loss": 0.0079,
+      "step": 3184
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 6.00906128213587e-06,
+      "loss": 0.0098,
+      "step": 3185
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 5.9886055993091164e-06,
+      "loss": 0.0206,
+      "step": 3186
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 5.968182574763259e-06,
+      "loss": 0.0147,
+      "step": 3187
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 5.947792223653048e-06,
+      "loss": 0.002,
+      "step": 3188
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 5.927434561108975e-06,
+      "loss": 0.0619,
+      "step": 3189
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 5.9071096022372595e-06,
+      "loss": 0.0195,
+      "step": 3190
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 5.886817362119862e-06,
+      "loss": 0.0373,
+      "step": 3191
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 5.86655785581447e-06,
+      "loss": 0.0255,
+      "step": 3192
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 5.84633109835449e-06,
+      "loss": 0.0033,
+      "step": 3193
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 5.826137104749002e-06,
+      "loss": 0.0063,
+      "step": 3194
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 5.8059758899827785e-06,
+      "loss": 0.0073,
+      "step": 3195
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 5.785847469016298e-06,
+      "loss": 0.0049,
+      "step": 3196
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 5.76575185678569e-06,
+      "loss": 0.0226,
+      "step": 3197
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 5.745689068202692e-06,
+      "loss": 0.0156,
+      "step": 3198
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 5.725659118154758e-06,
+      "loss": 0.0332,
+      "step": 3199
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 5.705662021504943e-06,
+      "loss": 0.0104,
+      "step": 3200
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 5.685697793091909e-06,
+      "loss": 0.0023,
+      "step": 3201
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 5.6657664477299535e-06,
+      "loss": 0.0228,
+      "step": 3202
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 5.645868000208953e-06,
+      "loss": 0.003,
+      "step": 3203
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 5.626002465294389e-06,
+      "loss": 0.0283,
+      "step": 3204
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 5.606169857727301e-06,
+      "loss": 0.0068,
+      "step": 3205
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 5.586370192224305e-06,
+      "loss": 0.0052,
+      "step": 3206
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 5.566603483477606e-06,
+      "loss": 0.0033,
+      "step": 3207
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 5.546869746154892e-06,
+      "loss": 0.0241,
+      "step": 3208
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 5.527168994899407e-06,
+      "loss": 0.014,
+      "step": 3209
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 5.507501244329954e-06,
+      "loss": 0.0023,
+      "step": 3210
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 5.487866509040818e-06,
+      "loss": 0.0522,
+      "step": 3211
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 5.468264803601758e-06,
+      "loss": 0.0511,
+      "step": 3212
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 5.448696142558079e-06,
+      "loss": 0.0273,
+      "step": 3213
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 5.429160540430533e-06,
+      "loss": 0.0033,
+      "step": 3214
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 5.409658011715346e-06,
+      "loss": 0.0033,
+      "step": 3215
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 5.390188570884203e-06,
+      "loss": 0.0058,
+      "step": 3216
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 5.370752232384229e-06,
+      "loss": 0.0337,
+      "step": 3217
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 5.351349010638024e-06,
+      "loss": 0.0289,
+      "step": 3218
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 5.331978920043551e-06,
+      "loss": 0.0335,
+      "step": 3219
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 5.31264197497422e-06,
+      "loss": 0.0039,
+      "step": 3220
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 5.293338189778874e-06,
+      "loss": 0.0113,
+      "step": 3221
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 5.2740675787817106e-06,
+      "loss": 0.0137,
+      "step": 3222
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 5.254830156282303e-06,
+      "loss": 0.0092,
+      "step": 3223
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 5.235625936555649e-06,
+      "loss": 0.0364,
+      "step": 3224
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 5.216454933852055e-06,
+      "loss": 0.0045,
+      "step": 3225
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 5.197317162397208e-06,
+      "loss": 0.0261,
+      "step": 3226
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 5.178212636392127e-06,
+      "loss": 0.0292,
+      "step": 3227
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 5.159141370013165e-06,
+      "loss": 0.0188,
+      "step": 3228
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 5.140103377411993e-06,
+      "loss": 0.0363,
+      "step": 3229
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 5.1210986727155855e-06,
+      "loss": 0.0289,
+      "step": 3230
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 5.102127270026219e-06,
+      "loss": 0.0079,
+      "step": 3231
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 5.083189183421483e-06,
+      "loss": 0.0039,
+      "step": 3232
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 5.064284426954213e-06,
+      "loss": 0.033,
+      "step": 3233
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 5.0454130146525e-06,
+      "loss": 0.0068,
+      "step": 3234
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 5.026574960519748e-06,
+      "loss": 0.0029,
+      "step": 3235
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 5.007770278534568e-06,
+      "loss": 0.0329,
+      "step": 3236
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 4.988998982650789e-06,
+      "loss": 0.002,
+      "step": 3237
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 4.970261086797517e-06,
+      "loss": 0.0108,
+      "step": 3238
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 4.951556604879048e-06,
+      "loss": 0.0054,
+      "step": 3239
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 4.932885550774868e-06,
+      "loss": 0.0676,
+      "step": 3240
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 4.9142479383396885e-06,
+      "loss": 0.0017,
+      "step": 3241
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 4.895643781403375e-06,
+      "loss": 0.0162,
+      "step": 3242
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 4.877073093771018e-06,
+      "loss": 0.0183,
+      "step": 3243
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 4.858535889222799e-06,
+      "loss": 0.0411,
+      "step": 3244
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 4.840032181514098e-06,
+      "loss": 0.0145,
+      "step": 3245
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 4.821561984375461e-06,
+      "loss": 0.0057,
+      "step": 3246
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 4.803125311512529e-06,
+      "loss": 0.0104,
+      "step": 3247
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 4.784722176606055e-06,
+      "loss": 0.0258,
+      "step": 3248
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 4.766352593311962e-06,
+      "loss": 0.0133,
+      "step": 3249
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 4.748016575261221e-06,
+      "loss": 0.0082,
+      "step": 3250
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 4.729714136059932e-06,
+      "loss": 0.0037,
+      "step": 3251
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 4.711445289289257e-06,
+      "loss": 0.0025,
+      "step": 3252
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 4.693210048505437e-06,
+      "loss": 0.0053,
+      "step": 3253
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 4.675008427239785e-06,
+      "loss": 0.0044,
+      "step": 3254
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 4.65684043899865e-06,
+      "loss": 0.0074,
+      "step": 3255
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 4.638706097263429e-06,
+      "loss": 0.0071,
+      "step": 3256
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 4.620605415490575e-06,
+      "loss": 0.0139,
+      "step": 3257
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 4.6025384071115395e-06,
+      "loss": 0.0737,
+      "step": 3258
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 4.58450508553277e-06,
+      "loss": 0.0036,
+      "step": 3259
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 4.566505464135767e-06,
+      "loss": 0.0156,
+      "step": 3260
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 4.548539556276998e-06,
+      "loss": 0.0068,
+      "step": 3261
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 4.5306073752878816e-06,
+      "loss": 0.0253,
+      "step": 3262
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 4.512708934474863e-06,
+      "loss": 0.0147,
+      "step": 3263
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 4.494844247119329e-06,
+      "loss": 0.0195,
+      "step": 3264
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 4.477013326477614e-06,
+      "loss": 0.0083,
+      "step": 3265
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 4.459216185780996e-06,
+      "loss": 0.0026,
+      "step": 3266
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 4.441452838235694e-06,
+      "loss": 0.0041,
+      "step": 3267
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 4.423723297022864e-06,
+      "loss": 0.0297,
+      "step": 3268
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 4.406027575298544e-06,
+      "loss": 0.0277,
+      "step": 3269
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 4.388365686193685e-06,
+      "loss": 0.0115,
+      "step": 3270
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 4.370737642814171e-06,
+      "loss": 0.0191,
+      "step": 3271
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 4.353143458240727e-06,
+      "loss": 0.0475,
+      "step": 3272
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 4.335583145528954e-06,
+      "loss": 0.034,
+      "step": 3273
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 4.318056717709357e-06,
+      "loss": 0.0035,
+      "step": 3274
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 4.300564187787259e-06,
+      "loss": 0.0322,
+      "step": 3275
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 4.283105568742846e-06,
+      "loss": 0.0201,
+      "step": 3276
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 4.265680873531136e-06,
+      "loss": 0.0198,
+      "step": 3277
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 4.248290115081976e-06,
+      "loss": 0.0267,
+      "step": 3278
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 4.230933306300033e-06,
+      "loss": 0.0236,
+      "step": 3279
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 4.213610460064782e-06,
+      "loss": 0.0197,
+      "step": 3280
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 4.196321589230473e-06,
+      "loss": 0.0227,
+      "step": 3281
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 4.1790667066262e-06,
+      "loss": 0.0188,
+      "step": 3282
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 4.1618458250557875e-06,
+      "loss": 0.0038,
+      "step": 3283
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 4.14465895729782e-06,
+      "loss": 0.0135,
+      "step": 3284
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 4.127506116105701e-06,
+      "loss": 0.0171,
+      "step": 3285
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 4.110387314207537e-06,
+      "loss": 0.0109,
+      "step": 3286
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 4.0933025643061895e-06,
+      "loss": 0.0145,
+      "step": 3287
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 4.076251879079252e-06,
+      "loss": 0.0051,
+      "step": 3288
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 4.0592352711790346e-06,
+      "loss": 0.0313,
+      "step": 3289
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 4.042252753232579e-06,
+      "loss": 0.0079,
+      "step": 3290
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 4.025304337841607e-06,
+      "loss": 0.0141,
+      "step": 3291
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 4.008390037582543e-06,
+      "loss": 0.0048,
+      "step": 3292
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.991509865006521e-06,
+      "loss": 0.0191,
+      "step": 3293
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 3.974663832639308e-06,
+      "loss": 0.0277,
+      "step": 3294
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.957851952981361e-06,
+      "loss": 0.0275,
+      "step": 3295
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 3.941074238507802e-06,
+      "loss": 0.0034,
+      "step": 3296
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.9243307016684e-06,
+      "loss": 0.0249,
+      "step": 3297
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 3.907621354887519e-06,
+      "loss": 0.0023,
+      "step": 3298
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.890946210564222e-06,
+      "loss": 0.0044,
+      "step": 3299
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 3.874305281072144e-06,
+      "loss": 0.0687,
+      "step": 3300
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.85769857875955e-06,
+      "loss": 0.005,
+      "step": 3301
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 3.841126115949301e-06,
+      "loss": 0.0056,
+      "step": 3302
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.824587904938842e-06,
+      "loss": 0.0207,
+      "step": 3303
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 3.808083958000225e-06,
+      "loss": 0.0018,
+      "step": 3304
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 3.791614287380063e-06,
+      "loss": 0.0205,
+      "step": 3305
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.7751789052995146e-06,
+      "loss": 0.0093,
+      "step": 3306
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 3.7587778239543457e-06,
+      "loss": 0.0036,
+      "step": 3307
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.7424110555148274e-06,
+      "loss": 0.0234,
+      "step": 3308
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 3.7260786121257663e-06,
+      "loss": 0.0024,
+      "step": 3309
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.709780505906535e-06,
+      "loss": 0.0052,
+      "step": 3310
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 3.6935167489509923e-06,
+      "loss": 0.0026,
+      "step": 3311
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.6772873533275185e-06,
+      "loss": 0.018,
+      "step": 3312
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 3.6610923310790136e-06,
+      "loss": 0.031,
+      "step": 3313
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.644931694222836e-06,
+      "loss": 0.0113,
+      "step": 3314
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 3.6288054547508667e-06,
+      "loss": 0.0342,
+      "step": 3315
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.6127136246294323e-06,
+      "loss": 0.0175,
+      "step": 3316
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 3.5966562157993377e-06,
+      "loss": 0.1344,
+      "step": 3317
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.580633240175868e-06,
+      "loss": 0.0268,
+      "step": 3318
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 3.564644709648707e-06,
+      "loss": 0.0145,
+      "step": 3319
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.548690636082014e-06,
+      "loss": 0.0056,
+      "step": 3320
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 3.5327710313143914e-06,
+      "loss": 0.0019,
+      "step": 3321
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 3.5168859071588402e-06,
+      "loss": 0.0078,
+      "step": 3322
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.5010352754027644e-06,
+      "loss": 0.023,
+      "step": 3323
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 3.4852191478080053e-06,
+      "loss": 0.0192,
+      "step": 3324
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.469437536110787e-06,
+      "loss": 0.0075,
+      "step": 3325
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 3.4536904520217185e-06,
+      "loss": 0.001,
+      "step": 3326
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.4379779072257868e-06,
+      "loss": 0.0188,
+      "step": 3327
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 3.4222999133823484e-06,
+      "loss": 0.0267,
+      "step": 3328
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.4066564821251367e-06,
+      "loss": 0.0242,
+      "step": 3329
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 3.391047625062216e-06,
+      "loss": 0.0051,
+      "step": 3330
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.375473353776004e-06,
+      "loss": 0.0068,
+      "step": 3331
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 3.3599336798232683e-06,
+      "loss": 0.0086,
+      "step": 3332
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.344428614735096e-06,
+      "loss": 0.0711,
+      "step": 3333
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 3.3289581700168605e-06,
+      "loss": 0.0134,
+      "step": 3334
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.3135223571482966e-06,
+      "loss": 0.0146,
+      "step": 3335
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 3.2981211875834195e-06,
+      "loss": 0.0059,
+      "step": 3336
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 3.2827546727505267e-06,
+      "loss": 0.0051,
+      "step": 3337
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.2674228240522133e-06,
+      "loss": 0.0108,
+      "step": 3338
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 3.2521256528653454e-06,
+      "loss": 0.0101,
+      "step": 3339
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.2368631705410655e-06,
+      "loss": 0.0557,
+      "step": 3340
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 3.221635388404759e-06,
+      "loss": 0.0039,
+      "step": 3341
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.2064423177560697e-06,
+      "loss": 0.0098,
+      "step": 3342
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 3.191283969868908e-06,
+      "loss": 0.0414,
+      "step": 3343
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.1761603559913755e-06,
+      "loss": 0.0055,
+      "step": 3344
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 3.1610714873458238e-06,
+      "loss": 0.0107,
+      "step": 3345
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.1460173751288236e-06,
+      "loss": 0.022,
+      "step": 3346
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 3.1309980305111675e-06,
+      "loss": 0.0042,
+      "step": 3347
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.116013464637796e-06,
+      "loss": 0.0476,
+      "step": 3348
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 3.1010636886279087e-06,
+      "loss": 0.0115,
+      "step": 3349
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.086148713574849e-06,
+      "loss": 0.0138,
+      "step": 3350
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 3.071268550546147e-06,
+      "loss": 0.0042,
+      "step": 3351
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 3.0564232105835034e-06,
+      "loss": 0.055,
+      "step": 3352
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0416127047027787e-06,
+      "loss": 0.0117,
+      "step": 3353
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 3.0268370438939766e-06,
+      "loss": 0.0228,
+      "step": 3354
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 3.0120962391212536e-06,
+      "loss": 0.0029,
+      "step": 3355
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 2.9973903013228887e-06,
+      "loss": 0.0217,
+      "step": 3356
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.9827192414113193e-06,
+      "loss": 0.0178,
+      "step": 3357
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 2.96808307027307e-06,
+      "loss": 0.0045,
+      "step": 3358
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.9534817987687756e-06,
+      "loss": 0.0194,
+      "step": 3359
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 2.9389154377332086e-06,
+      "loss": 0.0083,
+      "step": 3360
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.9243839979751943e-06,
+      "loss": 0.0064,
+      "step": 3361
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 2.9098874902776684e-06,
+      "loss": 0.0197,
+      "step": 3362
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.895425925397649e-06,
+      "loss": 0.0814,
+      "step": 3363
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 2.8809993140662074e-06,
+      "loss": 0.0082,
+      "step": 3364
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.8666076669884966e-06,
+      "loss": 0.0148,
+      "step": 3365
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 2.852250994843708e-06,
+      "loss": 0.0112,
+      "step": 3366
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.837929308285081e-06,
+      "loss": 0.0168,
+      "step": 3367
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 2.8236426179399256e-06,
+      "loss": 0.0089,
+      "step": 3368
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 2.8093909344095448e-06,
+      "loss": 0.0095,
+      "step": 3369
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.795174268269268e-06,
+      "loss": 0.0172,
+      "step": 3370
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.780992630068474e-06,
+      "loss": 0.047,
+      "step": 3371
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.7668460303305167e-06,
+      "loss": 0.0081,
+      "step": 3372
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 2.752734479552749e-06,
+      "loss": 0.0452,
+      "step": 3373
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.738657988206539e-06,
+      "loss": 0.0253,
+      "step": 3374
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 2.7246165667372315e-06,
+      "loss": 0.0299,
+      "step": 3375
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.710610225564131e-06,
+      "loss": 0.0109,
+      "step": 3376
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 2.6966389750805344e-06,
+      "loss": 0.0065,
+      "step": 3377
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.682702825653688e-06,
+      "loss": 0.0171,
+      "step": 3378
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 2.6688017876247863e-06,
+      "loss": 0.0209,
+      "step": 3379
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.654935871308983e-06,
+      "loss": 0.0138,
+      "step": 3380
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 2.641105086995355e-06,
+      "loss": 0.0454,
+      "step": 3381
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.6273094449469292e-06,
+      "loss": 0.0069,
+      "step": 3382
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 2.6135489554006464e-06,
+      "loss": 0.0135,
+      "step": 3383
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 2.5998236285673382e-06,
+      "loss": 0.0037,
+      "step": 3384
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.586133474631791e-06,
+      "loss": 0.0048,
+      "step": 3385
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 2.5724785037526565e-06,
+      "loss": 0.003,
+      "step": 3386
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.5588587260624964e-06,
+      "loss": 0.0134,
+      "step": 3387
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 2.5452741516677427e-06,
+      "loss": 0.0354,
+      "step": 3388
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.531724790648715e-06,
+      "loss": 0.0052,
+      "step": 3389
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 2.518210653059605e-06,
+      "loss": 0.0373,
+      "step": 3390
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.5047317489284573e-06,
+      "loss": 0.0415,
+      "step": 3391
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 2.4912880882571822e-06,
+      "loss": 0.0067,
+      "step": 3392
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.4778796810215387e-06,
+      "loss": 0.0033,
+      "step": 3393
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 2.464506537171124e-06,
+      "loss": 0.0057,
+      "step": 3394
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.451168666629344e-06,
+      "loss": 0.002,
+      "step": 3395
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 2.4378660792934714e-06,
+      "loss": 0.0063,
+      "step": 3396
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.424598785034582e-06,
+      "loss": 0.0377,
+      "step": 3397
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 2.411366793697534e-06,
+      "loss": 0.0155,
+      "step": 3398
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 2.3981701151010395e-06,
+      "loss": 0.0087,
+      "step": 3399
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.3850087590375716e-06,
+      "loss": 0.0011,
+      "step": 3400
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 2.371882735273401e-06,
+      "loss": 0.0039,
+      "step": 3401
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.358792053548581e-06,
+      "loss": 0.0035,
+      "step": 3402
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 2.345736723576947e-06,
+      "loss": 0.0032,
+      "step": 3403
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.332716755046094e-06,
+      "loss": 0.0065,
+      "step": 3404
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 2.319732157617377e-06,
+      "loss": 0.0216,
+      "step": 3405
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.3067829409258947e-06,
+      "loss": 0.0061,
+      "step": 3406
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 2.293869114580527e-06,
+      "loss": 0.0098,
+      "step": 3407
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.280990688163859e-06,
+      "loss": 0.003,
+      "step": 3408
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 2.268147671232207e-06,
+      "loss": 0.0084,
+      "step": 3409
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.2553400733156314e-06,
+      "loss": 0.0595,
+      "step": 3410
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 2.2425679039179026e-06,
+      "loss": 0.018,
+      "step": 3411
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.2298311725165e-06,
+      "loss": 0.014,
+      "step": 3412
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 2.2171298885626026e-06,
+      "loss": 0.0195,
+      "step": 3413
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.204464061481093e-06,
+      "loss": 0.0203,
+      "step": 3414
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 2.1918337006705424e-06,
+      "loss": 0.0034,
+      "step": 3415
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 2.1792388155031983e-06,
+      "loss": 0.0053,
+      "step": 3416
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.1666794153249793e-06,
+      "loss": 0.0014,
+      "step": 3417
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 2.154155509455502e-06,
+      "loss": 0.0186,
+      "step": 3418
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.1416671071880222e-06,
+      "loss": 0.0154,
+      "step": 3419
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 2.1292142177894327e-06,
+      "loss": 0.0027,
+      "step": 3420
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.116796850500313e-06,
+      "loss": 0.0145,
+      "step": 3421
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 2.1044150145348553e-06,
+      "loss": 0.0052,
+      "step": 3422
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.0920687190808976e-06,
+      "loss": 0.0118,
+      "step": 3423
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 2.0797579732999063e-06,
+      "loss": 0.004,
+      "step": 3424
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.067482786326963e-06,
+      "loss": 0.0135,
+      "step": 3425
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 2.0552431672707704e-06,
+      "loss": 0.0049,
+      "step": 3426
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.043039125213625e-06,
+      "loss": 0.0331,
+      "step": 3427
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 2.0308706692114433e-06,
+      "loss": 0.0151,
+      "step": 3428
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.0187378082937135e-06,
+      "loss": 0.0036,
+      "step": 3429
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 2.006640551463529e-06,
+      "loss": 0.0098,
+      "step": 3430
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 1.994578907697542e-06,
+      "loss": 0.026,
+      "step": 3431
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.9825528859460206e-06,
+      "loss": 0.0358,
+      "step": 3432
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.970562495132755e-06,
+      "loss": 0.0429,
+      "step": 3433
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.958607744155105e-06,
+      "loss": 0.0062,
+      "step": 3434
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 1.946688641884009e-06,
+      "loss": 0.0103,
+      "step": 3435
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 1.934805197163925e-06,
+      "loss": 0.0125,
+      "step": 3436
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 1.9229574188128598e-06,
+      "loss": 0.0228,
+      "step": 3437
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.91114531562237e-06,
+      "loss": 0.0696,
+      "step": 3438
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.8993688963575119e-06,
+      "loss": 0.0057,
+      "step": 3439
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 1.8876281697568832e-06,
+      "loss": 0.0214,
+      "step": 3440
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 1.8759231445325931e-06,
+      "loss": 0.0172,
+      "step": 3441
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.8642538293702438e-06,
+      "loss": 0.0368,
+      "step": 3442
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.85262023292897e-06,
+      "loss": 0.0223,
+      "step": 3443
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 1.841022363841377e-06,
+      "loss": 0.021,
+      "step": 3444
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 1.8294602307135477e-06,
+      "loss": 0.0044,
+      "step": 3445
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 1.8179338421250857e-06,
+      "loss": 0.0149,
+      "step": 3446
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 1.8064432066290437e-06,
+      "loss": 0.0617,
+      "step": 3447
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 1.7949883327519457e-06,
+      "loss": 0.0288,
+      "step": 3448
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.7835692289937866e-06,
+      "loss": 0.0377,
+      "step": 3449
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.7721859038280164e-06,
+      "loss": 0.0016,
+      "step": 3450
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 1.7608383657015337e-06,
+      "loss": 0.0325,
+      "step": 3451
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 1.7495266230346808e-06,
+      "loss": 0.0023,
+      "step": 3452
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.738250684221243e-06,
+      "loss": 0.0042,
+      "step": 3453
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.7270105576284335e-06,
+      "loss": 0.0104,
+      "step": 3454
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.7158062515968908e-06,
+      "loss": 0.0158,
+      "step": 3455
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.7046377744406706e-06,
+      "loss": 0.0077,
+      "step": 3456
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.6935051344472596e-06,
+      "loss": 0.0068,
+      "step": 3457
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 1.6824083398775393e-06,
+      "loss": 0.0012,
+      "step": 3458
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.6713473989657668e-06,
+      "loss": 0.0042,
+      "step": 3459
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.6603223199196493e-06,
+      "loss": 0.0343,
+      "step": 3460
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 1.6493331109202315e-06,
+      "loss": 0.0014,
+      "step": 3461
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 1.638379780121979e-06,
+      "loss": 0.0031,
+      "step": 3462
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.627462335652702e-06,
+      "loss": 0.0126,
+      "step": 3463
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 1.616580785613614e-06,
+      "loss": 0.0041,
+      "step": 3464
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 1.6057351380792619e-06,
+      "loss": 0.0121,
+      "step": 3465
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.594925401097569e-06,
+      "loss": 0.0229,
+      "step": 3466
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.5841515826898134e-06,
+      "loss": 0.0102,
+      "step": 3467
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 1.5734136908506114e-06,
+      "loss": 0.0401,
+      "step": 3468
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 1.562711733547939e-06,
+      "loss": 0.0186,
+      "step": 3469
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 1.5520457187230665e-06,
+      "loss": 0.0068,
+      "step": 3470
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 1.5414156542906399e-06,
+      "loss": 0.0052,
+      "step": 3471
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.5308215481386e-06,
+      "loss": 0.0232,
+      "step": 3472
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.520263408128214e-06,
+      "loss": 0.0021,
+      "step": 3473
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.5097412420940592e-06,
+      "loss": 0.0129,
+      "step": 3474
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.4992550578440233e-06,
+      "loss": 0.0384,
+      "step": 3475
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.4888048631592877e-06,
+      "loss": 0.0079,
+      "step": 3476
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.478390665794327e-06,
+      "loss": 0.0283,
+      "step": 3477
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 1.4680124734769151e-06,
+      "loss": 0.0039,
+      "step": 3478
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.4576702939080976e-06,
+      "loss": 0.0157,
+      "step": 3479
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.447364134762197e-06,
+      "loss": 0.0466,
+      "step": 3480
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.437094003686812e-06,
+      "loss": 0.0175,
+      "step": 3481
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 1.4268599083028133e-06,
+      "loss": 0.0061,
+      "step": 3482
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.4166618562043266e-06,
+      "loss": 0.019,
+      "step": 3483
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.4064998549587095e-06,
+      "loss": 0.0085,
+      "step": 3484
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 1.3963739121066077e-06,
+      "loss": 0.0052,
+      "step": 3485
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 1.386284035161889e-06,
+      "loss": 0.0229,
+      "step": 3486
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 1.3762302316116526e-06,
+      "loss": 0.0055,
+      "step": 3487
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 1.3662125089162425e-06,
+      "loss": 0.0046,
+      "step": 3488
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.356230874509229e-06,
+      "loss": 0.0334,
+      "step": 3489
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.346285335797387e-06,
+      "loss": 0.0116,
+      "step": 3490
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 1.33637590016073e-06,
+      "loss": 0.0015,
+      "step": 3491
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 1.3265025749524652e-06,
+      "loss": 0.0197,
+      "step": 3492
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 1.3166653674990147e-06,
+      "loss": 0.0253,
+      "step": 3493
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.3068642850999957e-06,
+      "loss": 0.0062,
+      "step": 3494
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.2970993350282012e-06,
+      "loss": 0.0093,
+      "step": 3495
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.2873705245296463e-06,
+      "loss": 0.0036,
+      "step": 3496
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.277677860823512e-06,
+      "loss": 0.0516,
+      "step": 3497
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.2680213511021499e-06,
+      "loss": 0.002,
+      "step": 3498
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 1.2584010025310955e-06,
+      "loss": 0.0221,
+      "step": 3499
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.2488168222490381e-06,
+      "loss": 0.0128,
+      "step": 3500
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.2392688173678501e-06,
+      "loss": 0.0161,
+      "step": 3501
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 1.2297569949725418e-06,
+      "loss": 0.0226,
+      "step": 3502
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 1.2202813621212728e-06,
+      "loss": 0.0048,
+      "step": 3503
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 1.2108419258453686e-06,
+      "loss": 0.0158,
+      "step": 3504
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 1.2014386931492704e-06,
+      "loss": 0.0083,
+      "step": 3505
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.1920716710105695e-06,
+      "loss": 0.003,
+      "step": 3506
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.1827408663799999e-06,
+      "loss": 0.0148,
+      "step": 3507
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 1.17344628618139e-06,
+      "loss": 0.0226,
+      "step": 3508
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 1.164187937311706e-06,
+      "loss": 0.0339,
+      "step": 3509
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 1.1549658266410302e-06,
+      "loss": 0.0224,
+      "step": 3510
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 1.1457799610125498e-06,
+      "loss": 0.0199,
+      "step": 3511
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 1.1366303472425622e-06,
+      "loss": 0.011,
+      "step": 3512
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.1275169921204532e-06,
+      "loss": 0.0071,
+      "step": 3513
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.1184399024087189e-06,
+      "loss": 0.0361,
+      "step": 3514
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.109399084842927e-06,
+      "loss": 0.0054,
+      "step": 3515
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.10039454613175e-06,
+      "loss": 0.012,
+      "step": 3516
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.0914262929569153e-06,
+      "loss": 0.0108,
+      "step": 3517
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.0824943319732494e-06,
+      "loss": 0.0043,
+      "step": 3518
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 1.0735986698086453e-06,
+      "loss": 0.0119,
+      "step": 3519
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 1.0647393130640337e-06,
+      "loss": 0.0138,
+      "step": 3520
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.0559162683134393e-06,
+      "loss": 0.003,
+      "step": 3521
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 1.0471295421039251e-06,
+      "loss": 0.0013,
+      "step": 3522
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.038379140955603e-06,
+      "loss": 0.0059,
+      "step": 3523
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.029665071361635e-06,
+      "loss": 0.033,
+      "step": 3524
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 1.0209873397882209e-06,
+      "loss": 0.0146,
+      "step": 3525
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.0123459526746038e-06,
+      "loss": 0.0096,
+      "step": 3526
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.0037409164330492e-06,
+      "loss": 0.0074,
+      "step": 3527
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 9.951722374488436e-07,
+      "loss": 0.0097,
+      "step": 3528
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 9.86639922080307e-07,
+      "loss": 0.0065,
+      "step": 3529
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 9.781439766587797e-07,
+      "loss": 0.0044,
+      "step": 3530
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 9.696844074885858e-07,
+      "loss": 0.0057,
+      "step": 3531
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 9.612612208470983e-07,
+      "loss": 0.0232,
+      "step": 3532
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 9.528744229846675e-07,
+      "loss": 0.0236,
+      "step": 3533
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 9.445240201246319e-07,
+      "loss": 0.0055,
+      "step": 3534
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 9.36210018463346e-07,
+      "loss": 0.0184,
+      "step": 3535
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 9.279324241701415e-07,
+      "loss": 0.0176,
+      "step": 3536
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 9.196912433873439e-07,
+      "loss": 0.0034,
+      "step": 3537
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 9.114864822302449e-07,
+      "loss": 0.0072,
+      "step": 3538
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 9.033181467871188e-07,
+      "loss": 0.0138,
+      "step": 3539
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 8.95186243119206e-07,
+      "loss": 0.0067,
+      "step": 3540
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 8.87090777260724e-07,
+      "loss": 0.0076,
+      "step": 3541
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 8.790317552188287e-07,
+      "loss": 0.0394,
+      "step": 3542
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 8.710091829736699e-07,
+      "loss": 0.0319,
+      "step": 3543
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 8.630230664783245e-07,
+      "loss": 0.0064,
+      "step": 3544
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 8.550734116588133e-07,
+      "loss": 0.0102,
+      "step": 3545
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 8.471602244141119e-07,
+      "loss": 0.0168,
+      "step": 3546
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 8.392835106161345e-07,
+      "loss": 0.0254,
+      "step": 3547
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 8.314432761097335e-07,
+      "loss": 0.0048,
+      "step": 3548
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 8.236395267126884e-07,
+      "loss": 0.0169,
+      "step": 3549
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 8.158722682157005e-07,
+      "loss": 0.0146,
+      "step": 3550
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 8.081415063823983e-07,
+      "loss": 0.0019,
+      "step": 3551
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 8.004472469493318e-07,
+      "loss": 0.0065,
+      "step": 3552
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 7.927894956259563e-07,
+      "loss": 0.0361,
+      "step": 3553
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 7.851682580946429e-07,
+      "loss": 0.0083,
+      "step": 3554
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 7.775835400106679e-07,
+      "loss": 0.0104,
+      "step": 3555
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 7.700353470022015e-07,
+      "loss": 0.0097,
+      "step": 3556
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 7.625236846703243e-07,
+      "loss": 0.0142,
+      "step": 3557
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 7.55048558589e-07,
+      "loss": 0.0123,
+      "step": 3558
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 7.476099743050691e-07,
+      "loss": 0.0082,
+      "step": 3559
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 7.402079373382887e-07,
+      "loss": 0.0258,
+      "step": 3560
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 7.32842453181265e-07,
+      "loss": 0.024,
+      "step": 3561
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 7.255135272994984e-07,
+      "loss": 0.0086,
+      "step": 3562
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 7.182211651313553e-07,
+      "loss": 0.0224,
+      "step": 3563
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 7.109653720880683e-07,
+      "loss": 0.0171,
+      "step": 3564
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 7.03746153553736e-07,
+      "loss": 0.005,
+      "step": 3565
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 6.965635148853233e-07,
+      "loss": 0.0153,
+      "step": 3566
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 6.894174614126392e-07,
+      "loss": 0.0825,
+      "step": 3567
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 6.82307998438364e-07,
+      "loss": 0.0055,
+      "step": 3568
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 6.752351312380056e-07,
+      "loss": 0.0112,
+      "step": 3569
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 6.681988650599157e-07,
+      "loss": 0.0206,
+      "step": 3570
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 6.61199205125318e-07,
+      "loss": 0.0086,
+      "step": 3571
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 6.542361566282296e-07,
+      "loss": 0.0052,
+      "step": 3572
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 6.473097247355342e-07,
+      "loss": 0.0217,
+      "step": 3573
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 6.404199145869316e-07,
+      "loss": 0.0472,
+      "step": 3574
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 6.335667312949378e-07,
+      "loss": 0.005,
+      "step": 3575
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 6.267501799449072e-07,
+      "loss": 0.0418,
+      "step": 3576
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 6.199702655949991e-07,
+      "loss": 0.0272,
+      "step": 3577
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 6.132269932761947e-07,
+      "loss": 0.0036,
+      "step": 3578
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 6.065203679922749e-07,
+      "loss": 0.007,
+      "step": 3579
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 5.998503947198475e-07,
+      "loss": 0.0155,
+      "step": 3580
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.932170784082979e-07,
+      "loss": 0.0825,
+      "step": 3581
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 5.866204239798223e-07,
+      "loss": 0.0018,
+      "step": 3582
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.800604363294215e-07,
+      "loss": 0.0095,
+      "step": 3583
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 5.735371203248685e-07,
+      "loss": 0.0286,
+      "step": 3584
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 5.670504808067412e-07,
+      "loss": 0.0118,
+      "step": 3585
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 5.606005225883893e-07,
+      "loss": 0.0059,
+      "step": 3586
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 5.541872504559509e-07,
+      "loss": 0.0223,
+      "step": 3587
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 5.478106691683471e-07,
+      "loss": 0.0084,
+      "step": 3588
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 5.414707834572541e-07,
+      "loss": 0.0027,
+      "step": 3589
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 5.35167598027142e-07,
+      "loss": 0.0029,
+      "step": 3590
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 5.28901117555225e-07,
+      "loss": 0.1192,
+      "step": 3591
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 5.226713466915e-07,
+      "loss": 0.0026,
+      "step": 3592
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 5.164782900587084e-07,
+      "loss": 0.0056,
+      "step": 3593
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 5.103219522523683e-07,
+      "loss": 0.0176,
+      "step": 3594
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 5.042023378407201e-07,
+      "loss": 0.0135,
+      "step": 3595
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.981194513647869e-07,
+      "loss": 0.0063,
+      "step": 3596
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 4.92073297338308e-07,
+      "loss": 0.007,
+      "step": 3597
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.860638802477946e-07,
+      "loss": 0.0275,
+      "step": 3598
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.800912045524797e-07,
+      "loss": 0.0289,
+      "step": 3599
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.741552746843292e-07,
+      "loss": 0.019,
+      "step": 3600
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 4.6825609504805854e-07,
+      "loss": 0.0066,
+      "step": 3601
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.6239367002109956e-07,
+      "loss": 0.0157,
+      "step": 3602
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.565680039536169e-07,
+      "loss": 0.0046,
+      "step": 3603
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.5077910116848585e-07,
+      "loss": 0.0119,
+      "step": 3604
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.4502696596133153e-07,
+      "loss": 0.0592,
+      "step": 3605
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 4.393116026004618e-07,
+      "loss": 0.0099,
+      "step": 3606
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.3363301532692304e-07,
+      "loss": 0.0036,
+      "step": 3607
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.279912083544557e-07,
+      "loss": 0.0091,
+      "step": 3608
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.2238618586951663e-07,
+      "loss": 0.0491,
+      "step": 3609
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.168179520312676e-07,
+      "loss": 0.0126,
+      "step": 3610
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.112865109715647e-07,
+      "loss": 0.0138,
+      "step": 3611
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 4.057918667949745e-07,
+      "loss": 0.0128,
+      "step": 3612
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.0033402357874116e-07,
+      "loss": 0.0066,
+      "step": 3613
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 3.9491298537281394e-07,
+      "loss": 0.0488,
+      "step": 3614
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.895287561998362e-07,
+      "loss": 0.0069,
+      "step": 3615
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.841813400551175e-07,
+      "loss": 0.0073,
+      "step": 3616
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 3.788707409066672e-07,
+      "loss": 0.0097,
+      "step": 3617
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 3.7359696269517185e-07,
+      "loss": 0.0285,
+      "step": 3618
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.683600093339956e-07,
+      "loss": 0.0142,
+      "step": 3619
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.6315988470916884e-07,
+      "loss": 0.0186,
+      "step": 3620
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 3.579965926794049e-07,
+      "loss": 0.0058,
+      "step": 3621
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.528701370760779e-07,
+      "loss": 0.0151,
+      "step": 3622
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.4778052170323374e-07,
+      "loss": 0.016,
+      "step": 3623
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.4272775033757364e-07,
+      "loss": 0.0195,
+      "step": 3624
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.377118267284707e-07,
+      "loss": 0.0066,
+      "step": 3625
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.3273275459794197e-07,
+      "loss": 0.0146,
+      "step": 3626
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 3.277905376406654e-07,
+      "loss": 0.0187,
+      "step": 3627
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.2288517952397404e-07,
+      "loss": 0.0032,
+      "step": 3628
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.1801668388784514e-07,
+      "loss": 0.0285,
+      "step": 3629
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.131850543449111e-07,
+      "loss": 0.0048,
+      "step": 3630
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.0839029448043177e-07,
+      "loss": 0.0085,
+      "step": 3631
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.0363240785232227e-07,
+      "loss": 0.0213,
+      "step": 3632
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 2.9891139799113623e-07,
+      "loss": 0.0199,
+      "step": 3633
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 2.942272684000491e-07,
+      "loss": 0.028,
+      "step": 3634
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.8958002255488615e-07,
+      "loss": 0.025,
+      "step": 3635
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 2.849696639040944e-07,
+      "loss": 0.0143,
+      "step": 3636
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.80396195868754e-07,
+      "loss": 0.0041,
+      "step": 3637
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 2.7585962184256685e-07,
+      "loss": 0.0115,
+      "step": 3638
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.713599451918569e-07,
+      "loss": 0.0084,
+      "step": 3639
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 2.668971692555755e-07,
+      "loss": 0.0373,
+      "step": 3640
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.6247129734528476e-07,
+      "loss": 0.0069,
+      "step": 3641
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 2.580823327451576e-07,
+      "loss": 0.0253,
+      "step": 3642
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.537302787120055e-07,
+      "loss": 0.0076,
+      "step": 3643
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 2.4941513847521747e-07,
+      "loss": 0.0015,
+      "step": 3644
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.4513691523680994e-07,
+      "loss": 0.0067,
+      "step": 3645
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 2.408956121714101e-07,
+      "loss": 0.0021,
+      "step": 3646
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.3669123242623381e-07,
+      "loss": 0.0184,
+      "step": 3647
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 2.3252377912110212e-07,
+      "loss": 0.01,
+      "step": 3648
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.283932553484469e-07,
+      "loss": 0.0261,
+      "step": 3649
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 2.2429966417328308e-07,
+      "loss": 0.0092,
+      "step": 3650
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.2024300863322523e-07,
+      "loss": 0.0199,
+      "step": 3651
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.1622329173848211e-07,
+      "loss": 0.0504,
+      "step": 3652
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 2.1224051647185105e-07,
+      "loss": 0.0217,
+      "step": 3653
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.0829468578871247e-07,
+      "loss": 0.0146,
+      "step": 3654
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.0438580261704088e-07,
+      "loss": 0.0248,
+      "step": 3655
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.0051386985738273e-07,
+      "loss": 0.0113,
+      "step": 3656
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.9667889038288423e-07,
+      "loss": 0.0027,
+      "step": 3657
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.9288086703924678e-07,
+      "loss": 0.0066,
+      "step": 3658
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.8911980264477158e-07,
+      "loss": 0.0052,
+      "step": 3659
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.8539569999032058e-07,
+      "loss": 0.0066,
+      "step": 3660
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.8170856183933326e-07,
+      "loss": 0.0044,
+      "step": 3661
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.7805839092781552e-07,
+      "loss": 0.0096,
+      "step": 3662
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.7444518996435066e-07,
+      "loss": 0.0129,
+      "step": 3663
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.7086896163007738e-07,
+      "loss": 0.0282,
+      "step": 3664
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.6732970857871179e-07,
+      "loss": 0.0114,
+      "step": 3665
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.638274334365253e-07,
+      "loss": 0.0036,
+      "step": 3666
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.6036213880234464e-07,
+      "loss": 0.0107,
+      "step": 3667
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.56933827247574e-07,
+      "loss": 0.0037,
+      "step": 3668
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.5354250131615622e-07,
+      "loss": 0.0015,
+      "step": 3669
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 1.5018816352458942e-07,
+      "loss": 0.032,
+      "step": 3670
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.4687081636194368e-07,
+      "loss": 0.0004,
+      "step": 3671
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.435904622898221e-07,
+      "loss": 0.0066,
+      "step": 3672
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 1.4034710374238868e-07,
+      "loss": 0.0075,
+      "step": 3673
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 1.3714074312634051e-07,
+      "loss": 0.0026,
+      "step": 3674
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.3397138282093548e-07,
+      "loss": 0.0201,
+      "step": 3675
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.3083902517797008e-07,
+      "loss": 0.0066,
+      "step": 3676
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.277436725217851e-07,
+      "loss": 0.0241,
+      "step": 3677
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.2468532714925983e-07,
+      "loss": 0.0331,
+      "step": 3678
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.2166399132981233e-07,
+      "loss": 0.0049,
+      "step": 3679
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.1867966730539914e-07,
+      "loss": 0.064,
+      "step": 3680
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 1.1573235729051556e-07,
+      "loss": 0.0061,
+      "step": 3681
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.1282206347218439e-07,
+      "loss": 0.0135,
+      "step": 3682
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.09948788009967e-07,
+      "loss": 0.0169,
+      "step": 3683
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.0711253303594682e-07,
+      "loss": 0.021,
+      "step": 3684
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.0431330065475142e-07,
+      "loss": 0.0265,
+      "step": 3685
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 1.0155109294352483e-07,
+      "loss": 0.0128,
+      "step": 3686
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 9.882591195193302e-08,
+      "loss": 0.0045,
+      "step": 3687
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 9.61377597021862e-08,
+      "loss": 0.0035,
+      "step": 3688
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 9.348663818899428e-08,
+      "loss": 0.012,
+      "step": 3689
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 9.087254937960033e-08,
+      "loss": 0.0171,
+      "step": 3690
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 8.829549521376934e-08,
+      "loss": 0.0227,
+      "step": 3691
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 8.575547760378278e-08,
+      "loss": 0.0103,
+      "step": 3692
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 8.32524984344385e-08,
+      "loss": 0.0217,
+      "step": 3693
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 8.078655956305636e-08,
+      "loss": 0.0178,
+      "step": 3694
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 7.835766281945045e-08,
+      "loss": 0.0092,
+      "step": 3695
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 7.596581000597903e-08,
+      "loss": 0.001,
+      "step": 3696
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 7.361100289748901e-08,
+      "loss": 0.0129,
+      "step": 3697
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 7.129324324134378e-08,
+      "loss": 0.0216,
+      "step": 3698
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 6.90125327574176e-08,
+      "loss": 0.0014,
+      "step": 3699
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 6.676887313809555e-08,
+      "loss": 0.0012,
+      "step": 3700
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 6.456226604826254e-08,
+      "loss": 0.0009,
+      "step": 3701
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 6.239271312531435e-08,
+      "loss": 0.002,
+      "step": 3702
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 6.02602159791521e-08,
+      "loss": 0.007,
+      "step": 3703
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 5.816477619217109e-08,
+      "loss": 0.0049,
+      "step": 3704
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 5.6106395319277525e-08,
+      "loss": 0.0215,
+      "step": 3705
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 5.4085074887877393e-08,
+      "loss": 0.0219,
+      "step": 3706
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 5.21008163978709e-08,
+      "loss": 0.0038,
+      "step": 3707
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 5.015362132166357e-08,
+      "loss": 0.0142,
+      "step": 3708
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 4.824349110414961e-08,
+      "loss": 0.0178,
+      "step": 3709
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 4.63704271627341e-08,
+      "loss": 0.0253,
+      "step": 3710
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 4.4534430887299694e-08,
+      "loss": 0.0107,
+      "step": 3711
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 4.2735503640234376e-08,
+      "loss": 0.018,
+      "step": 3712
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 4.0973646756414795e-08,
+      "loss": 0.0057,
+      "step": 3713
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 3.9248861543217385e-08,
+      "loss": 0.0114,
+      "step": 3714
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 3.7561149280501696e-08,
+      "loss": 0.0008,
+      "step": 3715
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3.591051122061595e-08,
+      "loss": 0.0249,
+      "step": 3716
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 3.429694858840815e-08,
+      "loss": 0.0037,
+      "step": 3717
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 3.2720462581209424e-08,
+      "loss": 0.0054,
+      "step": 3718
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 3.118105436882846e-08,
+      "loss": 0.0051,
+      "step": 3719
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.9678725093579273e-08,
+      "loss": 0.0051,
+      "step": 3720
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 2.821347587025347e-08,
+      "loss": 0.0043,
+      "step": 3721
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.6785307786120207e-08,
+      "loss": 0.0197,
+      "step": 3722
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 2.5394221900942872e-08,
+      "loss": 0.0233,
+      "step": 3723
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.4040219246962424e-08,
+      "loss": 0.0071,
+      "step": 3724
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 2.27233008289085e-08,
+      "loss": 0.0113,
+      "step": 3725
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.1443467623982748e-08,
+      "loss": 0.0208,
+      "step": 3726
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 2.02007205818866e-08,
+      "loss": 0.0158,
+      "step": 3727
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 1.899506062477685e-08,
+      "loss": 0.0106,
+      "step": 3728
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.782648864731007e-08,
+      "loss": 0.0355,
+      "step": 3729
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.6695005516614847e-08,
+      "loss": 0.0034,
+      "step": 3730
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.560061207229735e-08,
+      "loss": 0.0238,
+      "step": 3731
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 1.4543309126446858e-08,
+      "loss": 0.0089,
+      "step": 3732
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.3523097463619127e-08,
+      "loss": 0.0205,
+      "step": 3733
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.2539977840858586e-08,
+      "loss": 0.0128,
+      "step": 3734
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 1.1593950987676128e-08,
+      "loss": 0.0143,
+      "step": 3735
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 1.0685017606071324e-08,
+      "loss": 0.0027,
+      "step": 3736
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 9.81317837049911e-09,
+      "loss": 0.01,
+      "step": 3737
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 8.978433927914198e-09,
+      "loss": 0.0086,
+      "step": 3738
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 8.180784897715565e-09,
+      "loss": 0.0261,
+      "step": 3739
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 7.420231871807515e-09,
+      "loss": 0.0218,
+      "step": 3740
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 6.696775414538614e-09,
+      "loss": 0.0177,
+      "step": 3741
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 6.010416062751656e-09,
+      "loss": 0.0042,
+      "step": 3742
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.36115432575035e-09,
+      "loss": 0.0202,
+      "step": 3743
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 4.748990685321531e-09,
+      "loss": 0.0127,
+      "step": 3744
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.173925595701844e-09,
+      "loss": 0.0134,
+      "step": 3745
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.6359594836277156e-09,
+      "loss": 0.0371,
+      "step": 3746
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 3.135092748279833e-09,
+      "loss": 0.0398,
+      "step": 3747
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.6713257613331098e-09,
+      "loss": 0.0243,
+      "step": 3748
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 2.2446588669122747e-09,
+      "loss": 0.0045,
+      "step": 3749
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.8550923816307298e-09,
+      "loss": 0.0214,
+      "step": 3750
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.5026265945572438e-09,
+      "loss": 0.003,
+      "step": 3751
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.1872617672326058e-09,
+      "loss": 0.0298,
+      "step": 3752
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 9.089981336807274e-10,
+      "loss": 0.0038,
+      "step": 3753
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 6.678359003808865e-10,
+      "loss": 0.0192,
+      "step": 3754
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 4.637752462843814e-10,
+      "loss": 0.0298,
+      "step": 3755
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 2.9681632280897933e-10,
+      "loss": 0.0075,
+      "step": 3756
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.6695925385001864e-10,
+      "loss": 0.024,
+      "step": 3757
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 7.420413576375573e-11,
+      "loss": 0.0167,
+      "step": 3758
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 1.855103738401809e-11,
+      "loss": 0.0092,
+      "step": 3759
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0,
+      "loss": 0.0173,
+      "step": 3760
+    },
+    {
+      "epoch": 20.0,
+      "step": 3760,
+      "total_flos": 1.6562315532959744e+18,
+      "train_loss": 0.3441217174872439,
+      "train_runtime": 7819.0799,
+      "train_samples_per_second": 61.25,
+      "train_steps_per_second": 0.481
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3760,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "total_flos": 1.6562315532959744e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}